The full text on this page is automatically extracted from the file linked above and may contain errors and inconsistencies.
BLS Handbook
of Methods
U.S. Department of Labor
Bureau of Labor Statistics
April 1988
Bulletin 2285
BLS Handbook
of Methods
U.S. Department of Labor
Ann McLaughlin, Secretary
Bureau of Labor Statistics
Janet L. Norwood, Commissioner
April 1988
Bulletin 2285
For sale by the Superintendent of Documents, U.S. Government P rinting Office Washington, D.C. 20402
Preface
The BLS Handbook o f Methods presents detailed
explanations of how the Bureau of Labor Statistics
obtains and prepares the economic data it publishes. BLS
statistics are used for many purposes, and sometimes data
well suited to one purpose may have limitations for
another. This edition of the Handbook, like its predeces
sors, aims to provide users of b l s data with the infor
mation necessary to evaluate the suitability of the statistics
for their needs.
Chapters for each major Bureau program give a brief
account of the program’s origin and development and
then follow with comprehensive information on con
cepts and definitions, sources of data and methods of
collection, statistical procedures, where the data are
published, and their uses and limitations. Sources of addi
tional technical information are given at the end of
most chapters.
The Handbook was written by members of the staffs
of the various b l s program offices. It was prepared
for publication by Rosalind Springsteen and Rosalie
Epstein in the Division of Special Publications, Office
of Publications.
Material in this publication is in the public domain
and, with appropriate credit, may be reproduced without
permission.
Contents
Introduction .................................................................................................................................................
Page
1
Chapter:
Labor Force Statistics
1.
2.
3.
4.
5.
Labor force, employment, and unemployment from the Current Population Survey . . .
Employment, hours, and earnings from the establishment su rv e y ......................................
Occupational employment statistics .........................................................................................
Measurement of unemployment in States and local a r e a s ....................................................
Employment and wages covered by unemploymentinsurance...............................................
3
13
28
31
35
Wages and industrial Relations
6.
7.
8.
9.
Occupational pay and supplementary b e n e fits.......................................................................
Negotiated wage and benefit ch an g es.......................................................................................
Employment Cost I n d e x .............................................................................................................
Employee benefits survey ...........................................................................................................
41
49
53
63
Productivity and Technology
10.
11.
12.
13.
Productivity measures: Business sector and major subsectors...............................................
Productivity measures: Industries and governm ent................................................................
Technological c h a n g e ...................................................................................................................
Foreign labor statistics ...............................................................................................................
69
78
88
92
Occupational Safety and Health
14.
Occupational safety and health statistics.................................................................................
98
Economic Growth and Employment Projections
15.
Economic growth and employment p rojections......................................................................
115
Prices and Living Conditions
16.
17.
18.
19.
Producer p ric e s .............................................................................................................................
International price indexes .........................................................................................................
Consumer expenditures and in c o m e .........................................................................................
Consumer Price In d e x .................................................................................................................
125
138
148
154
Appendixes:
A.
Seasonal adjustment methodology at b l s ......................................................................................
B.
Industrial classification ...............................................................................................................
C.
Geographic classification ...........................................................................................................
209
211
213
v
Introduction
When U.S. Commissioner of Labor Carroll Wright
issued his first annual report in March 1886, he estab
lished the policy of explaining his statistical methods to
his readers and of seeking to avoid misinterpretation of
the figures presented. During the more than 100 years
which have followed that initial report, the definitions,
methods, and limitations of the data published by the
Bureau of Labor and its successor, the Bureau of Labor
Statistics, have been explained again and again. The
reason for this is not merely to make the readers aware
of the known limitations of the statistics, but also to guide
them in the appropriate use of the information and to
assure them that proper standards have been observed.
This volume continues that tradition by providing
detailed descriptions of the Bureau’s statistical series. The
Bureau’s role, organization, and staff, and its approach
to its data collection activities also are discussed briefly.
advice and assistance to State agencies and other
cooperating organizations. An important aspect of the
work of the regional staffs has been explaining the con
cepts and techniques which the Bureau uses in compiling
the statistics.
Staff
The Bureau’s work extends beyond the initial collec
tion and processing of data. Its findings frequently
influence, and sometimes are crucial to, the determining
and shaping of public policy. Over the years, it has
developed a staff of professionals—economists, statisti
cians, mathematical statisticians, computer analysts, and
administrative specialists, among others—each playing a
significant role to ensure that the information issued by
the Bureau is of the highest quality.
An atmosphere of professional growth is fostered at
the Bureau. Education is a continual process reached
through technical and on-the-job training, attendance at
seminars and conferences, and university classes. The end
result is a highly skilled staff whose competence is well
regarded throughout the community of users of b l s data.
BLS role
Among Federal agencies collecting and issuing
statistics, the Bureau of Labor Statistics has been termed
a general-purpose statistical collection agency. The
Bureau’s figures are prepared to serve the needs of
business, labor, Congress, the general public, and the
administrative and executive agencies for information on
economic and social trends, b l s statistics are often quite
specialized, yet they meet general economic and social
data requirements. As the needs of users are likely to dif
fer, no statistic is ideal for all. This makes it important
that the characteristics of the measures and their limita
tions be well understood.
Consultation and advice
A statistical program too much detached from the users
of its data may fail in its principal mission. To avoid
sterility, the Bureau continuously invites advice and ideas
from users and experts in business, labor, professional,
and academic organizations and from members of the
public. Over the years, several commissions and commit
tees have been appointed to review specific Bureau pro
grams and have made valuable suggestions. Of course,
the Commissioner of Labor Statistics always retains final
responsibility for all decisions on statistical policy.
The Commissioner established two standing research
advisory committees in 1947. These groups, now called
the Business Research Advisory Council and the Labor
Research Advisory Council, advise on technical problems
and provide perspectives on Bureau programs in relation
to needs of their members. The councils accomplish their
work in general sessions and also through committees
on specialized subject-matter fields. Committees are
augmented by persons in industry or labor who, al
though not council members, have special expertise. The
councils may take formal action through resolutions or
Organization
The statistical programs of the Bureau were developed,
for the most part, independently of each other, taking
on characteristics suited to the requirements of the sub
ject under observation. As a result, the Bureau was
organized according to subject matter areas, an arrange
ment which has proved efficient and has been continued
over the years. Expertise in techniques, economic analy
sis, and other staff activities across subject-matter lines
was added to provide better use of the Bureau’s resources.
As the Bureau’s collection activities increased, regional
offices were established to administer the field programs,
to disseminate data to local users, and to furnish technical
1
recommendations on appropriate matters, but such
resolutions are merely advisory. Members of the coun
cils and the subcommittees serve in their individual
capacities, not as representatives of their organizations.
The members of the Business Research Advisory Coun
cil are designated by the Commissioner under authori
zation of the Secretary of Labor, after nomination by
the National Association of Manufacturers, the U.S.
Chamber of Commerce, the Business Round Table, and
the National Federation of Independent Business. The
members of the Labor Research Advisory Council are
designated by the Commissioner under authorization of
the Secretary of Labor, from nominations by the Direc
tor of Research, AFL-CIO. All research directors of inter
national unions represented in the a f l - c io are invited to
attend the general meetings of the council.
desired information, the Bureau has gone back often for
later information on the same subject or for new types
of information. The Bureau’s respondents have been
remarkable in their generosity. In no small measure, their
cooperation has been due to the great care taken to avoid
identifying the firm or the person supplying the infor
mation. Bureau employees pledge themselves to protect
these data and understand the adverse long-run conse
quences of even a single lapse. The only inducement is
to tell respondents that their contributions are important
to the success of the survey and that they may find the
survey results useful in their own pursuits.
The policy of not identifying respondents is imple
mented by combining the data reported by the different
sources and issuing the findings in summary form. Thus,
respondents are assured that their reports will be used for
statistical purposes only. All efforts to obtain legal access
to individual respondents’ reports have been successfully
resisted.1
Bureau Commissioners and their staffs have been con
vinced over the years that these policies contribute to the
reliability of b l s statistics.
Voluntary reporting and confidentiality
Voluntary reporting and the preserving of the confiden
tial nature of reported data are important characteristics
of b l s programs. For more than a century, the Bureau
has asked hundreds of thousands of firms and individuals
to provide information closely related to their daily affairs
and their personal lives. To some who have supplied the
1 For example, see Hustead v. Norwood, 529 F. SUPP. 323 (S.D.
Fla. 1981).
2
Chapter 1. Labor Force,
Employment, and Unemployment
from the Current Population Survey
sample survey of households, called the Monthly Report
of Unemployment, initiated by the Works Progress
Administration in 1940.
The household survey was transferred to the Bureau
of the Census in late 1942, and its name was changed to
the Monthly Report on the Labor Force. The survey title
was changed once more in 1948 to the present Current
Population Survey in order to reflect its expanding role
as a source for a wide variety of demographic, social, and
economic characteristics of the population. In 1959,
responsibility for analyzing and publishing the c p s labor
force data was transferred to b l s , although the Bureau
of the Census continues to collect and tabulate the
statistics.
Each month, the Bureau analyzes and publishes
statistics on the labor force, employment, unemployment,
and persons not in the labor force, classified by a variety
of demographic, social, and economic characteristics.
These statistics are derived from the Current Population
Survey ( c p s ), which is conducted by the Bureau of the
Census for the BLS. This monthly survey of the popula
tion is conducted using a scientifically selected sample of
households, representative of the civilian noninstitutional
population of the United States.
Background
Specific concepts of the labor force, employment, and
unemployment were introduced in the later stages of the
depression of the 1930’s. Before the 1930’s, aside from
attempts in some of the decennial censuses, no direct
measurements were made of the number of jobless per
sons. Mass unemployment in the early 1930’s increased
the need for statistics, and widely conflicting estimates
based on a variety of indirect techniques began to appear.
Dissatisfied with these methods, many research groups,
as well as State and municipal governments, began experi
menting with direct surveys of the population or samples
of the population. In these surveys, an attempt was made
to classify the population as employed, unemployed, or
out of the labor force by means of a series of questions
addressed to each individual. In most of the surveys, the
unemployed were defined as those who were not work
ing but were “ willing and able to work.’’ This concept,
however, did not meet the standards of objectivity that
many technicians felt were necessary to measure either
the level of unemployment at a point in time or changes
over periods of time. The criterion “ willing and able to
work,” when applied in specific situations, appeared to
be too intangible and too dependent upon the interpreta
tion and attitude of the persons being interviewed.
A set of precise concepts was developed in the late
1930’s to meet these various criticisms. The classification
of an individual depended principally upon his or her
actual activity within a designated time period; i.e., was
he or she working, looking for work, or engaged in other
activities? These concepts were adopted for the national
Description of Survey
The CPS provides statistics on the civilian noninstitu
tional population 16 years of age and over. Figures on
the resident Armed Forces (obtained monthly from the
Department of Defense) are added to the CPS estimates
to derive estimates of the “ total employed,” “ labor
force,” and the “ noninstitutional population.” Persons
under 16 years of age are excluded from the official
definition of the labor force because child labor laws,
compulsory school attendance, and general social custom
prevent most of these children in the United States from
working. The institutional population, which is also
excluded from coverage, consists of inmates of penal and
mental institutions, sanitariums, and homes for the aged,
infirm, and needy.
The CPS is collected each month from a probability
sample of approximately 59,500 occupied households.
Respondents are assured that all information obtained
is completely confidential and is used only for the pur
pose of statistical analysis. Although the survey is con
ducted on a strictly voluntary basis, refusals to cooperate
have averaged about 2Vi percent or less since its
inception.
The time period covered in the monthly survey is a
calendar week. A calendar week was selected as the survey
reference period because the period used must be short
enough so that the data obtained are “ current” but not
3
the termination of their most recent employment. A
period of 2 weeks or more during which a person was
employed or ceased looking for work is considered to
break the continuity of the present period of seeking
work. Two useful measures of the duration of unemploy
ment are the mean and the median. Mean duration is the
arithmetic average computed from single weeks of
unemployment. Median duration is the midpoint of a
distribution of weeks of unemployment.
The reasons for unemployment are divided into four
major groups. (1) Job losers are persons whose employ
ment ended involuntarily and who immediately began
looking for work, including those on layoff. (2) Job
leavers are persons who quit or otherwise terminated their
employment voluntarily and immediately began looking
for work. (3) Reentrants are persons who previously
worked at a full-time job lasting 2 weeks or longer but
who were out of the labor force prior to beginning to look
for work. (4) New entrants are persons who never
worked at a full-time job lasting 2 weeks or longer.
so short that the occurrence of holidays or other acciden
tal events might cause erratic fluctuations in the informa
tion obtained. A calendar week fulfills these conditions
as well as being a convenient and easily defined period
of time. Since July 1955, the calendar week, Sunday
through Saturday, which includes the 12th day of the
month has been defined as the reference week. The actual
survey is conducted during the following week, which is
the week containing the 19th day of the month.
Concepts
The criteria used in classifying persons on the basis of
their labor force activity are as follows:
Employment. Employed persons comprise (1) all those
who, during the survey week, did any work at all as paid
employees, or in their own business, profession, or on
their own farm, or who worked 15 hours or more as
unpaid workers in a family-operated enterprise; and
(2) all those who did not work but had jobs or businesses
from which they were temporarily absent due to illness,
bad weather, vacation, labor-management dispute, or
various personal reasons—whether or not they were paid
by their employers for the time off and whether or not
they were seeking other jobs. Members of the Armed
Forces stationed in the United States are also included
in the employed total. Each employed person is counted
only once. Those who held more than one job are counted
in the job at which they worked the greatest number of
hours during the survey week. Included in the total are
employed citizens of foreign countries, temporarily in the
United States, who are not living on the premises of an
embassy. Excluded are persons whose only activity con
sisted of work around their own home (such as house
work, painting, repairing, etc.) or volunteer work for
religious, charitable, and similar organizations.
Labor force. The civilian labor force comprises the total
of all civilians classified as employed and unemployed.
The labor force, in addition, includes members of the
Armed Forces stationed in the United States.
Unemployment rate. The overall unemployment rate
represents the number of unemployed as a percent of the
labor force including members of the Armed Forces sta
tioned in the United States. The unemployment rate for
all civilian workers represents the number of unemployed
as a percent of the civilian labor force. This measure is
also computed for various groups within the labor force
classified by sex, age, race, Hispanic ethnicity, industry,
occupation, etc., or for combinations of these character
istics. Because there is no comparable labor force, the
job-loser, job-leaver, reentrant, and new entrant rates are
each calculated as a percent of the total civilian labor
force; the sum of the rates for the four groups thus equals
the unemployment rate for all civilian workers.
Unemployment. Unemployed persons include those
who did not work at all during the survey week, were
looking for work, and were available for work during the
reference period (except for temporary illness). Those
who had made specific efforts to find work within the
preceding 4-week period—such as by registering at a
public or private employment agency, writing letters of
application, canvassing for work, etc.—are considered
to be looking for work. Persons who were waiting to be
recalled to a job from which they had been laid off or
were waiting to report to a new job within 30 days need
not be looking for work to be classified as unemployed.
Duration of unemployment represents the length of
time (through the current survey week) during which per
sons classified as unemployed had been continuously
looking for work and thus is a measure of an in-progress
spell of joblessness. For persons on layoff, duration of
unemployment represents the number of full weeks since
Not in laborforce. All civilians 16 years of age and over
who are not classified as employed or unemployed are
defined as “ not in the labor force.” These persons are
further classified as “ engaged in own housework,” “ in
school,” “ unable to work” because of long-term physical
or mental illness, “ retired,” and “ other.” The “ other”
group includes the voluntarily idle, seasonal workers for
whom the survey week fell in an “ off” season and who
were not reported as looking for work, and persons who
did not look for work because they believed that no jobs
were available because of personal factors—age, lack of
education or training, etc.—or because of the prevailing
job market situation.
In addition to students with no current interest in labor
force activity, the category “ not in labor force—in
4
school” includes persons attending school during the
survey week who had new jobs to which they were sched
uled to report within 30 days. It also includes students
looking for jobs for some period in the future, such as
the summer months. All persons—whether or not attend
ing school—who had new jobs not scheduled to begin
until after 30 days (and who were not working or look
ing for work) are also classified as not in the labor force.
For persons not in the labor force, detailed questions
are asked about previous work experience, intentions to
seek work, desire for a job at the time of interview, and
reasons for not looking for work. These questions are
asked only in those households that are in the fourth and
eighth months of the sample; i.e., the “ outgoing” rota
tion groups, those which had been in the sample for 3
previous months and would not be in for the subsequent
month. Prior to 1970, the detailed not-in-labor force
questions were asked of persons in the first and fifth
months in the sample; i.e., the “ incoming” groups. (See
Sampling.)
reliable subnational estimates, the survey was redesigned
as 51 separate samples, one for each of the States and the
District of Columbia. At the same time, the redesign main
tained the statistical reliability of the national estimates.1
Selection o f sample areas. The entire area of the United
States, consisting of 3,137 counties and independent
cities, is divided into 1,973 primary sampling units
(PSU’s). With some minor exceptions, a psu consists of
a county or a number of contiguous counties.
Metropolitan areas within a State are used as a basis
for forming psu ’s. Outside of metropolitan areas, coun
ties normally are combined, except where the geographic
area of the sample county is very large. Combining coun
ties to form psu ’s provides greater heterogeneity; a typi
cal PSU includes urban and rural residents of both high
and low economic levels and encompasses, to the extent
feasible, diverse occupations and industries. Another
important consideration is to have the psu sufficiently
compact so that, with a small sample spread throughout,
it can be efficiently canvassed without undue travel cost.
In the sample, the 1,973 psu ’s are grouped into strata
within each State. Then one psu is selected from each
stratum with the probability of selection proportionate
to the population size of the psu . psu’s in strata by them
selves are self-representing, and generally are the most
populated psu ’s in each State. Other strata are formed
by combining psu ’s that are similar in such charac
teristics as population growth; proportions of blacks and
of Hispanics; and population distribution by occupation,
industry, age, and sex. psu ’s selected from these strata
are non-self-representing, since each one chosen
represents the entire stratum.
Sampling
The CPS sample is located in 729 areas comprising over
1.000 counties and independent cities with coverage in
every State and the District of Columbia. In all, about
71.000 housing units and other living quarters are
designated for the sample each month, of which about
59,500 are occupied and thus eligible for interview. The
remainder are units found to be vacant, converted to
nonresidential use, containing persons who reside else
where, or ineligible for other reasons. Of the occupied
units eligible for enumeration, about 4 to 5 percent are
not interviewed in a given month because the residents
are not found at home after repeated calls, are tem
porarily absent, refuse to cooperate, or are unavailable
for other reasons. Information is obtained each month
for approximately 113,000 individuals 16 years and over.
The c p s sample was redesigned in April 1984-July 1985
to incorporate data from the 1980 census into the sam
pling frame, as is done after every decennial census. At
the same time, the structure of the sample design was
changed. Previously, the CPS had been designed as a
national sample with the goal of providing the best esti
mates of employment and unemployment for the United
States as a whole. During the 1970’s, however, growing
demands were placed on the CPS for the development of
State and local labor force estimates used in the alloca
tion of Federal revenues to States and areas and for other
purposes. The CPS sample was selectively expanded on
several occasions to improve the ability to provide State
and local area labor force estimates. Even with these
efforts, it was still difficult to obtain very reliable subna
tional data from the c p s except in large States and metro
politan areas. Therefore, to provide more accurate and
Selection o f sample households. Since the sample design
is essentially State based, the sampling ratio differs by
State and depends on the reliability requirements for
estimates for each State. The State sampling ratios range
roughly from 1 in every 200 households to 1 in every 2,500
households in each stratum of the State. The sampling
ratio occasionally is modified slightly to hold the size of
the sample relatively constant given the overall growth
of the population. The sampling ratio used within a sam
ple PSU depends on the probability of selection of the PSU
and the sampling ratio for the State. In a sample PSU
with a probability of selection of 1 in 10 in a State with
a sampling ratio of 1 in 2,500, the within-PSU sampling
ratio that results is 1 in 250, thereby achieving the desired
ratio of 1 in 2,500 for the stratum.
Within each designated psu , several steps are involved
in selecting the housing units to be enumerated. First, the
1980 census enumeration districts (ED’s), which are
1
For a complete description of the CPS redesign, see “ Redesign of
the Sample for the Current Population Survey,” E m ploym en t a n d Earn
ings, May 1984, pp. 7-10.
5
administrative units and contain on the average about 300
housing units, are ordered so that the sample would
reflect the demographic and residential characteristics of
the p s u . Within each ED, the housing units are sorted
geographically and are grouped into clusters of approx
imately four housing units. Then a systematic sample of
these clusters of housing units is selected.
The identification of the sample housing units within
an ED is made wherever possible from the list of ED ad
dresses compiled during the 1980 census or, if the address
es are incomplete or inadequate, by area sampling
methods. The address lists are used in about two-thirds
of the cases, primarily in urban areas, and area sampling
is applied in the remainder. In using the census lists, an
effort is made to have all small multiunit addresses (2-4
units) included within the same segment. This improves
the ability of the interviewer to cover all units designated
for the sample. Subject to this restriction, clusters con
sist of geographically contiguous addresses to the extent
possible.
This address list sample is supplemented by a selection
of the appropriate proportion of units newly constructed
in the p s u since the census date. The addresses of these
units are obtained mainly from records of building per
mits in that area.
In those enumeration districts where area sampling
methods are used, mainly rural areas, the e d ’s are sub
divided into segments; that is, small land areas having
well-defined boundaries and, in general, an expected
“ size” of about 8 to 12 housing units or other living
quarters. For each subdivided ED, one segment is desig
nated for the sample. When a selected segment contains
about four households, for example, all units are included
in the sample. When the size of the segment is several
times four units, an interviewer does not conduct inter
views at all housing units in the segment but uses a
systematic sampling pattern to achieve the equivalent of
a four-household cluster which is canvassed completely.
The remaining housing units in the segment are then
available for further samples.2
on; the last eighth are in for the eighth time, the fourth
month of the second period of enumeration. Under this
system, 75 percent of the sample segments are common
from month to month and 50 percent from year to year.
This procedure provides a substantial amount of monthto-month and year-to-year overlap in the panel, thus
reducing discontinuities in the series of data without
burdening any specific group of households with an
unduly long period of inquiry.
Collection Methods
Each month, during the calendar week containing the
19th day, interviewers contact some responsible person
in each of the sample households in the CPS. At the time
of the first enumeration of a household, the interviewer
visits the household and prepares a roster of the house
hold members, including their personal characteristics
(date of birth, sex, race, ethnic origin, marital status,
educational attainment, veteran status, etc.) and their
relationship to the person maintaining the family. This
roster is brought up to date at each subsequent interview
to take account of new or departed residents, changes in
marital status, and similar items. The information on
personal characteristics is thus available each month for
identification purposes and for cross-classification with
economic characteristics of the sample population.
Personal visits are required in the first, second, and
fifth month that the household is in the sample. In other
months, the interview may be conducted by telephone if
the respondent agrees to this procedure. Also, if no one
is at home when the interviewer visits, the respondent may
be contacted by telephone after the first month. Approx
imately 67 percent of the households in any given month
are interviewed by telephone.
At each monthly visit, a questionnaire is completed for
each household member 16 years of age and over. The
interviewer asks a series of standard questions on eco
nomic activity during the preceding week. The primary
purpose of these questions is to classify the sample
population into the three basic economic groups—the
employed, the unemployed, and those not in the labor
force. (See facsimile of the CPS standard questionnaire
at the end of this chapter.)
Additional questions are asked each month to help
clarify the information on labor force status. For the
employed, information is obtained on hours worked dur
ing the survey week, together with a description of the
current job. For those temporarily away from their jobs,
the enumerator records their reason for not working dur
ing the survey week, whether or not they were paid for
their time off, and whether they usually work full or part
time. For the unemployed, information is obtained on
(1) method(s) used to find work during the 4 weeks prior
to the interview, (2) the reasons the unemployed persons
Rotation o f sample. Part of the sample is changed each
month. For each sample, eight systematic subsamples
(rotation groups) or segments are identified. A given rota
tion group is interviewed for a total of 8 months, divided
into two equal periods. It is in the sample for 4 consecu
tive months 1 year, leaves the sample during the follow
ing 8 months, and then returns for the same 4 calendar
months of the next year. In any 1 month, one-eighth of
the sample segments are in their first month of enumera
tion, another eighth are in their second month, and so
2 For more detail on the selection of sample households, see “The
Current Population Survey: Design and Methodology,” Technical
P a p er N o . 40 (U.S. Department of Commerce, Bureau of the Census,
1978).
6
had started to look for work, (3) the length of time they
had been looking for work, (4) whether they were seek
ing full- or part-time work, and (5) a description of their
last full-time civilian job. For those outside the labor
force, their principal activity during the survey week—
keeping house, going to school, etc.—is recorded. In
addition, all households in the outgoing rotation groups
are asked questions on the work history, reasons for non
participation, and jobseeking intentions of individuals not
in the labor force. In 1979, questions were added to col
lect data on hourly and weekly earnings from a quarter
of the sample households—those in the two outgoing
rotation groups.
The information obtained for each person in the sam
ple is subjected to an edit by the regional offices of the
Bureau of the Census. The field edit serves to catch omis
sions, inconsistencies, illegible entries, and errors at the
point where correction is possible.
After the field edit, the questionnaires are forwarded
to the Jeffersonville, Indiana, office of the Bureau of the
Census by the end of the week after enumeration. The
raw data are transferred to computer tape and transmit
ted to the computers in the Bureau of the Census’
Washington office where they are checked for com
pleteness and consistency.
Although the CPS interviewers are chiefly part-time
workers, most have had several years of experience on
the survey. They are given intensive training when first
recruited and further training each month before the
survey. Through editing of their completed question
naires, repeated observation during enumeration, and a
systematic reinterview of part of their assignments by the
field supervisory staff, the work of the interviewers is
monitored and errors or deficiencies are brought directly
to their attention.
roads, refusals, or unavailability of the respondents for
other reasons. This noninterview adjustment is made
separately for combinations of similar sample areas that
are not necessarily contained within a State. Similarity
of sample areas is based on Metropolitan Statistical Area
(MSA) status and size. Within each combination of sam
ple areas, there is a further breakdown by residence. MSA
sample areas are categorized by “ central city” and
“ balance of the msa .” Residence categories of non-MSA
areas are “ urban” and “ rural.” The proportion of sam
ple households not interviewed varies from 4 to 5 per
cent depending on weather, vacations, etc.
2. Ratio estimates. The distribution of the population
selected for the sample may differ somewhat, by chance,
from that of the population as a whole, in such char
acteristics as age, race, sex, and residence. Since these
characteristics are closely correlated with labor force par
ticipation and other principal measurements made from
the sample, the survey estimates can be substantially
improved when weighted appropriately by the known
distribution of these population characteristics. This is
accomplished through two stages of ratio estimates as
follows:
a. First-stage ratio estimate. In the cps, a portion of
the 729 sample areas is chosen to represent other areas
not in the sample; the remainder of the sample areas
represent only themselves. The first-stage ratio estima
tion procedure was designed to reduce the portion of the
variance resulting from requiring sample areas to repre
sent nonsample areas. Therefore, this procedure is not
applied to sample areas which represent only themselves.
The adjustment is made at the State level for each of the
43 States that contain nonsample areas by race cells of
black and nonblack. The procedure corrects for the dif
ferences that existed in each cell at the time of the 1980
census between the race distribution of the population
in sample areas and the known race distribution of the
State.
Estimating Methods
Under the estimating methods used in the CPS, all o f
the results for a given month become available simulta
neously and are based on returns from the entire panel
of respondents. The estimation procedure involves
weighting the data from each sample person by the inverse
of the probability of the person being in the sample. This
gives a rough measure of the number of persons the
sample person represents. Beginning in 1985, almost all
sample persons within the same State will have the same
probability of selection. These estimates are then adjusted
for noninterviews, and the ratio estimation procedure is
applied.
b. Second-stage ratio estimate. In this stage, the sam
ple proportions of persons in specific categories are
adjusted to the distribution of independent current
estimates of the civilian noninstitutional population in
the same categories. The second-stage ratio adjustment,
which is performed to further reduce variability of the
estimates and to correct to some extent for cps under
coverage relative to the decennial census, is carried out
in three steps.
In the first step, the sample estimates are adjusted
within each State and the District of Columbia to an
independent control for the population 16 years and over.
The second step involves an adjustment by Hispanic
origin to a national estimate for eight age-sex categories
by Hispanic and non-Hispanic. This step was initiated
in January 1985. The adjustment is prepared by carrying
1. Noninterview adjustment. The weights for all inter
viewed households are adjusted to the extent needed to
account for occupied sample households for which no
information was obtained because of absence, impassable
7
the estimate date from the total including Armed Forces
overseas. The institutional population is computed by
applying institutional proportions derived from the 1980
census to the total population, including Armed Forces
overseas, for the estimate date. All computations
described above are performed in cells defined by single
year of age, race, and sex. The independent national con
trol totals are then obtained by collapsing these cells into
broader age groups for the population 16 years and older.
Beginning in January 1986, two changes were intro
duced into the estimation of the independent population
controls. For the first time, an explicit allowance for net
undocumented immigration since April 1, 1980 (the
census date), was added to the estimated level of legal
immigration. In addition, an increase in the estimate of
emigration of legal foreign-born residents has been incor
porated into the postcensal population estimates since
1980. The nature and effect of these changes are discussed
in detail in “ Changes in the Estimation Procedure in the
Current Population Survey Beginning in January 1986”
in the February 1986 issue of Employment and Earnings.
forward the 1980 census count for Hispanics by adding
estimated Hispanic births and immigrants and subtrac
ting estimated Hispanic deaths and emigrants to yield an
estimate of the Hispanic population by age and sex.
In the third step, a national adjustment is made by the
race categories of white, black, and other races to inde
pendent estimates by age and sex. The white and black
categories contain 32 age-sex groups each while the other
races category has 6 age-sex cells. The entire second-stage
adjustment procedure is iterated six times, each time
beginning at the weights developed the previous time. This
ensures that the sample estimates of the population for
both State and national age-sex-race-origin categories will
be virtually equal to the independent population control
totals. This second-stage adjustment procedure incor
porates changes instituted in January 1985. The nature
and effect of these changes are discussed in detail in
“ Changes in Estimation Procedure in the Current
Population Survey Beginning in January 1985” in the
February 1985 issue of Employment and Earnings.
The controls by State for the civilian noninstitutional
population 16 years and over are an arithmetic extrapola
tion of the trend in the growth of this segment of the
population from the April 1, 1980, census through the
latest available July 1 estimate, adjusted as a last step to
a current estimate of the U.S. population of this group.
The “ inflation-deflation” method is used in the
preparation of the independent national controls used for
the age-sex-race groups in the third step of the secondstage ratio estimation procedure. With the “ inflationdeflation” method, the independent controls are prepared
by inflating the 1980 census counts to include estimated
undercounts by age, sex, and race, aging this population
forward to each subsequent month and later age by
adding births and net migration and subtracting deaths.
These postcensal population estimates are then deflated
to census level to reflect the pattern of net undercount
in the most recent census by age, sex, and race. Because
an estimate of undercount is first added and then sub
tracted, the size of each race-sex group is unaffected by
the “ inflation-deflation” method. Similarly, the final
estimate is affected only by the age structure of the under
count, but not the level. This feature of the method is
important since the exact amount of undercount in the
1980 census remains unknown.
Data on births and deaths between April 1, 1980, and
the estimate date are based on tabulations of vital sta
tistics for the resident population made by the National
Center for Health Statistics and data on deaths of military
personnel overseas from the Department of Defense.
Estimates of net civilian immigration are based on data
provided by the Immigration and Naturalization Service,
the Department of Defense, the Office of Personnel
Management, and the Puerto Rico Planning Board. The
civilian noninstitutional population is derived by subtract
ing the Armed Forces and the institutional population for
Composite estimate. The last step in the preparation of
estimates makes use of a composite estimating procedure.
The composite estimate for the CPS is a weighted average
of the noncomposited estimate for the current month
and of the composite estimate for the previous month,
adjusted for the net month-to-month change (based on
the continuing 75 percent of the households in the sam
ple from the previous month). Also included is an addi
tional term which is an estimate of the net difference
between incoming and continuing parts of the current
month’s sample.
The composite estimate results in a reduction in the
sampling error beyond that which is achieved after the
two stages of ratio estimates described; for some items,
the reduction is substantial. The resultant gains in
reliability are greatest in estimates of month-to-month
change, although gains are also usually obtained for
estimates of level in a given month, change from year to
year, and change over other intervals of time.
Presentation and Uses
The CPS provides a large amount of detail on the
economic and social characteristics of the population. It
is the source of monthly estimates of total employment,
both farm and nonfarm; of nonfarm self-employed per
sons, domestics, and unpaid helpers in nonfarm family
enterprises, as well as wage and salaried employees; and
of total unemployment, whether or not covered by unem
ployment insurance. It is a comprehensive source of infor
mation on the personal characteristics such as age, sex,
race, Hispanic origin, educational attainment, and the
marital and family status of the total civilian population
8
(not in institutions) 16 years of age and over and of the
employed, the unemployed, and those not in the labor
force.
The survey provides distributions of workers by the
number of hours worked, as distinguished from aggregate
or average hours for an industry, permitting separate
analyses of part-time workers, workers on overtime, etc.
It is a comprehensive current source of information on
the occupation of workers, whether teachers, stenog
raphers, engineers, laborers, etc.; and the industries in
which they work. It also provides data on the usual
weekly earnings of wage and salary workers, which are
published on a quarterly basis because the monthly detail
is collected from only a quarter of the sample (the two
“ outgoing” rotation groups).
Information is available from the survey not only for
persons currently in the labor force but also for those who
are outside of the labor force, some of whom may be con
sidered to be a “ labor reserve.” The characteristics of
such persons—whether married women with or without
young children, disabled persons, students, retired
workers, etc.—can be determined. Also, through special
inquiries, it is possible to obtain information on their
skills and past work experience.
Each month, the employment and unemployment data
are published initially in The Employment Situation news
release 2 weeks after they are collected. The release in
cludes a narrative summary and analysis of the major
employment and unemployment developments together
with tables containing statistics for the principal data
series. Subsequently, more detailed statistics are published
in Employment and Earnings. Labor force data are avail
able in machine-readable form and on b l s data diskettes.
The detailed tables in this periodical provide information
on the labor force, employment, and unemployment by
a number of characteristics, such as age, sex, race, marital
status, industry, and occupation. Estimates of the labor
force status of selected population groups not published
on a monthly basis, such as poverty and nonpoverty
residents of the Nation’s metropolitan and nonmetro
politan areas, special data for Vietnam-era veterans, etc.,
are published every quarter. Additionally, data are pub
lished quarterly on employment and unemployment by
family relationship and on median weekly earnings classi
fied by a variety of characteristics. Approximately 250 of
the most important estimates from the CPS are presented
each month on a seasonally adjusted basis.3 Over 20,000
of the most important monthly labor force data series plus
quarterly and annual averages are maintained on a onereel tape. In many cases, these data are available from the
inception of the series through the current month.
The CPS is used also for a program of special inquiries
to obtain detailed information from particular segments,
or for particular characteristics of the population and
labor force. About four such special surveys are made
each year. The inquiries are repeated annually in the same
month for some topics, including the earnings and total
incomes of individuals and families (published by the
Bureau of the Census); the extent of work experience of
the population during the calendar year; the marital and
family characteristics of workers; the employment of
school age youth, high school graduates and dropouts,
and recent college graduates; and the educational attain
ment of workers. Surveys have been made periodically
on subjects such as job mobility, job tenure, job-search
activities of the unemployed, displaced workers, and work
schedules.
Generally, the persons who provide information for
the monthly CPS questions also answer the supplemen
tal questions. Occasionally, the kind of information
sought in the special survey requires the respondent to
be the person about whom the questions are asked.
Information obtained through the supplemental ques
tions is combined with data in the regular questionnaire
to provide tabulations of all the desired personal and
economic characteristics of the persons in the special
survey. Reports on these special surveys are first pub
lished as news releases and subsequently in the Monthly
Labor Review.4
In addition to the regularly tabulated statistics described
above, special data can be generated through the use of
the CPS individual record (micro) tapes. These tape files
contain records of the responses to the survey question
naire for all individuals in the survey. While the tapes
can be used simply to create additional cross-sectional
detail, an important feature of their use is the ability to
match the records of specific individuals at different
points in time during their participation in the survey.
By matching these records, data files can be created which
lend themselves to some limited longitudinal analysis and
the investigation of short-run labor market dynamics.
While a number of technical difficulties lie in the path
of more complete utilization of these data files for the
purposes of longitudinal analysis, this area is continually
being investigated and holds considerable promise.
3 Since 1980, the X -ll ARIMA seasonal adjustment method has been
used to seasonally adjust labor force data. For a detailed description
of the X -ll ARIMA method, see Estela Bee Dagum, The X - l l ARIMA
Seasonal A dju stm en t M ethod, Statistics Canada Catalogue No. 12-564E,
January 1983.
4
Historical data through 1981 for many of the CPS regular mon
thly data series as well as those derived from the supplemental ques
tions were published in L a b o r F orce Statistics D erived From the C ur
rent P opu lation Survey: A D a ta b o o k , Volumes I and II, Bulletin 2096
(U.S. Department of Labor, Bureau of Labor Statistics, 1982).
Limitations
Geographic. Although the present CPS sample is a
State-based design, the CPS continues to produce reliable
national monthly estimates. The sample does not permit
9
the production of reliable monthly estimates for all States.
Subnational data from the CPS are published monthly for
11 large States and annually for all States, 30 large metro
politan areas, and selected central cities. The production
of subnational labor force and unemployment estimates
is discussed in more detail in chapter 4 of this bulletin.
Sources o f errors in the survey estimates. The estimates
from the survey are subject to sampling errors, that is,
errors arising from the fact that the estimates each month
are based on information from a sample rather than the
whole population. In addition, as in any survey, the
results are subject to errors made in the field and in the
process of compilation.
Classification errors in labor force surveys may be par
ticularly large in the case of persons with marginal attach
ments to the labor force. These errors may be caused by
interviewers, respondents, or both, or may arise from
faulty questionnaire design. In spite of a continuous
quality control program, interviewers may not always ask
the questions in the prescribed fashion. To the extent that
varying the wording of the question causes differences
in response, errors or lack of uniformity in the statistics
may result. Similarly, the data are limited by the adequacy
of the information possessed by the respondent and the
willingness to report accurately.
The estimates from the survey also are subject to vari
ous other types of errors. Some of these are:
Nonresponse—about 4 to 5 percent of occupied units
are not interviewed in a typical month because of tem
porary absence of the occupants, refusals to cooperate, or
other reasons. Although an adjustment is made in weights
for interviewed households to account for noninterviews,
they still represent a possible source of bias. Similarly,
for a relatively few households, some of the information
is omitted because of lack of knowledge on the part of
the respondent or because of interviewer error. In process
ing the completed questionnaires, entries usually are sup
plied for omitted items on the basis of the distributions
of these items for persons of similar characteristics.
Independent population estimates—the independent
population estimates used in the estimation procedure
may be a source of error although, on balance, their use
substantially improves the statistical reliability of many
of the figures. (See Ratio estimates.) Errors may arise in
the independent population estimates because of under
enumeration of certain population groups or errors in age
reporting in the last census (which serves as the base for
the estimates) or similar problems in the components of
population change (mortality, immigration, etc.) since
that date.
Processing errors—although there is a quality control
program on coding and a close control on all other phases
of processing and tabulation of the returns, some process
ing errors are almost inevitable in a large statistical opera
tion of this type. However, the net error arising from
processing is probably fairly negligible.
Measuring the accuracy o f results. Modern sampling
theory provides methods for estimating the range of
errors due to sampling where, as in the CPS sample, the
probability of selection of each member of the popula
tion is known. Methods are also available for determin
ing the effect of response variability in the CPS. A
measure of sampling variability indicates the range of dif
ferences that may be expected because only a sample of
the population is surveyed. A measure of response
variability indicates the range of difference that may be
expected as a result of compensating types of errors
arising from practices of different interviewers and the
replies of respondents; these would tend to cancel out in
an enumeration of a large enough population. In practice,
these two sources of error—sampling and response
variability—are estimated jointly from the results of the
survey. The computations, however, do not incorporate
the effect of response bias, that is, any systematic errors
of response. Response biases occur in the same way in
a complete census as in a sample, and, in fact, may be
smaller in a well-conducted sample survey where it may
be feasible to collect the information more skillfully.
Estimates of sampling and response variability com
bined are provided in Employment and Earnings and in
other reports based on CPS data, thus permitting the user
to take this factor into account in interpreting the data.
In general, the smaller figures and small differences be
tween figures are subject to relatively large variation and
should be interpreted with caution.
Estimation of response bias is one of the most difficult
aspects of survey and census work. In many instances,
available techniques are not sufficiently precise to provide
satisfactory estimates. Continuing experimentation is car
ried out with the aim of developing more precise measure
ments and improving the overall accuracy of the series.
10
Technical References
National Commission on Employment and Unemployment
Statistics. Counting the Labor Force, 1979.
A comprehensive review of the entire labor market data
system; includes an appraisal of current concepts and
methodology and recommendations for further research
and improvements.
President’s Committee to Appraise Employment and Un
employment Statistics. Measuring Employment and
Unemployment, 1962.
A review of all Federal statistical series on employment
and unemployment and a comparison of the sources and
uses of each series; includes a brief history of the develop
ment of labor force statistics, an evaluation of concepts
and techniques, and recommendations for further
research and improvements.
U.S. Department of Commerce, Bureau of the Census. “ The
Current Population Survey: Design and Methodology,”
Technical Paper No. 40, 1978.
A comprehensive description of the c p s , based on the
design following the 1970 census.
U.S. Department of Commerce, Office of Federal Sta
tistical Policy and Standards. “ An Error Profile:
Employment as Measured by the Current Population
Survey,” Statistical Policy Working Paper 3, 1978.
A description of the potential sources of error in the
c p s as they affect the national employment statistics.
Bureau of Labor Statistics. Employment and Earnings,
Explanatory Notes, monthly.
An up-to-date, concise description of the concepts and
methods used in the labor force data from the Current
Population Survey. Provides tables which present the
sampling errors for labor force series.
Bureau of Labor Statistics. A Guide to Seasonal Adjustment
o f Labor Force Data, Bulletin 2114, 1982.
A description of the concepts and techniques used in
seasonally adjusting labor force statistics from the Cur
rent Population Survey.
Bureau of Labor Statistics. Labor Force Statistics Derived
From the Current Population Survey: A Databook,
Volumes I and II, Bulletin 2096, 1982.
A compilation of historical data through 1981 for many
of the data series obtained from the Current Population
Survey.
Bureau of Labor Statistics. How the Government Measures
Unemployment, Report 742, 1987.
A short, nontechnical discussion of the concepts and
methods used in obtaining labor force statistics from the
Current Population Survey.
Bureau of Labor Statistics. “ Redesign of the Sample for the
Current Population Survey,” Employment and Earnings,
May 1984, pp. 7-10.
A description of the c ps redesign following the 1980
census and the change in the structure of the c p s sam
ple from a national sample to a State-based design.
Bureau of Labor Statistics. Technical Description o f the
Quarterly Data on Weekly Earnings from the Current
Population Survey, Bulletin 2113, 1982.
A description of the collection, processing, and
reliability of the weekly and hourly earnings data obtained
from the c p s .
11
18. LIN E N U MBER
19. What was . . . doing most
of LAST W EEK -
or something else?
Working (Skip to 2 0 A ). . . . WK
With a job but not at work . . J
during the past 4 weeks?
(Note: i f farm or business
was temporarily absent or
O
No
O
*
q
did . . . work
O
LAST WEEK
8 8
at all jobs?
3
O
O
Going to school........................ S
O
Unable to work (Skip to 2 4 ).. U
O
Retired.........................................R
O
Other (Specify)....................... OT
O
20B. IN T E R V IE W E R
CHECK ITEM
49+ O
1 -3 4
(Skip to
item 23)
O
3 5 -4 8
fG o to
20C)
O
j
5
j
5
G6
??
8 8
99
(Go to 20D)
within 30 days
Yes
O
O
(Under 3 0 days)
Indefinite layoff
(30 days or more
or no def. recall
date) ..................
or slack work?
How many hours
did . . . take off? 7
O
/(Skip
to
\2 2 C 3 )
O
Placed or answered ads..............
O
Nothing (Skip to 2 4 ) ..................
O
Other (Specify In notes, e.g,
JT P A , union or prof,
register, e tc .)....................
O
•
Lost jo b ....................................
O
Quit j o b ....................................
O
•
Left school...............................
O
•
Change in home
•
Left military s e rv ice............
O
•
Other (Specify in notes). . . .
O
. . . U S U A L LY works
not already deducted;
or salary from his/her
i f 20A reduced below 35,
em ployer for an y of the
correct 20B and fill 20C)
q
q
j
j
for work?
No
O
O
No
O
-V-
(Mark the appropriate reason)
for work?
20E. Did . . . work any overtime
Slack w o r k ..................................
O
Material shortage.........................
O
Plant or machine repair..............
O
was . . . laid off?
35 hours or more a week
New job started during week . .
O
Job terminated during w e e k .. .
O
Could find only part-time work
O
C
How many extra
hours d id . . .work?
at this job?
Yes
No
(Correct 20A and 20B as
Labor d ispute..............................
O
O
Own illness...................................
O
On vacation...................................
O
O
O
Did not want full-tim e work. . .
O
held last week)
■
■
■
OCC U PA TION
0 0O
0 0
I I
3
O
1
5
3
5
Ref.
Ref.
Unc.
O
5
G G G
USU A LLY
work at this job?
No
O
Personal, family
find, pregnancy) or school.
Health...................................................
O
Retirement or old a g e .....................
O
25B. Is . . . paid by the hour
on this job?
Seasonal job com pleted...................
O
Slack work or business conditions
O
Temporary
nonseasonal job completed. . .
Unsatisfactory work
arrangements (Hours, pay, etc.)
O
O th er....................................................
O
Temporary illness___
O
O
O
22F. When did . . . last work at a
full-tim e job or business lasting
2 consecutive weeks or more?
(Month)
Yes
O
(Go to 25C)
No
O
(S k ip to 2 5 D )
25C. H ow much
Dollars
d o es. . .
Y e s.............................
O
(
Maybe - it depends O
(Specify in notes)
N o ............................. O
(
D on't k n o w ............
I
O
o
0
O
Cents
O 0
II
II
3 3
3 3
per hour?
24C. Does . . . want a regular job now,
either fu ll-o r part-time?
5 5
G G
(Go to 24D)
D
(Skip to 24E)
24D . What are the reasons . . is not
looking for work?
(Mark each reason m. ntioned)
O
25D . How much does . . .
U S U A LLY
earn per week
© 0 0 0
?
at this job
C ouldn't find any w o rk ......................
O
Lacks nec. schooling,
training, skills or experience
...
O
Employers
think too young or too o ld..........
O
Other pers. handicap in finding job
O
Can't arrange child care.....................
O
BEFORE
deductions?
Include any
•
In school or other training.................
O
•
III health, physical d isability.............
O
•
Other (Specify in notes)......................
O
•
D on't k n o w ...........................................
O
?
T
5 5 5
overtime pay,
G G G
commissions,
?
or tips usually
f l
T
8 8 8
3 3 3
received.
?
?
9 9 9
REF
. .. O
Family responsibilities..........
9 9
REF
(Ask^25D)
O
Going to school............
2
3
24B. Why did . . . leave that job?
W hy not?
/
;
3
3 5
(Skip to
24C)
Never worked.............
®
Other (Specify in notes)
3
9
Unc.
O
W ithin last 12 months (Specify) . .
\
9
Part
/-
per week does . . .
4 up to 5 years ago. .
. have taken a job LA ST
Already has a job.........
(Skip to 23)
Other reason (Specify).
o>
|
up to 2 years ago .
W EEK if one had been offered?
(Skip to 23 and enter job
III
Too busy with housework,
school, personal bus., etc. . .
O
Yes
not already included and
lo
O
22E . Could .
necessary i f extra hours
skip to 23.)
Bad weather.................................
O
\
22D . Has . . . been looking for full-tim e
or part tim e work?
Full
/
O
o
21C. Does . . . usually work
LAST WEEK?
O
1
^
? 3?
3) How many weeks ago
or at more than one job
Yes
^
1 or 5 (Go to 25 A )
2 up to 3 years ago .
5
did . . . start looking
Yes
2, 3, 4, 6, 7, 8 (Skip to 26)
O
3 up to 4 years ago .
3 3
or or
2) How many weeks ago
tim e o ff LA ST WEEK?
less than 35 hours
C
25 A . H ow many hours
W ithin past 12 months
O
has . . . been looking
■
number is:
part-time?
5 or more years ago .
•
21B . Is . . . receiving wages
(Correct 20A I f lost time
a week?
Full-time work week
under 35 h o u rs .....................
O
22C. 1) How many weeks
What is the reason
Holiday (Legal or religious) . . . .
employer directly. . .
or family responsibilities
Other (Specify) . .
. . . worked less than
No
O
e W anted temporary w o r k .. O
What is the reason
35 hours LAST WEEK?
O
pvt. employ agency
som e other reason?
\
First digit o f SE G M EN T
regular job or business, either fu ll-o r
lost or quit a job or w a s there
22B and
22C
(Rotation number)
2, 3, 4, 6, 7, 8 (Skip to 26)
O 1 or 5 (Go to 24A )
------------------------- / ----------------------------
for w o rk, w as it because he/she
fSklp to
Temporary layoff
such as illness, holiday
O
O
24A . When did . . . last work for pay at a
22 B . A t th e tim e . . . started looking
New job to begin
25. IN T E R V IE W E R C H EC K ITEM
First digit o f SE G M EN T number is:
(Go to 24)
friends or relatives . . O
W EEK for any reason
Yes
w it h -
3
take any tim e o ff LAST
hours or more a week at this job?
O
Checked pub employ, agency
work LAST WEEK?
20D. Did . . . lose any time or
20C. Does . . . U S U A L L Y work 35
No
4 weeks to find work? (Mark ah
methods used; do not read list.)
21 A. Why was . . . absent from
q
O
(Rotation number)
22A . What has . . been doing in the last
Yes O
No O (Go to 22)
/ --------------------------------------------
(Go to 21)
O
Keeping house...........................H
j — Yes
y'
/
on layoff LAST WEEK?
^
20A . How many hours
Looking for w ork...................LK
\
Has . . . been looking for work
businen from which he/she
^es
j Going to school
2 4 .IN T E R V IE W E R CHECK ITEMj
Did . . . have a job or
work around the house?
unpaid work.)
I Keeping house
22. ( I f L K In 19, Skip to 22 A .)
LAST W EEK, not counting
operator In hh., ask about
. Working
f
21. ( I f j in 19, skip to 21 A .)
20. Did . . . do any work at all
O
25E. On this job, is . . . a member
of a labor union or of an
O
employee association similar
to a union?
f
One to five years ag o.......................
O
More than 5 years ago.....................
Never worked
full-tim e 2 wks. or m ore..............
O
Never worked at a l l .........................
O
O
(SK IP to 23. I f layo ff entered in 21 A,
enter job, either full or part time, from
which laid off. Else enter last full time
job lasting 2 weeks or more, or
"never worked.")
24E. Does . . . intend to look for work
o f any kind in the next 12 months?
Yes
O
(Skip to 26)
No
O
(Ask 25 F)
25 F . On this job, is . . . covered
by a union or employee
It depends (Specify in notes)
asso ciatio n co n tract?
O
D on't know ..........................
( I f entry in 24B, describe jo b in 23,
otherwise, skip to 26)____________
Yes
O
{
No
O
l
(Go to 26)
23. D ESC R IP TIO N OF JOB OR BUSINESS
23A . For whom did . . . work? (Name o f company, business, organization or other employer.)
23E. Was this person
23B. What kind of business or industry is this? (For example: T V and radio m fg, retail shoe store, State Labor Dept., farm.j\
23F. IN T E R V IE W E R
CHECK ITEM
An employee of a P R IV A T E Co,
bus., or individual for wages, salary or comm. . . P
O
A F E D E R A L government em ployee............................. F
O
A STA TE government em ployee................................... S
O
A LOC A L government employee................................... L
O
/
(Go to
23C. What kind of work was . . . doing? (F or example: electrical engineer, stock clerk, typist, farmer.)
Self-empl. in OWN bus., prof, practice, or farm
23D . What were . . .'s most im portant activities or duties at this job? (For example: types, keeps account books, files,
sells cars, operates printing press, finishes concrete.)
| Yes....................... I
Is the business incorporated? \
| N o ..................... SE
O
_
O
Working W IT H O U T PA Y in fam. bus. or farm. . . .WP
O
N EVE R W O R K E D ......................................................N E V
Page 6
12
■
O
(Skip
to 26)
j
i
Entry (or NA)
in item 20A
l
Entry (or NA)
in item 2 1 B
O
All other cases
O
O
t
/
(Go to 25
■ at top o f
\ P°9‘ )
1
i
!
|
i
(Skip to 26)
Chapter 2. Employment,
Hours, and Earnings from
the Establishment Survey
cooperates with State employment security agen
cies in a survey collecting data each month on employ
ment, hours, and earnings from a sample of nonagricultural establishments (including government). In 1987,
this sample included approximately 290,000 reporting
units. From these data, a large number of employment,
hours, and earnings series in considerable industry and
geographic detail are prepared and published each month.
The employment data include series on all employees,
women workers, and production or nonsupervisory
workers; hours and earnings data include average weekly
hours, average weekly overtime hours, and average
hourly and weekly earnings. For many series, seasonally
adjusted data are also published.
bls
Background
The first monthly studies of employment and payrolls
by b l s began in 1915 and covered four manufacturing
industries. Before 1915, the principal sources of employ
ment data in the United States were the census surveys—
the decennial Census of Population and the quinquen
nial Census of Manufactures. No regular employment
data were compiled between the Census dates.
In 1916, the survey was expanded to cover employment
and payrolls in 13 manufacturing industries; by 1923, the
number had increased to 52, and by 1932, 91 manufac
turing and 15 nonmanufacturing industries were covered
by a monthly employment survey.
With the deepening economic crisis in 1930, President
Hoover appointed an Advisory Committee on Employ
ment Statistics which recommended extension of the
Bureau’s program to include the development of hours
and earnings series. In 1932, Congress granted an increase
in the b l s appropriation for the survey. In 1933, average
hourly earnings and average weekly hours were published
for the first time for total manufacturing, for 90
manufacturing industries, and for 14 nonmanufacturing
categories.
During the Great Depression, there was controversy
concerning the actual number of unemployed people; no
reliable measures of employment or unemployment
existed. This confusion stimulated efforts to develop
comprehensive estimates of total wage-and-salary
employment in nonagricultural industries, and, in 1936,
BLS survey data produced such a figure for the first
time.
Interest in employment statistics for States and areas
also grew. Even before BLS entered the field in 1915,
Massachusetts, New York, and New Jersey were prepar
ing employment statistics. In 1915, New York and
Wisconsin entered into cooperative agreements with b l s ,
whereby sample data collected from employers by a State
agency would be used jointly with b l s to prepare State
and national series. By 1928, five other States had entered
into such compacts, and another five were added by 1936.
By 1940, estimates of total nonagricultural employment
for all 48 States and the District of Columbia were
available.
Since 1949, the Current Employment Statistics (CES)
program has been a fully integrated Federal-State proj
ect which provides employment, hours, and earnings
information by industry on a national, State, and area
basis. BLS has begun a long-range project to improve the
Current Employment Statistics program. The CES revi
sion will assess all aspects of the program at the national,
State, and area levels, from collection and processing of
data through estimation and publication. In 1987,
cooperative arrangements were in effect with all 50 States,
the District of Columbia, Puerto Rico, and the Virgin
Islands.
Concepts
Establishment
An establishment is defined as an economic unit which
produces goods or services, such as a factory, mine, or
store. It is generally at a single location and engaged
predominantly in one type of economic activity. Where
a single location encompasses two or more distinct
activities, these are treated as separate establishments,
provided that separate payroll records are available and
certain other criteria are met.
13
Employment
Employment represents the total number of persons
employed full or part time in nonagricultural establish
ments during a specified payroll period. Temporary
employees are included. In general, data refer to persons
who worked during, or received pay for, any part of the
pay period that includes the 12th of the month, which
is standard for all Federal agencies collecting employment
data from business establishments. However, national
employment figures for Federal Government establish
ments represent the number of persons who occupied
positions on the last day of the calendar month; inter
mittent workers are counted if they performed any serv
ice during the month.
Workers on an establishment payroll who are on paid
sick leave (when pay is received directly from the em
ployer), on paid holiday, or paid vacation, or who work
during only a part of the specified pay period are counted
as employed. Persons on the payroll of more than one
establishment during the pay period are counted in each
establishment which reports them, whether the duplica
tion is due to turnover or dual jobholding. Persons are
considered employed if they receive pay for any part of
the specified pay period, but are not considered employed
if they receive no pay at all for the pay period. Since pro
prietors, the self-employed, and unpaid family workers
do not have the status of paid employees, they are not
included. Domestic workers in households are excluded
from the data for nonagricultural establishments. The
employment statistics for government refer to civilian
employees only.
All persons who meet these specifications are included
in the designation “ all employees.” Major categories of
employees are differentiated primarily to ensure the
expeditious collection of current statistics on hours and
earnings; these groups of employees are designated pro
duction workers, construction workers, or nonsupervisory workers, depending upon the industry.
In manufacturing industries, data are collected for
production workers. This group, in general, covers
employees, up through the level of working super
visors, who are engaged directly in the manufacture of
the product of the establishment. Among those ex
cluded from this category are persons in executive and
managerial positions and persons engaged in activities
such as accounting, sales, advertising, routine office
work, professional and technical functions, and force
account construction. (Force-account construction is
construction work performed by an establishment,
primarily engaged in some business other than construc
tion, for its own account and use by its own employees.)
Production workers in mining are defined in a similar
manner. A more detailed description of the classes of
employees included in the production and nonproduction
worker categories in manufacturing is shown on the
facsimile of the b l s 790 C schedule at the end of this
chapter.
In construction, the term construction workers covers
workers, up through the level of working supervisors,
who are engaged directly on the construction project
either at the site or in shops or yards at jobs ordinarily
performed by members of construction trades. Excluded
from this category are executive and managerial person
nel, professional and technical employees, and workers
in routine office jobs.
In the remaining industries (transportation, com
munication, and public utilities; retail and wholesale
trade; finance, insurance, and real estate; and services),
data are collected for nonsupervisory workers. Nonsupervisory workers include most employees except those in
top executive and managerial positions. (See facsimile of
BLS 790 E, the reporting form for wholesale and retail
trade.)
An employment benchmark is defined as a reasonably
complete count of employment used to adjust estimates
derived from a sample. Adjustment is usually done
annually. The basic source of benchmark data for the
Current Employment Statistics program is data collected
from employers by State employment security agencies
as a byproduct of the unemployment insurance (ui)
system. About 98 percent of all employees on nonagri
cultural payrolls are covered by the ui system. The com
pilation and use of benchmark data are explained in detail
in later sections of this chapter.
Hours and earnings
The hours and earnings series are based on reports of
gross payrolls and the corresponding paid hours for pro
duction workers, construction workers, or nonsuper
visory workers. (See facsimile of BLS 790 C.) (In govern
ment and private educational institutions, payroll data
are for “ all employees.” )
Aggregate payrolls include pay before deductions for
Social Security, unemployment insurance, group
insurance, withholding tax, salary reduction plans, bonds,
and union dues. The payroll figures also include pay for
overtime, shift premiums, holidays, vacations, and sick
leave paid directly by the employer to employees for the
pay period reported. They exclude bonuses (unless earned
and paid regularly each pay period) or other pay not
earned in the pay period concerned (e.g., retroactive pay).
Tips and the value of free rent, fuel, meals, or other pay
ment in kind are not included.
Total hours during the pay period include the hours
worked, overtime hours, hours paid for standby or re
porting time, and equivalent hours for which employees
received pay directly from the employer for sick leave,
holidays, vacations, and other leave. Overtime or other
premium pay hours are not converted to straight-time
equivalent hours. Total hours differ from scheduled
14
hours or hours worked. The average weekly hours derived
from the total hours reflect the effects of such factors
as absenteeism, labor turnover, part-time work, and
strikes.
Overtime hours are hours worked for which premiums
were paid because they were in excess of the number of
hours of either the straight-time workday or workweek.
Saturday and Sunday hours (or 6th and 7th day hours)
are included as overtime only if overtime premiums were
paid. Holiday hours worked as overtime are not included
unless they are paid for at more than the straight-time
rate. Hours for which only shift differential, hazard,
incentive, or similar types of premiums were paid are
excluded from overtime hours. Overtime hours data are
collected only from establishments in manufacturing
industries.
Average hourly earnings series, derived by dividing
gross payrolls by total hours, reflect the actual earnings
of workers, including premium pay. They differ from
wage rates, which are the amounts stipulated for a given
unit of work or time. Average hourly earnings do not
represent total labor costs per hour for the employer,
because they exclude retroactive payments and irregular
bonuses, various fringe benefits, and the employer’s share
of payroll taxes. Earnings for those employees not
covered under the production worker and nonsupervisory
categories are, of course, not reflected in the estimates.
Real earnings data (those expressed in 1977 dollars),
resulting from the adjustment of average weekly earn
ings by means of the Bureau’s Consumer Price Index,
indicate the changes in the purchasing power of money
earnings as a result of changes in prices for consumer
goods and services. These data cannot be used to measure
changes in living standards as a whole, which are affected
by other factors such as total family income, the exten
sion and incidence of various social services and benefits,
and the duration and extent of employment and unem
ployment. The long-term trends of these earnings data
are also affected by changing mixes of full-time/part-time
workers, high-paid/low-paid workers, etc.
Straight-time average hourly earnings are approx
imated by adjusting average hourly earnings by elimi
nating only premium pay for overtime at a rate of time
and one-half. Thus, no adjustment is made for other
premium payment provisions such as holiday work, lateshift work, and premium overtime rates other than at time
and one-half.
Industrial classification
Industrial classification refers to the grouping of re
porting establishments into industries on the basis of their
major product or activity as determined by the establish
ments’ percent of total sales or receipts for the previous
calendar year. This information is collected as an
administrative byproduct of the ui reporting system. All
data for an establishment making more than one product
or engaging in more than one activity are classified under
the industry of the most important product or activity,
based on the percentages reported.
Data are currently classified in accordance with the
Standard Industrial Classification Manual, Office of
Management and Budget, 1972, as modified by the 1977
Supplement. (See appendix B of this bulletin for a
description of this system.) Beginning in 1989, data will
be reclassified in accordance with the 1987 SIC manual.
Data Sources and Collection Methods
Sample data
Each month, the State agencies cooperating with BLS
in the survey collect data by mail on employment,
payrolls, and hours paid for, from a sample of estab
lishments. The respondents extract these data from their
payroll records, which must be maintained for a variety
of tax and accounting purposes. Despite the voluntary
nature of the survey, numerous large establishments have
reported regularly for many years.
A “ shuttle” schedule is used (BLS form 790 series), that
is, one which is submitted each month by the respondent,
edited by the State agency, and returned to the respond
ent for use again the following month. The shuttle
schedule has been used since 1930, but there have been
substantial changes in its design and in the data collected.
A major redesign was completed in 1986 and introduced
with 1987 data collection.
All aspects of the schedule—its format, the wording
of the requested items and definitions, and the concepts
embodied therein—are subjected to a continuing review,
not only by b l s and the State agencies, but also by other
government agencies, private business, and labor
organizations. The report forms are basically alike for
each industry, but there are several variants tailored to
the characteristics of different industries.
The technical characteristics of the shuttle schedule are
particularly important in maintaining continuity and con
sistency in reporting from month to month. The shuttle
design automatically exhibits the trends of the reported
data during the year covered by the schedule, and
therefore, the relationship of the current data to the data
for the previous months. The schedule also has opera
tional advantages; for example, accuracy and economy
are obtained by entering the identification codes and the
address of the reporter only once a year.
All schedules are edited by the State agencies each
month to make sure that the data are correctly reported
and that they are consistent with the data reported by the
establishment in earlier months and with the data reported
by other establishments in their industry. This editing
15
process is carried out in accordance with detailed instruc
tions from b l s . The State agencies use the information
provided on the forms to develop State and area estimates
of employment, hours, and earnings, and then forward
the data, either on the schedules themselves or in machinereadable form, to b l s - Washington. At b l s , they are
edited again by computer to detect processing and report
ing errors which may have been missed in the initial State
editing. Questionable reports discovered at any stage of
the editing process are returned, if necessary, to the
respondent for review and correction. When all questions
have been resolved, the data are used to prepare national
estimates.
Benchmark data
Since about 1940, the basic source of benchmark infor
mation for “ all employees” has been the periodic tabula
tions compiled by State employment security agencies
from reports of establishments covered under State ui
laws.
The State employment security agencies receive
quarterly reports from each employer subject to the ui
laws showing total employment in each month of the
quarter, and the total quarterly wages for all employees.
The State agencies submit tabulations of these reports to
b l s - Washington each quarter. (See chapter 4.)
For the few industries exempt from mandatory ui
coverage, other sources are used for benchmark infor
mation. For example, data on employees covered under
Social Security laws, published by the Bureau of the
Census in County Business Patterns, are used to augment
the Ui data for nonoffice insurance sales workers. Data
for interstate railroads are obtained from the Interstate
Commerce Commission. Employment figures for
religious organizations are obtained from data provided
by the National Council of Churches, the Bureau of the
Census, and special surveys conducted by the State
agencies.
In benchmarking the Federal Government sector, b l s
uses monthly employment data compiled by the Office
of Personnel Management. The ui data for State and
local government employment are supplemented as
necessary with Bureau of the Census data derived from
the Census of Governments for local elected officials and
certain other groups.1
Sample Design
Sampling is used by BLS in the Current Employment
Statistics survey to collect data in most industries, since
full coverage would be prohibitively costly and time
1 For a more detailed description of the benchmarks, see Fred R.
Cronkhite, “ BLS Establishment Estimates Revised to March 1986
Benchmarks,” E m p lo ym en t an d Earnings, June 1987, pp. 6-23.
consuming. The sampling plan for the program, must:
(a) provide for the preparation of reliable monthly
estimates of employment, hours of work, and weekly and
hourly earnings, which can be published promptly and
regularly; (b) through a single, general system, yield con
siderable industry detail for metropolitan areas, States,
and the Nation; (c) be appropriate for the existing frame
work of operating procedures, administrative practices,
resource availability, and other institutional characteris
tics of the program; and (d) be efficient, that is, provide
maximum accuracy at minimum cost.
The primary sampling design is “ optimum allocation,”
which produces an efficient and equitable sample distribu
tion by stratifying the universe of establishments into
homogeneous groups. The strata are arranged according
to industry and size characteristics. Under optimum allo
cation, a larger sample is usually required for a size
stratum if the stratum has a greater number of units in
the universe or if it has a high degree of variability. The
optimum number of establishments to be included in each
size stratum of the national c e s sample is determined by
the number of establishments in a stratum’s universe and
the standard deviation of the establishments in that
universe.
A specific form of optimum allocation, called alloca
tion proportional to employment, is used in the CES
survey. This requires that the universe of establishments
for each industry be stratified into employment-size
classes. Then a total sample size sufficient to produce
satisfactory employment estimates is determined and
distributed among the size classes in each industry on the
basis of the average employment per establishment and
the relative importance of each size class to its industry.
In practice, this amounts to distributing the total number
of establishments needed in the sample among the cells
on the basis of the ratio of the employment in each cell
to the total employment in the industry.
The likelihood that a certain establishment will be
selected depends upon its employment level. Large
establishments are certain of selection; smaller ones have
less chance. Within each cell, sample members are selected
at random. Sampling ratios are determined in order to
aid this selection process. In nearly all industries,
establishments with 250 or more employees are included
in the sample with certainty; in many industries, the
cutoff is lower. In a manufacturing industry in which a
high proportion of total employment is concentrated in
a relatively few large establishments, a high percentage
of total employment is included in the sample. Conse
quently, the sample design for such industries provides
for a complete census of the large establishments with
only a few chosen from among the smaller establish
ments. On the other hand, in an industry where a large
proportion of total employment is in small establish
ments, the sample design calls for the inclusion of all large
establishments, and also for a substantial number of the
16
smaller establishments. Many industries in the trade and
services divisions fall into this category.
This sample design, although aimed primarily at meet
ing the needs of the national program, provides a tech
nical framework within which State and area needs can
be met. It incorporates the trends in all size classes,
reduces geographic bias, and reduces large-firm bias by
giving smaller firms proper representation in the sample.
Since the estimates for States and areas generally are not
prepared at the same degree of industry detail as the
national estimates, it may be necessary to modify the na
tional sampling ratios in order to obtain a sufficient sam
ple. The additional reports needed for State and area
samples are added to the sample required by the national
design.
Estimating Procedures
Employment
Employment estimates are made at what is termed the
basic estimating cell level and aggregated upward to
broader levels of industry detail by simple addition. Basic
cells are defined by industry (usually at the 3- or 4-digit
sic level) and in some cases are stratified within industry
by geographic region and/or size class. Within the whole
sale trade, retail trade, and services divisions, most
industries are stratified into five size classes (beginning
in 1984) because research demonstrated that estimates
produced under this scheme require less benchmark revi
sion. (See earlier section on benchmarks.) For other divi
sions, size and region strata are used when they improve
the hours and earnings estimates.
To obtain “ all employee” estimates for a basic esti
mating cell, the following three steps are necessary:
1. A total employment figure (benchmark) for the
basic estimating cell as of a specified month (usually
March) is obtained.
2. For each cell, the ratio of all employees in 1 month
to all employees in the preceding month (i.e., the link
relative) is computed for sample establishments which
reported for both months.
3. Beginning with the benchmark month, the all
employee estimate for each month is obtained by
multiplying the all-employee estimate for the previous
month by the link relative for the current month.
Application of the estimating procedure in preparing
a series is illustrated by the following example. Assume
that the estimate of all employees for a given cell was
50,000 in July. The reporting sample, composed of 60
establishments, had 25,000 employees in July and 26,000
in August, a 4-percent increase. To derive the August
estimate, the change for identical establishments reported
in the July-August sample is applied to the July estimate:
50,000 X 26,000 (or 1.04) = 52,000
25,000
This procedure, known as the link relative technique,
is efficient in that it takes advantage of a reliable, com
plete count of employment and of the high correlation
between levels of employment in successive months in
identical establishments.
Most employment estimates are multiplied by bias
adjustment factors to produce the monthly published
estimates. Bias adjustment factors are used to compen
sate for the inability to capture the entry of new firms
on a timely basis, and other biases of the survey method.
The bias factors are derived based on a 3-year average
of differences between benchmarks and sample estimates,
and the rate of employment change in the most recent
quarter.
To obtain estimates of production, construction, or
nonsupervisory worker employment, the sample ratio of
production workers to all employees for the current
month is used. For example, the 60 sample firms which
had 26,000 employees in August reported an August
production-worker figure of 19,500, resulting in a ratio
of 19,500/26,000 or 0.750. Using this ratio, the number
of production workers in August is estimated to be 39,000
(52,000 multiplied by 0.750 = 39,000). A similar ratio
method is used to estimate the number of women
workers.
If permanent changes in the composition of the sam
ple occur, the “ production worker to all employee” ratios
and the “ women worker to all employee” ratios calcu
lated from the sample data are modified using the wedg
ing technique described in the hours and earnings section.
The estimates for each type of series (all employees,
production workers, and women workers) for individual
basic estimating cells are summed to obtain the corre
sponding totals for broader industry groupings and
divisions.
All estimates back to the previous benchmark month
are subject to revision each year when new benchmarks
become available. Because of the complexity of develop
ing benchmarks, they are not available until at least 15
months after the benchmark month (usually March). For
example, the revised estimates based on the March 1986
benchmarks were released in June 1987. The revision
period extended from April 1985 through February 1986.
To determine the appropriate revisions, the new bench
marks for March are compared to the estimates for that
month based on the previous benchmarks. The dif
ferences represent estimating errors that accumulated
since the previous benchmark revision. These differences
are assumed to have accumulated at a regular rate. The
all-employee estimates are wedged, or tapered, in order
to smooth out the differences between the new and old
benchmarks. Estimates subsequent to the benchmark
month are revised by applying the sample link relative
17
to the new benchmark level. Estimates for women
workers and production workers are recomputed using
the revised all-employee estimates.
Although most national all-employee series are
adjusted by this wedging technique, in some cases the CES
estimates are replaced by the benchmark source figures
if this results in more accurate levels and trends. (In many
States, the replacement technique predominates.)
A comparison of the national revisions made in recent
years is presented in table 1.
Table 1. Percent differences between nonagricultural employ
ment estimates and benchmarks by industry, March 1984-86
Industry
1984
1985
1986
Total ...............................................
Mining .......................................................
Construction.............................................
Manufacturing...........................................
Transportation and public utilities........
0.4
-1.6
3.1
-.9
.2
(1)
-3.1
1.4
-.5
-1.0
-0.5
-1.2
-.6
-1.1
-.3
Wholesale trade.......................................
Retail trade.................................................
Finance, insurance, and real estate.. . .
Services.....................................................
Government...............................................
.5
1.6
.4
.4
.1
-.5
-.2
.5
.1
.8
-1.9
-.5
-.1
.2
-.4
1 Less than 0.05 percent.
Hours and earnings
Independent benchmarks are not available for the
hours and earnings series. Consequently, the levels are
derived directly from the CES sample averages.
Since 1959, when all-employee benchmark data
stratified by employment size became available, estimates
have been prepared using a cell structure which makes
use of size and, in some cases, regional stratification.
In preparing the estimates, the nine standard size classes
are usually combined into no more than three size classes
when stratification by size is needed.
Size classes are combined because the preliminary
estimates are based on only partially reported samples.
Experience indicates that estimates of hours and earnings
prepared from the CES sample using a maximum of three
size strata generally do not differ significantly from those
computed with four or more size strata.
At the same time that the national benchmark revisions
for employment are made, national estimates of average
weekly hours and average hourly earnings are prepared
using eight size strata and four regional strata (Northeast,
Midwest, South, and West). These estimates are used as
a standard against which the published averages are com
pared. If this comparison indicates that modification of
the stratification pattern is needed, a change is introduced
into the estimating cell structure at the time of the next
benchmark revision.
In the wholesale trade, retail trade, and services divi
sions, size stratification based on employment estimation
requirements takes precedence over stratification for
hours and earnings purposes; hence, the procedures
described above are not used for these divisions. The cell
stratification developed for employment estimates is also
used for hours and earnings estimates.
A verage weekly hours and average hourly earnings. To
obtain average weekly hours for a basic estimating cell,
the sum of the worker hours reported by the establish
ments classified in the cell is divided by the total number
of production workers reported for the same establish
ments. In computing average hourly earnings, the
reported payroll is divided by the reported worker hours.
The first estimates, which equal the unmodified sam
ple averages, of average hourly earnings and average
weekly hours are modified at the basic estimating cell level
by a wedging technique designed to compensate for
changes from month to month in the sample of report
ing establishments.
For example, a first estimate of average hourly earn
ings for the current month, xc, is obtained from aggre
gates from a matched sample of establishments reporting
for both the current month and the previous month.
Similarly, a first estimate of average hourly earnings for
the previous month, xp , is calculated from the same
matched sample. xc - xp is a measure of the change be
tween the 2 months.
Note is then taken of the final estimate of average
hourly earnings for the previous month, Xp . Because the
panel of establishments reporting in the sample is not
fixed from month to month, there may be differences
between_Xp and x^. A final estimate for the current
month, Xc, is obtained by making use of both pieces of
information:
Xc = (0.9 Xp -f0.1xp) + (xc-xp)
The procedure reflected in this formula has the follow
ing advantages: (1) it uses matched sample data; (2) it
tapers the estimate for the previous month (Xp) towards
the sample average for the previous month of the cur
rent matched sample (xp); and (3) it promotes continuity
by heavily favoring the estimate for the previous month
(Xp) when applying the numerical factors.
The results of the formula may be modified if the dif
ference between Xp and xp is too great. This is done by
changing the numerical factors from 0.9 and 0.1 to 0.8
and 0.2, or 0.7 and 0.3, etc., or by using a special wedg
ing procedure when the difference exceeds 3 percent in
the same direction for 3 consecutive months.
Average weekly hours and average hourly earnings for
industries and groups above the basic estimating cell level
are weighted averages of the figures for component cells.
The average weekly hours for each basic estimating cell
are multiplied by the corresponding estimate of the
number of production workers to derive aggregate worker
hours. Payroll aggregates are the product of the aggregate
18
worker hours and average hourly earnings. The payroll
and worker hour aggregates for industry groups and divi
sions represent the sum of the aggregates for the compo
nent industries.
Average weekly hours for industry groups are obtained
by dividing the worker hour aggregates by the corre
sponding production worker estimates. Average hourly
earnings for industry groups are computed by dividing
the payroll aggregates by the worker hour aggregates.
This method is equivalent to weighting average weekly
hours by the estimated number of production workers
in the universe and weighting average hourly earnings by
the estimated worker hours for the universe.
For all levels, from basic estimating cells to major
industry divisions, average weekly earnings are computed
by multiplying average hourly earnings by average weekly
hours.
Overtime hours. Average weekly overtime hours are
estimated in basically the same way as average weekly
hours. Overtime worker hour sample averages are used
in the computations rather than the sample averages for
total worker hours. The sample totals for production
workers used in the computations are those for the reports
containing overtime worker hours as well as production
workers, total payroll, and total worker hours. The wedg
ing technique and the summary level estimating technique
are also comparable to those used to estimate average
weekly hours.
Average weekly earnings in 1977 dollars. Average
weekly earnings are computed and published in terms of
1977 dollars to give an approximate measure of changes
in “ real” average weekly earnings (earnings in constant
dollars). These series are computed by dividing the
average weekly earnings (in current dollars) by the b l s
Consumer Price Index for Urban Wage Earners and
Clerical Workers ( c p i -W ) for the same months.
Average hourly earnings, excluding overtime, for
manufacturing industries. These are computed by
dividing the total production worker payroll for an
industry group by the sum of the total production worker
hours and one-half of the total overtime worker hours,
which is equivalent to the payroll divided by straight-time
hours. This method excludes overtime earnings at an
assumed rate of 1^-times the straight-time rates; no fur
ther adjustment is made for other premium payment
provisions.
Hourly earnings indexes. These indexes reflect the
average change of the component industries’ average
hourly earnings since the base period (1977). The method
used to derive these indexes adjusts for the effects of fluc
tuations and varying trends in employment activity
(aggregate worker hours) in higher wage versus lower
wage industries. This is done by assigning a fixed weight
to each of the component industries. The weights are
derived from the industries’ base period estimates of
aggregate worker hours. The hourly earnings indexes
utilize a further adjustment to the component industries
in the manufacturing sector (the only sector for which
overtime data are available) to adjust for the varying
impact that changes in overtime hours have on the
estimates of average hourly earnings. No attempt is made
to adjust the hourly earnings indexes for the impact of
the fluctuations and varying trends in occupational
employment within industries and other factors which
also influence the trends in average hourly earnings.
Hourly earnings indexes are published for “total private”
and the major industry divisions except government.
Indexes o f aggregate weekly worker hours and pay
rolls. These indexes are prepared by dividing the cur
rent month’s aggregates by the average of the monthly
aggregates for 1977.
Indexes o f diffusion o f changes in the number o f
employees on nonagricultural payrolls. These indexes
measure the percentage of industries which posted in
creases in employment over the specified time span. The
indexes are calculated from 185 seasonally adjusted
employment series (2-digit nonmanufacturing industries
and 3-digit manufacturing industries) covering all non
agricultural payroll employment in the private sector.
Seasonally adjusted series
Many economic statistics reflect a regularly recurring
seasonal movement which can be measured on the basis
of past experience. By eliminating that part of the change
which can be ascribed to the normal seasonal variation,
it is possible to observe the cyclical and other nonseasonal
movements in these series. Seasonally adjusted series are
published regularly for selected employment, hours, and
earnings series.2
The seasonally adjusted series are computed by
multiplying the corresponding unadjusted series by the
appropriate seasonal adjustment factors. Seasonally
adjusted series for broader industry groups are obtained
by summing the seasonally adjusted data for the compo
nent industries. Seasonally adjusted hours and earnings
averages for broader level industry groups are weighted
averages of the component series.
Presentation
The c e s program has continued to improve and
expand since its inception; it currently uses payroll reports
2
See appendix A of this bulletin for a description of the seasonal
adjustment methodology.
19
from over 290,000 establishments to provide monthly
estimates of employment, hours, and earnings in con
siderable industry detail. Estimates are prepared at the
4-digit sic level for manufacturing industries and the
3-digit Sic level for most nonmanufacturing industries.
However, as a result of an expansion in the SIC coding
structure, b l s began publication of 82 additional
industries in the service-producing sector, including 58
at the 4-digit sic level, in June 1984, coincident with the
introduction of the March 1983 benchmarks.3
At the national level, the CES program currently pro
duces more than 3,500 separately published series each
month. Tables 2, 3, and 4 provide a summary of the
published national detail by major industry division.
Table 2 describes the primary series produced by the pro
gram, that is, those computed directly from the sample
and benchmark data. Table 3 indicates the special series
which are obtained from the primary series by the applica
tion of special adjustments, while table 4 lists the
seasonally adjusted series.
In addition to the series published on a current month
ly basis, employment in March of each year (based on
benchmark data) is published in the June issue of
Employment and Earnings for a number of industries for
which monthly estimates do not meet established publica
tion standards. In June 1987, following revisions of the
establishment-based data to March 1986 benchmarks,
data for nearly 330 such industries were published.
In addition to the national estimates, BLS publishes
employment estimates for all 50 States, the District of Co
lumbia, Puerto Rico, the Virgin Islands, and 278 areas.4
Approximately 12,000 employment series and about
10,000 hours and earnings series (for 184 areas) are pub
lished for these States and areas. (These estimates were
adjusted to March 1986 benchmarks with the publication
of January 1987 data.) The employment series usually
cover total nonagricultural employment, major industry
divisions (e.g., mining, construction, manufacturing), and
major industry groups (e.g., textile mill products, transpor
tation equipment, retail trade) for each State and area.
Additional industry detail is frequently provided for the
larger States and areas, particularly for industries which
are locally important in the various jurisdictions.
The series on employment, hours, and earnings appear
in several b l s publications. The summary data are first
published each month in The Employment Situation news
release which contains preliminary national estimates of
nonagricultural employment, average weekly hours, and
average weekly and hourly earnings in the preceding
month, for major industries. The release also includes
seasonally adjusted data on employment, average weekly
3 See John T. Tucker, “Publication of Employment Data for Addi
tional Service-Producing Industries,” E m ploym en t a n d Earnings, June
1984, pp. 24-27.
4 Data for Puerto Rico and the Virgin Islands are not used in compil
ing national estimates.
hours, and average weekly overtime hours. The prelimi
nary estimates are based on tabulations of data for less
than the full sample to permit early release of these widely
used economic indicators. This release is normally issued
3 weeks after the week of reference for the data. The news
release also includes a brief analysis of current trends in
employment, hours, and earnings, highlighting current
trends as compared with the data for the previous month
and for the same month in the prior year.
Most of the national estimates in the detail described
in tables 2, 3, and 4 are published monthly in the
periodical Employment and Earnings, issued about 5
weeks after the week of reference. Employment data for
total nonagricultural employment and for the major
industry divisions, as well as hours and earnings for total
manufacturing, are published for the States and areas in
Employment and Earnings 1 month later than employ
ment data for the Nation. Special articles describe tech
nical developments in the program. Many of the national
series are also published in the Monthly Labor Review.
Historical national statistics (monthly data and annual
averages) derived from the c e s program are published
in Employment, Hours, and Earnings, United States,
1909-84 (Bulletin 1312-12). Following each benchmark
revision, a supplement has been published which contains
all revised data. The latest supplement, issued in July
1987, reflects all revisions resulting from the introduction
of the March 1986 benchmarks. A companion volume,
Employment, Hours, and Earnings, States and Areas,
1939-82 (Bulletin 1370-17) and its most recent supple
ment, Bulletin 1370-19, provide annual averages on all
employees and on production-worker hours and earnings.
Plans are underway to publish a new edition of the histor
ical bulletin incorporating monthly and annual data for
States and areas through 1987. Detailed industry data are
available each month in releases published by the cooper
ating State agencies.
Employment, hours, and earnings (national) data are
available in machine-readable form and on data diskettes .
The data are also disseminated in the publications of other
Federal agencies; e.g., the Department of Commerce, the
Board of Governors of the Federal Reserve System, and
the Council of Economic Advisers. They are also regu
larly republished in summary form or for specific indus
tries in many trade association journals, the labor press,
and in general reference works.
Comparison with the Current
Population Survey
Total employment in nonagricultural establishments
from the CES or payroll survey is not directly comparable
with the Bureau’s estimates of the number of persons
employed in nonagricultural industries obtained from the
monthly household survey. (See chapter 1 for a description
20
Table 2. Number of “primary” national series on employment, hours, and earnings published from the Current Employment Statistics
program by industry, June 1987
Industry
All
employees
Production
workers1
Women
workers
Average
weekly
hours
Average
weekly
overtime
hours
Average
hourly
earnings
Average
weekly
earnings
Total .................................................
597
452
522
451
323
451
451
Total nonagriculturai ................................
1
—
1
—
-
-
-
Total private ..........................................
1
1
1
1
Goods-producing ......................................
Mining .................................................
Construction .......................................
Manufacturing ....................................
1
13
15
324
1
11
15
323
1
9
15
269
Service-producing ......................................
Private service-producing ....................
Transportation and public utilities .
Wholesale trade ................................
Retail trade ........................................
Finance, insurance, and real estate
Services ...............................................
1
1
32
46
45
30
65
—
1
16
20
31
10
23
1
1
27
46
45
30
64
—
11
15
323
—
—
17
20
31
10
23
Government .............................................
22
-
12
-
1
—
1
—
—
—
—
323
11
15
323
11
15
323
—
—
—
—
—
—
—
—
—
17
20
31
10
23
—
—
17
20
31
10
23
-
-
-
1 Production workers in manufacturing and mining; construction
workers in construction; and nonsupervisory workers in all other
industries.
Table 3. Number of “special” national series on employment, hours, and earnings published from the Current Employment Statistics
program by industry, June 1987
Indexes of
aggregate
weekly
hours
In d u s try
Indexes of
aggregate
weekly
payrolls
..................................................
1
1
G o o d s -p ro d u c in g ............................................
M in in g
C o n s tru c tio n ............................................
M a n u fa c tu rin g ..........................................
1
1
1
25
1
1
1
25
1
1
1
1
1
1
1
1
1
1
1
1
T o ta l p riv a te
S e rv ic e -p ro d u c in g ............................................
T ra n s p o rta tio n an d p u b lic u tilitie s .
W h o le s a le tra d e .....................................
R e ta il tra d e ...............................................
F in a n c e , in s u ra n c e , an d real e s ta te
S e rv ic e s ......................................................
The
Hourly
Earnings
Index
Average
hourly earnings,
excluding
overtime
1
Average
hourly
earnings
(1977 = 100)
1
—
_
—
1
1
1
1
—
—
1
1
1
1
1
1
—
1
1
1
1
1
—
1
1
1
1
1
—
—
23
—
1
1
1
1
1
Average
weekly
earnings
(1977 = 100)
—
—
—
—
—
—
Table 4. Number of “seasonally adjusted” national series on employment, hours, and earnings published from the Current Employ
ment Statistics program by industry, June 1987
Industry
All
Production
employees
workers1
Women
workers
Indexes
Average
Average
of
weekly aggregate overtime
hours
weekly
hours
hours
Average hourly
earnings
Average weekly
earnings
Current
dollars
1977
dollars
Current
dollars
1977
dollars
Total nonagriculturai ............................
1
—
1
—
—
—
—
—
—
—
Total private ........ ..............................
1
1
1
1
1
1
1
1
Goods-producing ...................................
Mining .................. ......................
Construction ...................................
Manufacturing ...............................
1
2
2
25
1
1
1
25
1
1
1
23
_
—
—
21
1
1
1
25
—
—
—
—
3
—
—
1
1
—
1
—
—
—
Service-producing ................................
Transportation and public utilities
Wholesale trade ............................
Retail trade .....................................
Finance, insurance, and real
estate ...........................................
Services ...........................................
2
3
3
5
1
1
1
1
2
1
1
1
_
1
1
1
1
1
1
1
—
—
—
_
—
—
—
_
—
—
—
—
—
—
—
—
—
—
4
3
1
1
1
1
—
1
1
-
Government.........................................
4
4
-
-
1
1 Production workers in manufacturing and mining; construction
workers in construction; and nonsupervisory workers in all other
industries.
21
-
_
—
—
—
1
1
1
1
1
_
-
-
—
—
—
_
-
-
-
-
-
-
of the Current Population Survey, or household sur
vey.) The two surveys have differences in concept and
scope and employ different collection and estimating
techniques.
The payroll survey excludes unpaid family workers,
domestic workers in private homes, proprietors, and other
self-employed persons, all of whom are covered by the
household survey. Moreover, the payroll survey counts
a person who is employed by two or more establishments
at each place of employment, while the household survey
counts a person only once, and classifies him or her
according to the major activity. Certain persons on
unpaid leave are counted as employed under the house
hold survey, but are not included in the employment
count derived from the payroll survey. However, over
time, they show similar trends in employment.
The household survey places its primary emphasis on
the employment status of individuals and also provides
a great deal of information on the demographic character
istics (sex, age, race) of the labor force. The survey is not
well suited to providing detailed information on the
industrial and the geographic distribution of employment.
The establishment survey, while providing limited infor
mation on personal characteristics of workers, is an excel
lent vehicle for obtaining these detailed industrial and
geographic data; in addition, it provides hours and earn
ings information which is directly related to the employ
ment figures. The payroll and household surveys,
therefore, should be regarded as complementary.
Employment trends indicate changes in the structure
and growth of individual industries and, in conjunction
with trends in hours and other economic data, yield
measures of productivity.
Wide need has been demonstrated by both labor and
business for industry series on hourly earnings and weekly
hours, to provide a basis for labor-management negotia
tions. They not only furnish current and historical infor
mation on a given industry but provide comparative data
on related industries.
Reliability of Estimates
Although the relatively large size of the CES sample
assures a high degree of accuracy, the estimates derived
from it may differ from the figures that would be
obtained if it were possible to take a complete census
using the same schedules and procedures. While the
estimates are adjusted annually to new benchmarks,
changes between benchmark months are not reflected in
the data—new establishments, for example, or changes
in the industrial classification of establishments resulting
from changes in their product or activity. In addition,
small sampling and response errors may accumulate over
several months as a result of the link relative technique
of estimation between benchmarks.
One measure of the reliability of the employment
estimates for individual industries is the root-mean-square
error (rmse). This measure is the standard deviation
adjusted for the bias in the estimates:
Uses
RMSE = \/(Standard Deviation)2 + (Bias)2
The series are used by business firms, labor unions,
universities, trade associations, private research organiza
tions, and many government agencies to study economic
conditions and to develop plans for the future. Business
firms, for example, use the employment, hours, and earn
ings data for guidance in plant location, sales, and pur
chases. Also, firms negotiating long-term supply or con
struction contracts often use escalation clauses based on
the average hourly earnings series as an aid in reaching
equitable agreements; escalation clauses permit an adjust
ment of wages depending on the movement of average
hourly earnings in a selected industry.
Researchers use the trends reflected in these statistics
as economic indicators. The average weekly hours series,
for example, is a leading indicator of swings in the
business cycle.
If the bias is small, the chances are about 2 out of 3
that an estimate based on the sample would differ from
its benchmark by less than the root-mean-square error.
The chances are about 19 out of 20 that the difference
would be less than twice the root-mean-square error.
Hours and earnings estimates are not subject to bench
mark revisions although the broader industry groupings
may be affected slightly by changes in the productionworker weights. The hours and earnings estimates,
however, are subject to sampling errors which may be
expressed as relative errors of the estimates. (A relative
error is a standard error expressed as a percent of the
estimate.) Measures of root-mean-square errors for
employment estimates and relative errors for hours and
earnings estimates are provided in the Explanatory Notes
in Employment and Earnings.
22
Technical References
Bureau of Labor Statistics
Cronkhite, Fred R. “ bls Establishment Estimates Revised
to March 1986 Benchmarks,” Employment and Earnings,
June 1987.
Current Employment Statistics State Operating Manual,
October 1985.
Employment and Earnings, Explanatory Notes, monthly.
An up-to-date, concise description of the concepts and
methods used in establishment-based employment, hours,
and earnings data from the Current Employment
Statistics program. Provides tables which present
measures of the reliability of the data and the magnitude
of revisions due to benchmark adjustments.
Stinson, John F., Jr. “ Comparison of Nonagricultural
Employment Estimates From Two Surveys,” Employ
ment and Earnings, March 1984.
Tucker, John T. “ Publication of Employment Data for Addi
tional Service-Producing Industries,” Employment and
Earnings, June 1984.
Other
Dagum, Estela Bee. The X - l l a r i m a Seasonal Adjustment
Method. Ottawa, Statistics Canada, January 1983.
Statistics Canada Catalogue No. 12-564E.
Employment, Hours, and Earnings, United States, 1909-84,
Volumes I and II, Bulletin 1312-12, March 1985, Sup
plement, July 1987.
A compilation of historical and current data from all
national series published from the Current Employment
Statistics program.
National Commission on Employment and Unemployment
Statistics. Counting the Labor Force, 1979.
A comprehensive review and critique of the methods
and concepts used by various Federal Government
programs providing statistics on employment, unemploy
ment, and the labor force in the United States.
Green, Gloria P. “ Comparing Employment Estimates From
Household and Payroll Surveys.” Monthly Labor
Review, December 1969.
Shiskin, Julius; Young, Allan H.; and Musgrave, John C.
The X - l l Variant o f the Census M ethod II Seasonal
Adjustment Program. U.S. Department of Commerce,
Bureau of the Census, November 1967.
Manual on Series Available and Estimating Methods, bls
Current Employment Statistics Program, March 1986.
A summary of the sic codes, industry titles, sample
coverage, stratification pattern, etc., of the national
employment, hours, and earnings series produced and
published from the Current Employment Statistics
program.
23
Bureau of Labor Statistics Report on
Employment, Payroll, and Hours—Manufacturing
U.S. Department of Labor
This report is authorized by law 29 U.S.C. 2. Your voluntary cooperation is needed to make the results of this survey
Form Approved
comprehensive, accurate, and timely. The information collected on this form by the Bureau of Labor Statistics and the
O.M.B. No. 1220-0011
States cooperating in its statistical programs will be held in confidence and will be used for statistical purposes only._________________
Report Number
State
Return promptly each month in the enclosed envelope which requires no postage.
Change name and mailing address if incorrect—Include Zip code.
Return to:
r
~i
SAMPLE COPY
L
J
A.
Please provide the following Information in case questions arise concerning this report.
Your Name
Title
Phone Number
B.
Please provide the location of establishments covered by this report.
Number of establishments
City
(_____ )
C.
[I]
Please check one: Production workers are paid
□
each week
County
State
CD every 2 weeks CH twice a month
□
once a month
other, specify:_________________________ ___ ____
D. Please complete columns 1-6 for the pay period checked above which includes the 12th of the month. Detailed explanations are on the reverse side.
Reference
Period
Please report
data only for
the pay period
which includes
the 12th of the
month
(1)
All Employees:
Report the number
of paid employees
who worked during
or received pay for
any part of the pay
period which in
cludes the 12th of
the month
DEC 1986
JAN
(2)
Women
Employees:
Report the
number of em
ployees from
column 1 that
are women
(3)
(4)
Production
Production
Workers:
Worker Payroll:
Report the num Report the total pro
ber of employees duction worker pay
from column 1
roll, including over
that are produc time, for the pay
tion workers
period which includes
the 12th of the month
(omit cents)
(5)
Production
Worker Hours:
Report the total
production worker
hours, including
overtime, for the
pay period which
includes the 12th of
the month (omit
fractions)
$
1987
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
DEC
E. Please report comments on significant changes in your employment, payroll, or hours on the reverse.
BLS-790 C Rev Dec 86
24
(6)
Production Worker OFFIC E USE
ONLY
Overtime Hours:
Expl
UP
Report the total
production worker code
code
overtime hours
included in col
umn 5 (omit
fractions)
E xplanatio n* for Entering D ata on R everse Side_____________________
For what time period should I complete this form?
"Production workers” excludes:
executives
personnel
cafeterias
finance
medical
accounting
technical
professional
legal
advertising
credit
sales
collection
sales-delivery
purchasing
recordkeeping (clerical) not related to production
force account construction
installation of products
servicing of products
Complete this form only for the single pay period checked in Part C (weekly,
monthly, etc.) that includes the 12th day of the month. Payroll and hours
(Part D, columns 4-6) should be reported for the entire pay period checked
in Part C, regardless of its length.
If your pay period is Monday through Friday, and the 12th falls on a Satur
day, please report for the week of the 6th through the 12th. When the 12th
falls on a Sunday, report for the week of the 12th through the 18th.
Column 1 All Employees:
Enter the total number of persons who worked full- or part-time or received
pay for any part of the pay period including the 12th of the month.
“All Employees” includes:
Column 4 Production Worker Payroll:
Enter the total amount of pay earned during the entire pay period checked
in Part C (weekly, etc.) for all production workers in column 3.
Report pay before employee deductions for:
salaried officials of corporations
executives and their staff
persons on paid vacation
persons on paid sick leave
persons on other paid leave
part-time employees
trainees
“All Employees” excludes.
FICA (social security)
unemployment insurance
health insurance
pensions
pay deferral plans (401K plans)
Federal, State, and local income taxes
bonds
union dues
proprietors
pensioners
unpaid family workers
partners of unincorporated firms
persons on strike the entire pay period
persons on leave without pay the entire pay period
armed forces personnel on active duty the entire pay period
Include pay for:
Exclude:
overtime
holidays
vacations
sick leave
other paid leave
bonuses, unless paid regularly
lump sum payments
retroactive pay
pay advances
payments-in-kind
Column 2 Women Employees:
Enter the number of employees from column 1 that are women.
Column 3 Production Workers:
Enter the number of employees from column 1 that are production workers.
"Production workers” includes all nonsupervisory workers engaged in such
occupations as:
fabricating
storage
receiving
warehousing
shipping
processing
trucking
assembling
packing
janitorial
handling
repair
maintenance
product development
recordkeeping (clerical) related to production
“Production workers” also includes working supervisors and group leaders
who may be "in charge" of a group of employees, but whose supervisory
functions are only incidental to their regular work.
Column 5 Production Worker Hours:
Enter the total number of hours paid for during the entire pay period checked
in Part C (weekly, etc.) for all production workers in column 3. Do not con
vert overtime or other premium hours to straight-time equivalent hours.
“Hours paid for” is the sum of:
1. Hours worked, including overtime hours.
2. Hours paid for stand-by or reporting time.
3. Hours not worked, but for which pay was received directly from the
firm. Included are holidays, vacations, sick leave, or other paid leave.
Column 6 Production Worker Overtime Hours:
Enter the total number of hours from column 5 for which overtime premiums
were paid because the hours were in excess of the regularly scheduled hours.
Include Saturday, Sunday, 6th day, 7th day, and holiday hours only if over
time premiums were paid. Exclude hours for which only shift differential,
hazard, incentive, or other similar types of premiums were paid.
If there were no overtime hours, enter "O" in column 6.
E. Comments. Check the box which best
indicates the reason for significant
changes in employment (Emp),
payroll (PR), or hours (Hrs).
Circle the item(s) (Emp,
PR, or Hrs) to which
the comment applies.
JAN
Emp
PR
Hrs
FEB
Emp
PR
Hrs
MAR
Emp
PR
Hrs
APR
Emp
PR
Hrs
MAY
Emp
PR
Hrs
JUN
Emp
PR
Hrs
JUL
Emp
PR
Hrs
AUG
Emp
PR
Hrs
SEP
Emp
PR
Hrs
OCT
Emp
PR
Hrs
NOV
Emp
PR
Hrs
DEC
Emp
PR
Hrs
If
none of
the checkboxes
apply, write your
own comments here.
BLS-790 C Rev Dec 86
25
Bureau of Labor Statistics Report on
Employment, Payroll, and Hours—Trade
U.S. Department of Labor
This report is authorized by law 29 U.S.C. 2. Your voluntary cooperation is needed to make the results of this survey
Form Approved
comprehensive, accurate, and timely. The information collected on this form by the Bureau of Labor Statistics and the
O.M.I3. No. 1220-0011
States cooperating in its statistical programs will be held in confidence and will be used for statistical purposes only._______________
State
Report Number
Return promptly each month in the enclosed envelope which requires no postage.
Change name and mailing address if incorrect—Include Zip code.
Return to:
n
r
SAMPLE COPY
L
J
A. Please provide the following Information In case questions arise concerning this report.
Your Name
Title
B. Please provide the location of establishments covered by this report.
Number of establishments
City
C. Please check one.
Nonsupsrvlsory employees are paid:
D. Please check one.
Nonsupervlsory employees are paid commissions:
EH
EH
ED
EH
each week
Phone Number
(
)
County
EH
State
every 2 weeks
EH
twice a month
EH
once a month
EH
twice a month
EH
once a month
other, specify:
each week
EH
every 2 weeks
no commissions are paid
EH
other, specify:
c. m i w complete columns l-e ana o ror me pay penoa cnecnea in u aoove wnicn inciuaes me in n or me monin. uompieie column s ror me
commission period checked In D above which includes the 12th of the month. Detailed explanations are on the reverse side.
(3)
(1)
(•)
(2)
(4)
(5)
Women
Nonsupervlsory Nonsupervlsory
Nonsupervlsory
DO MOT
Rcltftnog
0 0 NOT
All Employees:
Commissions of
"•nou
Employee Hours:
Nonsupervlsory
U5IE
USE
Employees: Employees:
Employee Payroll:
Employees:
Please report
data only for
the pay period
which includes
the 12th of the
month
Report the number of Report the
paid employees who number of
worked during or re employees
ceived pay for any
from column
part of the pay period 1 that are
women
which includes the
12th of the month
DEC 1986
Report the num
ber of employees
from column 1
that are nonsupervisory
employees
Report the total nonsupervisory employee
payroll, excluding
commissions, for the
pay period including
the 12th of the
month (omit cents)
Report the total
commissions paid
for the period in
cluding the 12th
of the month
(omit cents)
$
$
OFFICE USE
ONLY
PR
Report the total nonsupervisory employee
hours, including over
time, for the pay
period including the
12th of the month
(omit fractions)
JAN 1987
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
DEC
F. Please report comments on significant changes In your employment, payroll, hours, or commissions on the reverse.
BLS-790 E Rev Dec 86
26
OFFIC E USE
ON LY
Expl
code
UP
code
Explanations for Entering Data on Reverse Side
For what time period should I complete this form?
Complete Part E, columns 1-4 and 6, only for the single pay period checked
in Part C (weekly, monthly, etc.) that includes the 12th day of the month.
Payroll and hours (Part E, columns 4 and 6) should be reported for the en
tire pay period checked in Part C, regardless of its length.
Commissions (Part E, column 5) should be reported for the entire commis
sion period checked in Part D, regardless of its length. If your commission
period ends more than 2 weeks after the end of the pay period checked
in Part C, do not delay this report. Instead, report commissions on a one
month lag, the next time you receive this form.
“Nonsupervisory employees” includes working supervisors and group
leaders who may be "in charge” of a group of employees, but whose super
visory functions are only incidental to their regular work
In other words, ‘‘nonsupervisory employees” includes every employee except
those whose major responsibility is to supervise, plan, or direct the work
of others.
Column 4 Nonsupervisory Employee Payroll:
Enter the total amount of pay earned during the entire pay period checked
in Part C (weekly, etc.) for all nonsupervisory employees in column 3. Do
not include commissions. Commissions are reported in column 5.
Report pay baton employee deductions for:
If your pay period checked in Part C is Monday through Friday, and the 12th
falls on a Saturday, please report for the week of the 6th through the 12th.
When the 12th falls on a Sunday, report for the week of the 12th through
the 18th.
FICA (social security)
unemployment insurance
health insurance
pensions
pay deferral plans (401K plans)
Federal, State, and local income taxes
bonds
union dues
Column 1 All Employees:
Enter the total number of persons who worked full- or part-time or received
pay for any part of the pay period including the 12th of the month.
“All Employees” includes:
salaried officials of corporations
executives and their staff
persons on paid vacation
persons on paid sick leave
persons on other paid leave
part-time employees
trainees
“All Employees” excludes:
proprietors
pensioners
unpaid family workers
partners of unincorporated firms
persons on strike the entire pay period
persons on leave without pay the entire pay period
armed forces personnel on active duty the entire pay period
Column 2 Women Employees:
Enter the number of employees from column 1 that are women.
Column 3 Nonsupervisory Employees:
Enter the number of employees from column 1 that are nonsupervisory
employees. Nonsupervisory employees are all employees in column 1 who
are NOT:
officers of corporations
executives
Emp
PR
Emp
PR
Hrs
MAR
Emp
PR
Hrs
APR
Emp
PR
Hrs
MAY
Emp
PR
Hrs
Emp
PR
Hrs
JUL
Emp
PR
Hrs
AUG
Emp
PR
Hrs
SEP
Emp
PR
Hrs
OCT
Emp
PR
Hrs
NOV
Emp
PR
Hrs
DEC
Emp
PR
Hrs
If
none of
the checkboxes
apply, write your
own comments here.
BLS-790 E Rev Dec 86
tips
commissions
lump sum payments
retroactive pay
pay advances
payments-in-kind (meals, etc.)
bonuses, unless paid regularly
travel expenses
1. Hours worked, including overtime hours.
2. Hours paid for stand-by or reporting time.
3. Hours not worked, but for which pay was received directly from the
firm. Included are holidays, vacations, sick leave, or other paid leave.
Hrs
JUN
overtime
holidays
vacations
sick leave
other paid leave
Column 6 Nonsupervisory Employee Hours:
Enter the total number of hours paid for during the entire pay period checked
in Part C (weekly, etc.) for all nonsupervisory employees in column 3. Do
not convert overtime or other premium hours to straight-time equivalent
hours. “Hours paid for” is the sum of:
F. Comments. Check the box which
indicates the reason for significant
changes in employment (Emp),
payroll (PR), or hours (Hrs).
Circle the item(s) (Emp,
PR, or Hrs) to which
the comment applies.
FEB
Exclude:
Column 5 Commissions of Nonsupervisory Employees:
Enter commissions (not base pay, drawing accounts, or basic guarantees)
paid to all nonsupervisory employees in column 3 during the entire com
mission period checked in Part D (weekly, etc.). If no commissions are paid,
check the appropriate box in Part D and leave column 5 blank.
department heads
managers
JAN
Include pay for:
27
Chapter 3. Occupational
Employment Statistics
The Occupational Employment Statistics (OES) survey
is a periodic mail survey conducted by State employment
security agencies of a sample of nonfarm establishments
to obtain wage and salary employment by occupation.
These data are used to estimate total employment by
occupation for the Nation, each State, and selected areas
within States.
Background
In 1971, questionnaires were sent to 50,000 manufac
turing establishments throughout the United States,
marking the beginning of the OES survey. This survey was
conducted in cooperation with the Employment and
Training Administration and 10 State employment
security agencies. It was designed to obtain occupational
estimates for the Nation and for the cooperating States.
Similar surveys were inaugurated for nonmanufacturing
industries with the participation of additional cooperating
State agencies. State and local governments were surveyed
as well.
o e s surveys follow a 3-year cycle. Between 1971 and
1987, three surveys, on average, were conducted alter
nately for manufacturing; nonmanufacturing; and trade,
transportation, communications, public utilities, and
government services industries. Surveys of hospitals were
added to the cycle in 1980 and educational services in
1985. Currently, the 50 States, the District of Columbia,
Puerto Rico, Guam, and American Samoa are
cooperating in this effort.
salaried officers, executives, and staffs of incorporated
firms; employees temporarily assigned to other units; and
employees for whom this unit is their permanent (home)
duty station, regardless of whether this unit prepares their
paycheck. Unit total employment excludes proprietors
(owners and partners) of unincorporated firms; unpaid
family workers; workers on extended leave (i.e., pen
sioners and members of the Armed Forces); and workers
on long-term layoff.
Employees are reported in the occupation in which they
are working, not in an occupation for which they may
have been trained, if that is different. For example, an
employee trained as an engineer but working as a drafter
is reported as a drafter.
Working supervisors (those spending 20 percent or
more of their time at work similar to that performed by
workers under their supervision) are reported in the
occupations which are most closely related to their work.
Part-time workers, learners, and apprentices are
reported in the occupation in which they ordinarily work.
Industrial classification
The classification system currently used for compiling
and publishing data is that described in the 1972 Stand
ard Industrial Classification (sic) Manual as revised in
1977. (See appendix B for a detailed description of this
system.) Beginning with the 1988 OES survey, data will
be collected in accordance with the 1987 sic manual.
Reporting establishments are classified on the basis of
major product or activity for the previous calendar year.
Concepts
An establishment is an economic unit which processes
goods or provides services, such as a factory, mine, or
store. It is generally at a single physical location and is
engaged predominantly in one type of economic activity.
Where a single physical location encompasses two or
more distinct activities, these are treated as separate
establishments, provided that separate payroll records are
available and certain other criteria are met.
Unit total employment includes full- or part-time
workers; workers on paid vacations or other types of
leave; workers on unpaid, short-term absences (i.e.,
illness, bad weather, temporary layoff, jury duty);
Occupational classification
The OES classification system introduced in 1983 is
based primarily on the Dictionary o f Occupational Titles,
(DOT) and is compatible with the 1980 Standard Occupa
tional Classification (SOC) system. The titles and descrip
tions of occupations are principally derived from the
d o t . The classification of occupations, with some excep
tions, follows the soc principles which group occupations
by function, industry, and skill.
The revision of the OES classification system achieved
four objectives: (1) it addressed general and specific user
needs for data, (2) it achieved compatibility with the
1980 soc at the lowest level of detail, (3) it maintained
28
historical com patibility w ith the previous o es system ,
and (4) it sim plified the classification structure.
Information compiled from industry officials and other
sources was incorporated and contributed to the classi
fication system. This system continues to allow for the
constant state of change that occupational terminology
and classification undergo. This flexibility permits
integration of the insights gained from each successive
round of OES surveys.
A “ crosswalk” which relates OES occupations to the
SOC, the 1980 Census of Population classification system,
and the DOT, has been developed so that users can inte
grate OES data with these other sources, using the DOT
as a common denominator.
Data Sources and Collection Methods
Sources of occupational data reported by respondents
are personnel records and, especially for the small report
ing units, personal knowledge of persons completing the
reports.
Employment benchmarks for this survey are derived
from employment data tabulated from the reports of the
unemployment insurance program. In some nonmanufac
turing industries, supplemental sources are used to obtain
lists of establishments that are not covered by unemploy
ment insurance laws.
Employment information is currently being collected
for more than 700 occupations. A list of occupations has
been designed for each industry or for each group of
industries having a similar occupational structure.
Two types of survey questionnaires—one long and one
short—are used. The short form was developed to reduce
the reporting burden in smaller establishments. Both
forms include specific occupational titles and definitions,
establishment identification information, and several
questions concerning the nature of the business. In addi
tion, the questionnaire provides descriptions of 3-digit
sic industries to reduce industry misclassifications.
The long form specifies an extensive list of occupations
selected for each industry grouped under broad headings
such as Clerical Occupations, Professional and Technical
Occupations, and Service Occupations. The long form
includes supplemental sheets for respondents to report
significant occupations that they could not place under
specific titles, and thus reported in the “ all other”
residual data lines. Experience with previous surveys has
shown that the supplemental sheets can be a valuable tool
in improving the occupational lists and definitions, as well
as clarifying and correcting reported data.
The short form includes abbreviated occupational lists
with accompanying definitions. No broad groups are
specified. Respondents are asked to identify and briefly
describe jobs that cannot be matched to the occupations
listed on the forms. When the questionnaires are returned,
these additional occupations are coded according to the
corresponding long-form occupational content prepara
tory to making estimates of employment by occupation.
Data are collected from respondents primarily by mail,
but visits are made to many large employers and to other
respondents who indicate particular difficulty in com
pleting the questionnaires. Normally, two mailings follow
the initial mailing and a subsample of residual nonre
spondents is contacted further by telephone.
Occupational employment data are requested for the
pay period including the 12th day of April, May, or June,
depending upon the industry surveyed.
Sampling
The OES sample is designed to yield reliable industry
occupational estimates for the participating States and
areas within those States. The sample members are
selected primarily from the lists of establishments report
ing to the State unemployment insurance program.
The sample design initially stratifies the universe of
establishments by industry. All establishments employ
ing 250 employees or more are included in the sample.
In some industries and States, the level of employment
for establishments included with certainty is less than 250
employees. For establishments not included in the sam
ple with certainty, an optimum allocation design is
obtained by stratifying the industry by size class and
sampling the size classes with probability proportionate
to the amount of employment contained in those size
classes. Within each industry size stratum, the sample
members are randomly selected.
Estimating Procedures
The occupational distribution of the respondents in
each industry by size class is determined by deriving the
ratio of the sum of the employment in each occupation
to the sum of the total employment of the corresponding
reporting establishments. These distributions are
multiplied by the corresponding benchmark estimates of
total employment in that size class. Estimates for occupa
tions in each industry group are derived by summing all
of the occupational size class estimates within that
industry group. Similarly, the estimates of combined
industry groups are derived by summing the individual
industry components.
Presentation
A report on the results of each OES survey is published
by the cooperating State employment security agencies.
bls published national estimates for the survey years
1971 and 1977-86, inclusive. Each report consisted of an
29
analytical interpretation of the findings supported by
statistical tables showing estimates of occupational
employment and measurements of the sampling error
associated with the estimates.
Uses and Limitations
The data enable analysis of the occupational composi
tion of different industries, of different plants in the same
industry, or of changes in an industry over time. Such
information is used in projecting employment require
ments by occupation and for vocational and educational
guidance. The occupational composition of various
industries is also needed to estimate the employment
implications of proposed new Government programs,
such as those in the fields of defense procurement, health,
or mass transit. Local employment service offices use
information on the occupational patterns of industries
to locate employment opportunities. Finally, occupa
tional employment and patterns data are used in analysis
by the firms and in industrial management.
All surveys are subject to response and processing
errors, although these are reduced through reviewing,
editing, and screening procedures and through contact
with respondents whose data are internally inconsist
ent or appear to involve misinterpretation of defini
tions or other instructions. In addition, estimates
derived from sample surveys are subject to sampling
error. Sampling errors for occupational employment
estimates are calculated and normally published with the
estimates.
Technical References
Thom pson, Jo h n .“ BLS Job Cross-classification System
Relates Information From Six Sources,” Monthly Labor
Review, November 1981.
Describes the relationships of several major classifica
tion systems to the Occupational Employment Statistics
classification system.
U.S. Department of Commerce, Bureau of the Census. 1970
Census o f Population Classified Index o f Industries and
Occupations, September 1971.
Bureau of the Census. 1980 Census o f Population Classified
Index o f Industries and Occupations, November 1982.
U.S. Department of Commerce, Office of Federal Statistical
Policy and Standards. Standard Occupational Classif
ication Manual, 1980.
U.S. Department of Education, Office of Educational Research
and Improvement, National Center for Education Sta
tistics. A Classification o f International Programs, 1981.
U.S. Department of Labor, Bureau of Labor Statistics. Cross
walk o f o e s Occupations to Other Occupational
Classification Systems, 1987 (available upon request).
Bureau of Labor Statistics. Occupational Employment in
Manufacturing Industries, 1983, Bulletin 2248, 1985.
Presents occupational employment data collected in
1983 for manufacturing industries.
Bureau of Labor Statistics. Occupational Employment in
Selected Nonmanufacturing Industries, Bulletin 2264,
1986.
Presents occupational employment data collected in
1984 for mining; construction; finance, insurance, and
real estate; and services industries.
Bureau of Labor Statistics. Occupational Employment in Se
lected Nonmanufacturing Industries, Bulletin 2284,
forthcoming.
Presents occupational employment data collected in
1985 for the transportation, communications, utilities,
trade, educational services, and State and local govern
ment industries.
U.S. Department of Labor, Employment and Training Adminis
tration. Dictionary o f Occupational Titles, fourth edi
tion, 1977, and its 1986 supplement.
Comprehensive descriptions of more than 13,000 jobs
coded by work requirements and duties performed.
30
Chapter 4. Measurement of
Unemployment In States and
Local Areas
Unemployment estimates for States and local areas
are developed by State employment security agencies
to measure local labor market activity. These estimates
are a key indicator of local economic conditions and are
used by State and local governments for planning and
budgetary purposes and as an indication of the need for
local employment and training services and programs.
Local area unemployment estimates are also used to
determine the eligibility of an area for benefits in various
Federal assistance programs. Under the Federal-State
cooperative program, the Department of Labor develops
the concepts, definitions, and technical procedures which
are used by State agencies for the preparation of labor
force and unemployment estimates.
Background
Unemployment estimates have been developed for
Labor Market Areas ( l m a ’s) for over 40 years. The pro
gram began during World War II under the War Man
power Commission to identify areas where labor market
imbalance was created as a result of an inadequate labor
supply, material shortages, and transportation diffi
culties. After World War II, emphasis was placed on
identifying areas of labor surplus, and the program of
classifying areas in accordance with the severity of unem
ployment was established.
In 1950, the Department of Labor’s Bureau of Em
ployment Security (now Employment and Training
Administration) published a handbook, Techniques for
Estimating Unemployment, in order that comparable
estimates of the unemployment rate could be produced
among the States. During the late 1950’s, their experiences
led to the formulation of the Handbook method, which
is a series of computational steps designed to produce
total employment and unemployment estimates and relies
heavily on data derived from the unemployment
insurance (ui) system.
In 1972, the Bureau of Labor Statistics was assigned
the responsibility for developing the concepts and methods
used by States to estimate labor force, employment, and
unemployment. In 1973, after extensive research, a new
system for developing labor force estimates was introduced
which combined the Handbook method with the con
cepts, definitions, and estimation controls from the Cur
rent Population Survey ( c p s ), the Bureau of Census
survey used to measure the labor force status of
individuals.
To improve the quality of State labor force estimates,
the CPS State samples have been increased in size several
times, beginning in 1976 when the use of the CPS as an
estimation control was extended to all States, bls
established, as a criterion for direct use of c p s data, a
maximum expected relative error of 10 percent for unem
ployment given an expected unemployment rate of 6 per
cent. Based on this criterion, in January 1978, monthly
c p s data were introduced as the official labor force
estimates at the statewide level for the 10 largest States—
California, Florida, Illinois, Massachusetts, Michigan,
New Jersey, New York, Ohio, Pennsylvania, and Texas;
and for two areas—Los Angeles-Long Beach Standard
Metropolitan Statistical Area (sm sa ) and New York City.
With the redesign of the c ps to reflect the 1980 census,
an 11th State, North Carolina, was added to those States
with direct monthly use of CPS data. Sample design
efficiencies resulted in a reduction in the maximum
expected relative error to 8 percent monthly for the 11
largest States. All other State and area estimates are
based on the Handbook method controlled to CPS state
wide estimates as explained below.
bls and the States also engaged in the u i data base
project to standardize for all States and areas the UI
claims data used in the Handbook method so that these
data would be more consistent with the concept and
definition of unemployment used in the CPS. The result
was the regular development, from computer files, of
data on u i claimants, based on their State/county/city
of residence, who certified to unemployment in the week
including the 12th of the month (the c p s reference week),
without earnings from employment in the certification
week. Currently, monthly estimates of employment and
unemployment are prepared in the State agencies for
some 5,400 geographic areas which include all States,
l m a ’s , and counties and cities with 25,000 or more
population.
31
Handbook Method
Until 1973, the Handbook method was the only means
used in developing State and local area labor force and
unemployment estimates. It is an effort to estimate
unemployment for a State or area, comparable to what
would be produced by a random sample of households
in the area, using available information without the
expense of the c p s . The Handbook presents a series of
estimating “ building blocks” where categories of
unemployed workers are classified by their previous
status. Three broad categories of unemployed persons
are: (1) those who were last employed in industries cov
ered by State ui laws; (2) those who were last employed
in noncovered industries; and (3) those who either
entered the labor force for the first time, or re-entered
after a period of separation.
In the current month, the estimate of unemployment
is an aggregate of the estimates for each of the three
building-block categories. The covered category consists
of those unemployed workers who are currently collect
ing ui benefits, have exhausted their benefits, have been
disqualified from receiving benefits, and have delayed fil
ing for benefits.
Within the covered category, only the insured
unemployed are derived directly from an actual count of
current Ui claimants for the reference week. All other
components in this and the other two covered categories
are based on special estimating equations. The estimates
of persons who have exhausted their benefits and those
in a disqualified status are based on the number actually
counted in the current period, plus an estimate of those
expected still to be unemployed from previous periods.
For the noncovered category, an estimate of unemploy
ment is developed for each industry or class-of-worker
subgroup based primarily on the “ State covered
unemployment rate” (the ratio of covered unemployment
to covered employment), and the estimate of employment
for the subgroup.
The third category, new entrants and reentrants into
the labor force, cannot be estimated directly from ui
statistics because unemployment for these persons is not
immediately preceded by the period of employment re
quired to receive ui benefits. Instead, total entrants into
the labor force are estimated on the basis of the national
historical relationship of entrants to the experienced
unemployed and the experienced labor force. The Hand
book estimate of entrants into the labor force is a func
tion of: (1) the particular month of the year; (2) the level
of the experienced unemployed; (3) the level of the
experienced labor force; and (4) the youth proportion of
the working-age population. The estimate of total
entrants for a given month is derived from the following
equation:
ENT = A(X + E) + BX
where:
ENT
E
X
A, B
=
=
=
=
total entrant unemployment
total employment
total experienced unemployment
synthetic factors incorporating seasonal variation,
and assumed relationship between the proportion
of youths in the working-age population and the
historical relationship of entrants to either the
experienced unemployed (B factor) or the experi
enced labor force (A factor).
The total employment estimate is based on data from
several sources. The primary source is a survey of
establishments designed to produce an estimate of the
total number of employees on payrolls in nonagricultural
industries. Estimates of agricultural workers, the selfemployed, unpaid family workers, and domestic workers
are developed synthetically.
Methodological Improvements—
Adjustments to the Handbook
Research has established that the Handbook pro
cedures alone produce seriously biased estimates of
unemployment and employment as measured by the c p s .
These biases are caused, in part, by methodological and
definitional problems. For example, the employment
estimates in the Handbook method are based primarily
on establishment payroll data and are place-of-work
estimates. The c p s estimates are based on a survey of
households in the area and are place-of-residence esti
mates. Also, a person on an unpaid absence is excluded
from the payroll estimate in the Handbook method but
is considered employed in the c p s . Further, a person
holding two jobs within the reference week is counted
twice in the payroll estimate but only once in the CPS
estimate.
The definitional and methodological differences be
tween the Handbook and c p s estimate of unemployment
are more difficult to reconcile. The Handbook method
does not count (or estimate) the number of persons in
covered industries who do not have sufficient time on the
job or earnings to qualify for benefits. Since ui laws vary
from State to State, the criteria for the determination of
eligibility for benefits and the treatment of persons who
fail to qualify for benefits for nonmonetary reasons
(quits, discharges, etc.), also vary from State to State.
More importantly, the CPS estimates are based on a
household sample selected in a way to provide unbiased
estimates. The Handbook is a nonsurvey method that uses
counts of ui claims at the area level and estimates other
unobservable components of unemployment using equa
tions developed primarily from historical national data.
These special equations are subject to numerous errors
related to the specification of functional form, the
32
method of estimation, and the use of national data which
do not reflect interarea differences in labor markets.
While the differences between the CPS and Handbook
estimates are often very large, as can be determined by
a comparison in selected areas where the CPS sample size
is adequate, it is not possible to measure the error in most
of the individual Handbook components since no com
parable CPS data exist.
In order to reduce the bias resulting from the use of
the Handbook method alone, and produce more consis
tent estimates across States and areas, bls has introduced
a number of adjustment procedures and changes in the
previous estimating methodology. These are described
below.
by State employment security agencies using the Hand
book estimating procedure are adjusted, or benchmarked,
to the annual average c p s State estimate. This is accom
plished in three stages. First, the monthly Handbook
estimates in each year are adjusted by the ratio of the
c p s to the Handbook annual averages for that year.
Second, the difference between the ratios of annual
averages for 2 consecutive years is wedged into the
monthly estimates in order to minimize the disturbance
to (month-to-month change in) the original series from
the first step. Finally, the second-stage monthly estimates
are adjusted to yield the c p s annual average for each
year.
Place-of-work adjustment
Direct use of CPS data
The most fundamental change was the use of CPS data
at the State level for controlling estimation error in the
Handbook method and Handbook estimates to distribute
the CPS-based State estimates to the l m a ’s that each State
comprises. In the 11 largest States (and 2 large areas
mentioned above), c p s data are used directly on a
monthly basis. In the 39 remaining States and the Dis
trict of Columbia, where the sample will not support the
direct monthly use of the c p s data, they are used as
follows.
Monthly a d ju stm en t to c p s . Each month, during the
current estimating year, Handbook employment and
unemployment estimates are adjusted to conform more
closely with the c p s estimates. The adjustment consists
of multiplying the current statewide Handbook estimates
by the ratio of a 6-month moving average of the CPS
estimate ending in the current month to the correspond
ing moving average for the Handbook estimate. This
is the so-called moving-average-ratio adjustment
procedure.
This adjustment is illustrated below using unemploy
ment as an example:
Another important modification is a procedure for
adjusting the place-of-work employment estimates used
in the Handbook method to place-of-residence estimates,
as in the CPS. Estimated adjustment factors for the major
categories of employment in the Handbook method have
been developed on the basis of employment relationships
which existed at the time of the 1980 decennial census.
These factors are applied to the preliminary employment
estimates for the current period to obtain the adjusted
estimates, which are then used in the Handbook method.
Consistency/additivity adjustment
Each month, Handbook estimates are prepared for
Labor Market Areas that exhaust the entire State area.
To obtain an estimate for a given area, a “ Handbook
share” is computed for that area which is defined as the
ratio of that area’s Handbook estimate to the sum of the
Handbook estimates for all l m a ’s in the State. This ratio
is then multiplied by the current, CPS-based statewide
estimate—either the moving-average-ratio-adjusted
Handbook estimate for 39 States or the c p s estimate for
the 11 largest States—to produce the final adjusted LMA
estimates:
5
Ua(t) = Us(t)
E UCPSs (t-K)
K= 0
Us(t) = UHBs(t) * --------------------
EUHBa(t)
where:
E UHBs(t-K)
K= 0
a = area
s = State
t = time
where:
t=
s=
Us(t)=
UHBs(t)=
UCPSs(t)=
time period
State
official State estimate
Handbook State estimate
c p s State estimate
Annual benchmark adjustments. Each year, monthly
State employment and unemployment estimates prepared
UHBa( t ) 1
The Handbook share procedure allocates the c p s ad
justed level to the l m a ’s within the State and insures
that area estimates of employment and unemployment
add to the more accurately estimated State total. In
California and New York, which also have areas taken
directly from the CPS, the Handbook share ratio for the
remaining areas is applied to the State total minus the
CPS area.
33
Producing estimates for parts of
Current labor force estim ates at the sub-LMA level are
required by several Federal program s. H ow ever, for
areas this sm all, the data required to com pute independ
ent H an d b ook estim ates are generally not available.
Based on data availability, tw o alternative m ethods are
used to disaggregate the
lm a
Uses and Limitations
l m a ’s
estim ates to the subarea
level.
The population-claims method is the preferred tech
nique. If residence-based ui claims data are available for
the subareas within the l m a , the ratio of the subarea to
the total number of claims within the LMA is used to
disaggregate the Handbook estimate of experienced
unemployed to the subarea level. The Handbook esti
mates of unemployed entrants are allocated based on the
latest available census distribution of adult and teenage
population groups. Employment is disaggregated using
current population distributions prepared by the Bureau
of the Census and weighted by each area’s 1980 census
relative share of employment to population.
If the necessary ui claims data are not available, the
census-share method is used. This method utilizes each
disaggregated area’s 1980 census shares of total LMA
employment and unemployment in order to disaggregate
Handbook employment and unemployment.
Estimates of unemployment and the unemployment
rate are used by Federal agencies to determine the
eligibility of an area for benefits in various Federal pro
grams. These include the Job Training Partnership Act
( jt p a ), the Emergency Food and Shelter Program ( e f sp ),
the Public Works and Economic Development Act
( p w e d a ), the Urban Development Action Grant Pro
gram ( u d a g p ), and Labor Surplus Area designations.
Under jt p a and e f s p , unemployment data are used with
other data to determine the distribution of funds to be
allocated to each eligible area. In the case of the p w e d a ,
the u d a g p , and Labor Surplus Area designations, the
data are used in the determination of area eligibility for
benefits.
The c p s estimates used on an annual average basis to
control labor force estimates, at the State level for 39
States and monthly for 11 States and 2 areas, are based
on a random sample of households and are subject to
sampling error. BLS does not accept sample estimates
unless the coefficient of variation (standard error divided
by the mean) of the estimate is 10 percent or less at 1
standard error. However, other types of nonsampling
errors and biases do occur that make these estimates less
reliable than what could be produced from the c p s , given
an adequate sample size.
Technical References
U.S. Department of Labor, Bureau of Employment Security.
Handbook on Estimating Unemployment, Employment
Security Research Methods, Handbook Series ( b l s
Reprint No. R-185), 1960.
U.S. Department of Labor, Bureau of Labor Statistics. Manual fo r Dev eloping Local Area Unemployment Statistics,
July 1979.
34
Chapter 5. Employment
and Wages Covered
by Unemployment Insurance
The Employment and Wages program, commonly
called the ES-202 program, is a cooperative endeavor of
b l s and the employment security agencies of the 50
States, the District of Columbia, Puerto Rico, and the
Virgin Islands. Using quarterly reports submitted by the
agencies, b l s summarizes employment and wage data for
workers covered by State unemployment insurance (Ui)
laws and for civilian workers covered by the program of
Unemployment Compensation for Federal Employees
( u c f e ).
The program is a comprehensive and accurate source of
employment and wage data, by industry, at the national,
State, and county levels. It provides a virtual census of
nonagricultural employees and their wages. In addition,
about 40 percent of all workers in agriculture are covered.
Background
The ES-202 program can trace its origins back to the
Social Security Act of 1935, which authorized collection
of information to determine if State unemployment com
pensation programs were in compliance with the act.
From the inception of the national UI system in 1938,
when the Federal Unemployment Insurance Tax Act
became effective, until 1972, collection of the data,
publication, and technical expertise were the responsibili
ties of the U.S. Department of Labor’s Manpower Ad
ministration or its predecessor agencies. Semiannual
reports summarizing the data were issued until 1950,
when the periodical Employment and Wages began quar
terly publication. In 1972, b l s assumed responsibility and
continued quarterly publication until 1975. Employment
and Wages then became an annual publication.
Concepts and Methodology
Scope of coverage
In 1938, ui coverage and, consequently, ES-202
reporting requirements, extended only to private firms
employing eight or more persons at least 20 weeks a year;
certain employee groups were exempt. Insurance coverage
was successively broadened, to include Federal civilian
employees1 (1955); firms employing four to seven
employees and ex-military personnel2 (1958); and firms
employing one to three employees, and State colleges,
universities, and hospitals (1972). In 1978, coverage was
extended to nearly all other State and local public
employees; to agricultural firms employing a minimum
of 10 workers or having a $20,000 quarterly payroll; and
to employers paying a quarterly minimum of $1,000 to
domestic workers.
ui coverage is broad and basically comparable from
State to State. In 1986, u i and u c f e covered just under
100 million workers, or 90 percent of civilian employ
ment. Covered workers received $1,848 billion in pay or
94.3 percent of the wage and salary component of per
sonal income.
Over the years, many States have legislated unemploy
ment insurance protection for additional categories of
workers above the base established through Federal
legislation. Details on coverage laws are provided in Com
parisons o f State Unemployment Insurance Laws,
available upon request from the Employment and Train
ing Administration of the Department of Labor.
When ui-covered private industry employment data
are compared directly with other employment series, the
industry exclusions also should be taken into account.
Excluded from private sector coverage in 1986 were
approximately 0.4 million wage and salary agricultural
employees, 1.4 million self-employed farmers, 7.9 million
nonagricultural workers, 0.8 million domestic workers,
and 0.4 million unpaid family workers. Also excluded
were 1.7 million members of the Armed Forces stationed
in the United States, 0.4 million workers covered by the
railroad unemployment insurance system, and about 0.6
million State and local government workers. In addition,
certain types of nonprofit employers, e.g., religious
organizations, are given a choice of coverage or non
coverage in a number of States.
1 Under the Unemployment Compensation for Federal Employees
(UCFE) program.
2 Under the Unemployment Compensation for Ex-Servicemen (UCX)
program.
35
Establishment
An establishment is an economic unit, such as a farm,
mine, factory, or store, which produces goods or pro
vides services. It usually is at a single physical location
and engaged in one, or predominantly one, type of
economic activity, for which a single industrial classifica
tion may be applied. Occasionally, a single physical loca
tion encompasses two or more distinct and significant
activities. Each activity should be reported as a separate
establishment if separate records are kept and the various
activities are classified under different 4-digit sic codes.
Reporting units
A reporting unit is the economic unit for which the
employer submits a contribution report or provides
employment and wage data for separate locations on a
supplemental form that is included with the regular con
tribution report.
Most employers covered under State ui laws operate
at only one location and primarily or entirely engage in
one activity. In such instances, the establishment and the
reporting unit are identical. Multiunit employers having
establishments in more than one county or classifiable
in more than one 4-digit industry ordinarily must submit
separate reports for each establishment. However, em
ployers having a total of fewer than 50 employees in all
secondary counties or industries may combine these units
with the primary county or industry report.
Employers having a number of similar units, par
ticularly in industries characterized by small branch
establishments (food stores, drug stores, banks) are
allowed to combine all branch establishments within a
county on a single report, regardless of employment.
In government, the reporting unit is the installation (a
single location at which a department, agency, or other
government instrumentality has civilian employees).
Federal agencies follow slightly different criteria from
private employers in breaking down their reports by
installation. They are permitted to combine as a single
statewide unit (1) all installations with 10 workers or
fewer, and (2) all installations which have a combined
total in the State of fewer than 50 workers. Also, when
there are fewer than 25 workers in all secondary installa
tions in a State, they may be combined and reported with
the major installations.
As a result of these reporting rules, the number of
reporting units is always larger than the number of
employers (or government agencies) but smaller than the
number of establishments (or installations).
the month. The pay period varies in length from employer
to employer; for most employers, it is a 7-day period,
but not necessarily a calendar week. An employer who
pays on more than one basis (such as weekly for produc
tion employees and semimonthly for office employees)
reports the sum of the number of workers on each type
of payroll for the period.
The employment count includes all corporation offi
cials, executives, supervisory personnel, clerical workers,
wage earners, pieceworkers, and part-time workers.
Workers are reported in the State and county of the
physical location of their job. Persons on paid sick leave,
paid holiday, paid vacation, and so forth are included,
but those on leave without pay for the entire payroll
period are excluded.
Persons on the payroll of more than one establishment
are counted each time reported. Workers are counted
even though their wages may be nontaxable for ui pur
poses during that period (having reached the taxable limit
for the year).
The employment count excludes employees who earned
no wages during the entire applicable period because of
work stoppages, temporary layoffs, illness, or unpaid
vacations; and employees who earned wages during the
month but not during the applicable pay period.
Total wages
Total wages, for purposes of the quarterly ui reports
submitted by employers in private industry in most States,
include gross wages and salaries, bonuses, tips and other
gratuities, and the value of meals and lodging, where sup
plied. In a majority of the States, employer contributions
to certain deferred compensation plans, such as 401 (k)
plans, were included in total wages. Total wages, how
ever, do not include employer contributions to old-age,
survivors’, disability, and health insurance (OASDHi);
unemployment insurance; workers’ compensation; and
private pension and welfare funds.3
In most States, firms report the total wages paid
during the calendar quarter, regardless of the timing of
the services performed. Under laws of a few States, how
ever, the employers report total wages earned during the
quarter (payable) rather than actual amounts paid.
For Federal workers, wages represent the gross amount
of all payrolls for all pay periods ending within the
quarter. This gross amount includes cash allowances and
the cash equivalent of any type of remuneration. It
includes all lump-sum payments for terminal leave,
withholding taxes, and retirement deductions. Federal
employee remuneration generally covers the same types
of services as those for workers in private industry.
Employment
Employment data represent the number of workers on
the payroll during the pay period including the 12th of
3
Employee contributions for the same purposes, as well as money
withheld from the employee’s gross pay for income taxes, union dues,
etc., are excluded in the UI reports.
36
activity. States assign a 4-digit industrial code to all new
units and review and update codes, where necessary, on
a 3-year cycle. Establishments or government installations
reporting more than one activity allocate the proper pro
portion of total production, revenue, sales, or payroll
costs (depending on the industry group) to each activity.
The State agency designates the proportionately largest
activity as the primary activity. Occasionally, two or more
relatively minor activities may be determined to fall within
the same industry classification and, when combined,
become the primary activity.
In some industries, separate establishments of the same
employer often carry on the same activities, in the same
proportions, and may be combined at the county level.
Sometimes, however, the proportions vary to such a
degree that the units must be classified in differing
industries and file separate reports.
Since 1938, the industrial classification of business
establishments and government installations has under
gone a number of modifications. (See table 1.) Until 1945,
classification was based on the Social Security Board
( s s b ) Classification Manual. At that time, the basis was
changed to the sic manual, which since has been revised
several times. Originally, establishments were classified
into 20 manufacturing and 60 nonmanufacturing groups,
on a 2-digit basis. The number of such groups has
remained fairly constant. Three-digit groupings were
added in 1942, and 4-digit groupings were added for
manufacturing in 1956 and for nonmanufacturing in
1968. Statewide 4-digit classifying for nonmanufacturing
Taxable wages and contributions
Taxable wages are that part of wages subject to the
State unemployment insurance tax. Contributions are
calculated on taxable wages and are reported quarterly.
Under Federal law, certain units of State and local
governments and certain nonprofit establishments may
elect to reimburse the State for any unemployment
insurance claims that have been filed against them. These
reimbursable accounts are not subject to the quarterly
assessment for unemployment insurance funds and,
therefore, their taxable wages and contributions are not
reported.
In mid-1986, approximately one-third of the States
required that employers pay ui taxes on the first $7,000
of employee wages—the minimum established by Federal
law. The remaining States established higher limits on tax
able earnings. The portion of wages subject to taxation
has varied substantially over time. In mid-1986 also,
about two-fifths of the States allowed employers to obtain
lower tax rates by making voluntary contributions to the
unemployment tax fund. The few States which tax
employees in addition to employers are requested to
include employee contributions in their ES-202 report.
industrial classification
State employment security agencies use the current
Standard Industrial Classification (SIC) Manual to
classify each reporting unit according to its primary
Table 1.
Industrial classification of employment and wage data, 1938-88
Number of industry groups by:
Period
2-digit
code
3-digit
code
..................
..................
..................
..................
..................
20
21
21
21
21
146
150
150
148
'469
1433
1968-74 ..................
1975-78 ..................
1979-87 ..................
1988 ........................
21
20
20
20
148
143
143
140
1417
451
452
459
..................
..................
..................
..................
60
56
58
62
256
236
235
2494
1975-77 ..................
1978 ........................
1979-87 ..................
1988 ........................
64
64
64
63
277
277
277
276
2553
553
553
546
4-digit
code
Basis of industrial classification
Social Security
Board (SSB)
1939
edition
Standard Industrial Classification (SIC)
1942
1945
edition edition
1957
edition
1967
edition
1972
edition
1977
supplement
1987
edition
Manufacturing
1938-41
1942-46
1947-55
1956-57
1958-67
X
X
X
X
X
X
X
X
X
Nonmanufacturing
1938-41
1942-57
1958-67
1968-74
X
X
X
X
X
X
X
1 January-March quarter only.
2 Not coded on a mandatory basis.
37
X
X
X
did not become mandatory until 1978. A few industry
exceptions allow 3-digit coding (34 4-digit sic’s are col
lapsed into 9 3-digit sic ’s). These few exceptions are
coded at the 3-digit level because it is difficult to get
systematic and accurate information sufficient to code
at the 4-digit level.
Beginning with the first quarter 1988 ES-202 report,
State agencies will be using the 1987 edition of the Sic
manual (See appendix B for more information on the new
sic manual.)
Collection methods
State agencies send to BLS each quarter the ES-202
report for approximately 5.5 million reporting units. These
reporting units are classified under 1 of 5 ownership cate
gories—private industry, 5.3 million reports; Federal
Government, 42,000; State government, 52,000; local gov
ernment, 98,000; and international government, about 25.
The State agencies summarize and codify the raw data;
check for missing information and errors; prepare
estimates of data for delinquent reports; and finally,
machine process the data onto magnetic tapes. Five
months following the end of each quarter, the agencies
are scheduled to send the tapes to Washington.
In order to assure accurate data, bls conducts several
edits of the data each quarter and then requests State
agencies to review questionable entries and correct errors.
Furthermore, an exportable macro-edit system was devel
oped by bls for State agency use so that there may be
consistent and efficient review of the ES-202 report. The
macro-edit permits State agencies to use effectively their
e d p resources in the processing, review, and correction
of data.
Comparison of the ES-202 Program with
Other Series
A number of other statistical data series are comparable
in some respects to those obtained in the ES-202 pro
gram. These series all have certain applications, strengths,
and shortcomings. Because of its broad universe cover
age, continuity, and currency, the ES-202 program is one
of the most useful.
County Business Patterns Reports
The Bureau of the Census conducts a census of most
industries every 5 years. These data, along with data from
the Internal Revenue Service and Social Security Admini
stration are combined to develop County Business
Patterns (c b p ) reports. The Census information is similar
to ES-202 data, although various differences in concepts
and methodology make comparisons difficult, particu
larly in some measurements, such as size of firm. The
Bureau of the Census separately tabulates central admini
strative offices and auxiliaries at the division level only.
Therefore, industry breakouts of private sector data at
the 2-digit, 3-digit, or 4-digit level will exclude these
groups. The ES-202 data are more frequently updated
and consequently the program maintains more continuity.
Current Employment Statistics
The Current Employment Statistics (CES) or 790 pro
gram of bls employs a sample of approximately 290,000
establishments to provide current estimates of monthly
nonagricultural employment, average hourly earnings,
average weekly earnings, and average weekly hours. The
790 program’s employment estimates are benchmarked
primarily to ES-202 records, which cover about 98 per
cent of all nonagricultural employees and 97 percent of
those in the private nonagricultural sector. For the
remaining industries, the CES program uses other sources
to estimate employment not covered by State Ul laws.
In addition to being sample-based as opposed to being
a universe count, the 790 program differs from ES-202
in that it provides hourly earnings for production (nonsupervisory) workers only whereas ES-202 provides total
payroll data for all employees, unrelated to hours.
Office of Personnel Management
The Office of Personnel Management (OPM) maintains
a statistical series on Federal employment and payroll
information by agency, type of position and appoint
ment, and employee demographic characteristics. Both
the OPM and ES-202 series exclude the Central Intelli
gence Agency and the National Security Agency, the
Armed Forces, temporary emergency workers, and crews
of certain vessels. The OPM, but not ES-202, includes
employees working in foreign countries, workers paid on
a fee or commission basis, and paid patients, inmates,
and certain employees of Federal institutions, whereas
the ES-202, but not o p m , includes Department of
Defense employees paid from nonappropriated funds,
employees with Federal appointments of the Agricultural
Extension Service, County Agricultural Stabilization and
Conservation Committees, and State and Area Marketing
Committees.
In comparison with the o p m data, ES-202 data pro
vide more industry and local employment and wage
detail, and more frequently updated detail on employ
ment by State, o p m , of course, has certain statistics that
have no parallel in ES-202.
Current Population Survey
The Current Population Survey ( c p s ) is a sample
survey of about 59,500 households chosen to represent
the entire civilian noninstitutional population. Therefore,
38
the sample includes categories of workers which are
entirely or partly excluded from the ES-202 program—
certain farm and domestic workers, the self-employed,
persons working 15 hours or more in the survey week as
unpaid workers in an enterprise operated by a member
of the family, employees of certain nonprofit organiza
tions, and railroads. The CPS also counts employees
uncompensated because of temporary absence, but
excludes workers under 16 years old. Because the c p s is
a sample and surveys households rather than establish
ments, it cannot present employment and wage data in
the industrial and geographical detail available under the
ES-202 program, but it does provide demographic
characteristics.
When providing geographic information, the CPS pro
gram tabulates data by the location of the residence. On
the other hand, the ES-202 program provides its State
and county data by the location of the job.
Presentation
Employment and Wages, an annual b l s publication,
presents State and national totals for covered employment
and wages by broad industry division, major industry
group, and detailed 4-digit industry. Data for Federal
workers also are shown by agency, industry, and State.
The publication includes distributions of employment
and wages by size of reporting unit for each major
industry division for the United States as a whole. These
data are distributed into 10 employment-size categories.
To preserve the anonymity of establishments, BLS
withholds publication of data for any geographic industry
level in which there are fewer than three reporting units,
or in which the employment of a single installation or
establishment accounts for over 80 percent of the
industry. At the request of a State, data are also withheld
where there is reason to believe that the “ fewer than
three” rule would not prevent disclosure of information
pertaining to an individual reporting unit or would other
wise violate the State’s disclosure provisions. Informa
tion concerning Federal employees, however, is fully
disclosable.
In addition to published information, county-level
data and historical data are available from the BLS
Washington office either on hard copy or on a computer
medium such as a d p tapes and floppy disks on a costreimbursable basis. The charge for this service varies
according to the complexity and volume of the request.
Write to the Division of Occupational and Administrative
Statistics, Office of Employment and Unemployment
Statistics, U.S. Department of Labor, Bureau of Labor
Statistics, 441 G Street NW., Washington, DC 20212.
The individual States, which have a wide variety of
users for the data, usually publish their own reports using
ES-202 data.
Uses
As the most complete universe of monthly employment
and quarterly wage information by industry, county, and
State, the ES-202 series has broad economic significance
in evaluating labor trends and major industry develop
ments in time series analyses and industry comparisons,
and in special studies such as analyses of wages by size
of unit.
The program provides data necessary to both the
Employment and Training Administration and the
various State employment security agencies in administer
ing the employment security program. The data
accurately reflect the extent of coverage of the State
unemployment laws and are used to measure ui revenues
and disbursements; national, State, and local area
employment; and total and taxable wage trends. The
information allows actuarial studies, determination of
experience ratings, maximum benefit levels, areas needing
Federal assistance, and also helps ensure the solvency of
unemployment insurance funds.
The ES-202 data are used by a variety of other BLS
programs. They serve, for example, as the basic source
of benchmark information for employment by industry
and by size of firm in the Current Employment Statistics
Program (b l s 790). The Unemployment Insurance Name
and Address File, compiled from ES-202 reports, also
serves as a national sampling frame for establishment
surveys by the Industry/Area Wage, Producer Price
Index, and Occupational Safety and Health Statistics
programs.
Additionally, the Bureau of Economic Analysis of the
Department of Commerce uses ES-202 wage data as a
base for estimating a large part of the wage and salary
component of national personal income and gross
national product. These estimates are instrumental in
Federal allocation of revenue-sharing funds to State and
local governments. The Social Security Administration
also uses ES-202 data in updating economic assumptions
and forecasting trends in the taxable wage base.
Finally, business and public and private research
organizations find the ES-202 program one of the best
sources of detailed employment and wage statistics.
39
Technical References
U.S. Department of Labor, Bureau of Labor Statistics. Em
ployment and Wages, Annual Averages, 1985, Bulletin
2272, November 1986.
U.S. Department of Labor, Employment and Training
Administration. “ Comparison of State Unemployment
Insurance Laws,” revised on a regular basis.
Bureau of Labor Statistics. “ ES-202, Operating Manual,”
Employment Security Manual, Part III, Sections 04000599, revised on a regular basis.
40
Chapter 6. Occupational Pay
and Supplementary Benefits
Also in 1960, the Bureau began conducting an annual
nationwide survey of professional, administrative,
technical, and clerical jobs in a broad spectrum of private
industries. The survey was begun in preparation for the
Federal Salary Reform Act of 1962 and is currently being
used in administering the Federal Pay Comparability Act
of 1970, which governs adjustments in salaries of most
Federal white-collar employees.
Background
For many decades, the Bureau of Labor Statistics has
conducted studies of wages by occupation and industry.
The best known of its early studies stemmed from a
Senate resolution of March 3, 1891, which instructed its
Committee on Finance to investigate the effects of tariff
legislation on wages and prices. At the request of the com
mittee, the Bureau developed detailed data for 1889-91
and a more limited wage rate history extending back con
tinuously to 1860, and in some cases to 1840. Systematic
collection of wage data by occupation and industry has
continued since the turn of the century; changes in
coverage have been dictated mainly by government
requirements. Thus, a large survey program undertaken
for the War Industries Board in 1919 produced occupa
tional pay rates by industry and State, and (for some
industries) by city. Between 1934 and 1940, the selection
of industries studied was determined largely by
administrative needs under the National Industrial
Recovery Act, the Public Contracts Act, and the Fair
Labor Standards Act, with emphasis on nationwide data
for relatively low-wage industries.
Survey activity shifted in the early 1940’s defense period
to heavy industries essential to war production. Imple
mentation of wage stabilization policy during the war
required a large-scale program of occupational wage
studies by industry and locality. The emphasis on data
by locality has continued since 1945 within the framework
of industry studies generally designed to yield national
and regional estimates. In addition, since World War II
the Bureau has developed two new types of occupational
wage surveys.
Area wage surveys, initiated in the late 1940’s, were
designed to meet the growing demand for pay data related
to office clerical and manual jobs that are common to
a wide variety of manufacturing and nonmanufacturing
industries within a metropolitan area. This survey pro
gram was firmly established and temporarily expanded
for use in the wage stabilization effort during the Korean
emergency. In 1960, the program was converted from a
study of metropolitan areas of special interest to a
statistically selected group of areas from which data could
be projected to represent all metropolitan areas of the
United States, excluding Alaska and Hawaii.
Description of Surveys
Although differing in industrial, geographic, and
occupational coverage, these surveys form an integrated
program of occupational wage studies based upon a com
mon set of administrative forms, a single manual of pro
cedures, and common concepts and definitions. Survey
data, collected largely by personal visits, are provided by
employers on a voluntary basis. In return, the Bureau
pledges confidentiality for the information and publishes
it in a manner that will avoid possible disclosure of an
establishment’s pay rates. In all surveys, establishments
are classified by industry as defined in the Standard
Industrial Classification Manual (sic) prepared by the
U.S. Office of Management and Budget.1 Survey
reports identify the minimum size of the establishments
(measured by total employment) studied. Metropolitan
Statistical Area definitions are employed in all
programs.2
Industry Wage Surveys, conducted in selected manu
facturing and nonmanufacturing industries, provide data
for occupations selected to represent a range of activities
performed by workers in the industry during a specified
payroll month. In selecting the occupations, primarily
nonsupervisory, consideration is given to their prevalence
in the industry, definiteness and clarity of duties, use as
reference points in collective bargaining, and importance
in representing the industry’s wage structure.
1 See appendix B.
2 The Office of Management and Budget replaced the designation
“ Standard Metropolitan Statistical Areas” with “ Metropolitan
Statistical Areas” (MSA’s) and “ Primary Metropolitan Statistical
Areas” (PMSA’s) in June 1983. These new designations are being
gradually introduced into the occupational wage surveys as the pro
gram schedules permit.
41
In addition to reporting straight-time first-shift wage
rates of individuals in the selected occupations (or hours
and earnings for incentive workers), surveys in most
industries also provide pay distributions for broad
employee groups, such as all production and related
workers or all nonsupervisory workers.
Weekly work schedules; shift operations and shift pay
differentials; paid holiday and vacation practices; and
incidence of health, insurance, and retirement plans are
included in the information collected, along with other
items of interest in a particular industry, for example,
incidence of cost-of-living adjustment ( c o l a ) provisions
or company-provided work clothing. The studies also
provide estimates of workers covered by labormanagement agreements, proportions employed under
incentive pay plans, and the extent to which establish
ments have a single pay rate or a range of rates for
individual job categories.
Twenty-five manufacturing and fifteen nonmanufac
turing industry surveys, accounting for about 22 million
employees, are conducted at the 3- or 4-digit sic level of
industry detail. A majority are on a 5-year cycle, but a
number of comparatively low-wage industries are on a
3-year cycle. The program covers a broad cross-section
of the Nation’s economy, including automobile and steel
manufacturing as well as banking, computer data serv
ices, and hospitals.
Nearly all of the manufacturing, utility, and mining
industries are studied on a nationwide basis, and estimates
are provided also for broad regions and major local areas
of employment concentration wherever possible. Surveys
in trade, finance, and service industries usually are limited
to about two dozen metropolitan areas. Nationwide
surveys generally develop separate employment and wage
estimates by size of establishment, type of area
(metropolitan or nonmetropolitan), labor-management
agreement coverage status, and type of product or plant
group.
Area Wage Surveys, conducted in a sample of metro
politan areas, provide wage data annually or every second
year for selected office clerical, professional, technical,
maintenance, toolroom, powerplant, material movement,
and custodial occupations common to a wide variety of
industries in the areas surveyed. The occupations studied
provide representation of the range of duties and respon
sibilities associated with white-collar jobs, skilled
maintenance trades, and custodial and material move
ment jobs. Weekly salaries reported for individuals in
white-collar jobs relate to regular straight-time salaries
paid for standard workweeks. Earnings information for
plant workers excludes late-shift differentials and
premium pay for overtime.
Industry divisions included in these surveys are: (1)
manufacturing; (2) transportation, communication, and
other public utilities; (3) wholesale trade; (4) retail trade;
(5) finance, insurance, and real estate; and (6) selected
service industries. Establishments employing fewer than
50 workers are excluded. However, in the 19 largest areas,
establishments in manufacturing; transportation, com
munication, and other public utilities; and retail trade
must employ a minimum of 100 workers to be included
in the survey.
In addition to the all-industry pay averages and
distributions of workers by earnings classes, separate data
are provided for manufacturing and nonmanufacturing
in each area, and for transportation, communication, and
other public utilities in all but nine areas. In 31 of the
larger areas, wage data are presented separately for
establishments that have 500 workers or more. In 1987,
the program increased its sample of areas from 70 to 90,
with 61 being studied each year. The 32 largest areas, in
terms of nonagricultural employment, are surveyed
annually, and two groups of 29 areas each are surveyed
in alternate years. All of the areas are Metropolitan
Statistical Areas or Primary Metropolitan Statistical
Areas as defined by the Office of Management and
Budget through October 1984.
Data on weekly work schedules; paid holiday and vaca
tion practices; and health, insurance, and retirement plans
are recorded separately for nonsupervisory office workers
and production and related workers. Information relating
to shift operations and shift pay differentials is published
for production workers in manufacturing, while data on
minimum entrance rates are collected for inexperienced
office workers in all industries. While the wage data are
collected annually or every 2 years, establishment prac
tices and benefit items are studied every 3 or 4 years.
Area type wage surveys have also been conducted since
1967 at the request of the Employment Standards
Administration of the U.S. Department of Labor for use
in administering the Service Contract Act of 1965. Survey
scope and method are the same as for the Bureau’s regular
area surveys, but a more limited number of occupations
and benefits are studied. In some cases, surveys relate to
geographic areas other than Metropolitan Statistical
Areas. Wage data are published annually or every second
year and benefits data every 3 or 4 years for all industries
combined.
Both programs of area wage surveys are conducted
throughout the calendar year, with each survey relating
to a specific month.
The National Survey o f Professional, Administrative,
Technical, and Clerical Pay ( p a t c ) provides broadly
based information on white-collar salary levels and
distributions in private employment, as of March each
year. Approximately 110 occupational work levels were
studied in 1985 selected from the following fields:
Accounting, legal services, personnel management,
engineering and chemistry, purchasing, photography,
drafting, computer science, and clerical. Definitions for
42
these occupations provide for classification of employees
into appropriate work levels. Although reflecting duties
and responsibilities in private industry, the definitions are
designed to be translatable to specific pay grades of
Federal white-collar employees. As a result, this survey
provides information suitable for use in comparing pay
of salaried employees in the Federal civil service with pay
of their counterparts in private industry.
Monthly and annual average salaries are reported by
occupational work level. Data relate to the straight-time
salary corresponding to the employee’s normal work
schedule, excluding overtime hours. Salary averages are
presented for all establishments covered by the survey,
establishments employing 2,500 workers or more, and for
metropolitan areas as a group.
Industry divisions included in the p a t c survey
are: (1) mining; (2) construction; (3) manufacturing; (4)
transportation, communications, electric, gas, and
sanitary services; (5) wholesale trade; (6) retail trade; (7)
finance, insurance, and real estate; and (8) services.
Limited to the Nation’s metropolitan areas during
1960-64, the annual survey was expanded in 1965 to
include nonmetropolitan counties. In 1985, the minimum
establishment size included in the survey was 50, 100, or
250 employees, depending on the industry. This minimum
has been adjusted at various times since 1961 in response
to the specifications of the President’s Pay Agent (the
Secretary of Labor and the Directors of the Office of Per
sonnel Management and the Office of Management and
Budget).3 Because the survey scope is subject to change,
users are directed to the published bulletins for a descrip
tion of current practice.
Unless stated otherwise, rates do not include tips or
allowances for the value of meals, room, uniforms, etc.
The earnings figures, thus, represent cash wages (prior
to deductions for Social Security and income taxes, sav
ings bonds, premium payments for group insurance,
meals, room, or uniforms) after the exclusion of premium
pay for overtime, weekend, holiday, or late-shift work.
Hours shown for salaried occupations relate to stand
ard weekly hours for which the employee receives regular
straight-time salary.
Survey occupations are defined in advance in a uniform
set of job descriptions. Because of the emphasis on com
parability of occupational content across establishments,
the Bureau’s job descriptions may differ significantly
from those in use in individual establishments or those
used for other purposes. The primary objective of the
description is to identify the essential elements of skill,
difficulty, and responsibility that establish the basic con
cept of the job. In general, the Bureau’s survey job
descriptions are more specific than those published in the
Standard Occupational Classification Manual, prepared
by the U.S. Office of Management and Budget.4
Although work arrangements in any one establishment
may not correspond precisely to those described, workers
meeting the basic requirements established for the job are
included.5
In applying the survey job descriptions, the Bureau’s
field representatives exclude working supervisors and those
paid less than the established job rate, such as apprentices,
learners, beginners, trainees, handicapped workers whose
rates are reduced because of their handicap, part-time or
temporary workers, and probationary workers unless
provision for their inclusion is specifically stated.
Concepts
4 An example of a job description: Maintenance Machinist: Pro
duces replacement parts and new parts in making repairs of metal parts
of mechanical equipment. Work involves most of the following: Inter
preting written instructions and specifications; planning and laying out
of work; using a variety of machinists’ handtools and precision measur
ing instruments; setting up and operating standard machine tools; shap
ing of metal parts to close tolerances; making standard shop computa
tions relating to dimensions of work, tooling, feeds, and speeds of
machining; knowledge of the working properties of the common metals;
selecting standard materials, parts, and equipment required for this
work; and fitting and assembling parts into mechanical equipment. In
general, the machinist’s work normally requires a rounded training in
machine-shop practice usually acquired through a formal apprenticeship
or equivalent training and experience.
5 In general, workers are included in a classification if the described
duties are performed a major part of the time and the remainder is spent
on related duties requiring similar or lesser skill and responsibility.
However, in some jobs, particularly office and skilled production-worker
categories, workers may regularly perform a combination of duties
involving more than one occupation. Unless indicated otherwise in the
description, in these situations consideration for classification purposes
is given to those elements of the job which are most important in deter
mining its level for pay purposes. Thus, a worker meets the basic con
cept of the stenographer classification if taking dictation is a regular
requirement of the job even though a majority of the time is spent on
routine typing.
The Bureau’s occupational wage surveys summarize a
highly specific wage measure—the rate of pay for individ
ual workers, excluding premium pay for overtime and for
work on weekends, holidays, and late shifts. Also
excluded are performance bonuses and lump-sum
payments of the type negotiated in the auto and aerospace
industries, as well as profit-sharing payments, attendance
bonuses, Christmas or year-end bonuses, and other non
production bonuses. Pay increases—but not bonuses—
under cost-of-living allowance clauses and incentive
payments, however, are included. For workers paid under
piecework or other types of production incentive pay
plans, an hourly earnings figure serves as a proxy for
the wage rate; it is computed by dividing straight-time
earnings over a time period by corresponding hours
worked.
3 The agent has responsibility for making salary comparisons be
tween Federal white-collar workers and their private-sector counterparts
and recommending pay increases for Federal white-collar workers based
on these comparisons.
43
Tabulations of the incidence of paid holidays, paid
vacations, and health, insurance, and retirement plans
are based on the assumption that plans are applicable
to all covered nonsupervisory production or office
workers if a majority of such workers are eligible or can
expect eventually to qualify for the practices listed. Data
for health, insurance, and retirement plans are limited
to plans for which at least a part of the cost is borne by
the employer. Informal provisions are excluded. (For a
description of the Bureau’s comprehensive study of
employee benefits in medium and large firms, see chapter
9 of this bulletin.)
Survey Methods
Planning. The needs of major users are a prime
consideration in designing the Bureau’s multipurpose
occupational wage surveys. Consultations are held with
appropriate management, labor, and government
representatives to obtain views and recommendations
related to scope, timing, selection and definitions of
survey items, and types of tabulations. Particularly in
planning surveys in specific industries, these discussions
supplement feedback received from the Bureau’s regional
offices on their experiences in collecting data for the
previous study. Reflecting its use in evaluation of Federal
white-collar pay, the design of the National Survey of
Professional, Administrative, Technical, and Clerical Pay
was developed in conjunction with the Office of Manage
ment and Budget and the Office of Personnel Manage
ment. Changes in the survey scope, item coverage, and
job definitions are initiated by these agencies.
The industrial scope of each survey is identified in
terms of the classification system provided in the Stand
ard Industrial Classification Manual. The scope may
range from part of a 4-digit code for an industry study
to a uniform combination of broad industry divisions and
specific industries for the area wage surveys or the
national survey of professional, administrative, technical,
and clerical jobs.
The minimum establishment size included in a survey
is set at a point where the possible contribution of the
excluded establishments to the pay averages is regarded
as negligible for most of the occupations surveyed.
Another practical reason for the adoption of size limita
tions is the difficulty encountered in classifying workers
in small establishments where they do not perform the
specialized duties indicated in the survey job definitions.
Considerations in timing of industry wage surveys
include expiration dates of major labor-management
agreements, deferred wage adjustments, seasonality of
production, and special needs of users. Wherever possi
ble, area wage surveys are timed to follow major wage
settlements as well as to meet the needs of government
agencies administering various wage statutes.
The types of occupations studied and the criteria used
in their selection are identified in the descriptions of the
various types of surveys. The job list for each survey is
selected to represent a reasonably complete range of rates
in the wage structure for the employment categories
involved; e.g., production and related workers in a
specific manufacturing industry or nonsupervisory office,
maintenance, material handling, and custodial workers
in a metropolitan area. The established hierarchy of job
rates to be found within establishments and industries per
mits the use of pay data for such key or benchmark jobs
for interpolating rates for other jobs. Technological
developments or user interests may dictate changes over
time in the job lists and definitions. New definitions for
jobs usually are pretested in a variety of establishments
prior to their use in a full-scale survey.
Questionnaires. Two basic reporting forms are used in
all surveys. The first ( b l s 2751 A) includes items relating
to products or services, employment, shift operations and
differentials, work schedules, wage payment plans, mini
mum entrance rates, paid holidays and vacations, insur
ance and retirement plans, union contract coverage, and
other items applicable to the establishments. The second
( b l s 2753G) is used in recording such information as
occupation, sex, method of wage payment, hours, and
pay rate or earnings for each worker studied. Supplemen
tary forms are used to meet particular needs.
Data collection. Bureau field representatives typically
visit the sample establishments in a survey and collect data
for a specified payroll period. They carefully compare
job functions and factors in the establishment with those
included in the Bureau job definitions. This job-matching
process may involve review of records (such as pay struc
ture plans, organizational charts, and company position
descriptions), interviews with appropriate officials, and,
on occasion, observation of jobs within establishments.
A satisfactory completion of job matching permits accep
tance of company-prepared reports where this procedure
is preferred by the respondent. Generally, however, the
field representative secures wage or salary rates (or hours
and earnings data, when needed) from payroll or other
records, and data on the selected employer practices and
employee benefits from company officials, company
booklets, or labor-management agreements.
Area wage surveys in each locality are conducted by
personal visits every third or fourth year, with partial col
lection by mail or telephone in the intervening years.
Establishments participating in the mail collection receive
a transcript of the job-matching and wage data obtained
previously, together with the job definitions. The returns
are scrutinized, and questionable entries are checked with
the respondent. Visits are made to establishments not
suitable for other types of collection, those not respond
ing to the mail or telephone request, and those reporting
unusual changes from previous survey data.
44
The work of field representatives is checked for com
pleteness and quality of reporting and accuracy in job
matching. Revisits are made by supervisory and senior
representatives on a selective basis. Systematic technical
audits of the validity of survey definitions, made by staff
with specialized training, also are maintained for the tech
nically complex nationwide white-collar salary survey.
Sampling
All surveys are conducted on a sample basis using a
suitable sampling “ frame,” that is, a list of establish
ments which fall within the designated scope of the sur
vey. The frame is as close to the universe as possible
but is often incomplete, b l s uses frames primarily com
piled from lists provided by administrative or regulatory
government agencies (primarily State unemployment
insurance agencies). These may be supplemented by data
from directories, trade associations, labor unions, and
other sources as needed. For survey purposes, an “ estab
lishment” generally refers to a single physical location
in manufacturing industries and to all outlets of a com
pany within an area or county in nonmanufacturing
industries.
The survey design employs a high degree of stratifica
tion. Each geographic-industry unit for which a separate
analysis is to be presented is sampled independently.
Within these broad groupings, a finer stratification by
product (or other pertinent attribute) and size of establish
ment is made. Stratification may be carried still further
in certain industries: Coal mines, for instance, are
classified into underground and surface mines. Such
stratification is important if the occupational structure
differs widely among the various industry segments.
The sample for each industry-area group is a prob
ability sample, that is, each establishment has a predeter
mined chance of selection. However, in order to secure
maximum accuracy at a fixed level of cost (or a fixed level
of accuracy at minimum cost) the sampling fraction used
in the various strata, or sampling cells, ranges downward
from all large establishments through progressively
declining proportions of the establishments in each
smaller size group. This procedure follows the principles
of optimum allocation using the average employment in
the stratum as the design variable. Thus, each sampled
stratum will be represented in the sample by a number
of establishments roughly proportionate to its share of
total employment. The method of estimation employed
yields unbiased estimates by the assignment of proper
weights to the sampled establishments.
The size of the sample in a particular survey depends
on the size of the universe, the diversity of occupations
and their distribution, the relative dispersion of earnings
among establishments, the distribution of the establish
ments by size, and the degree of accuracy required.
Area wage surveys are limited to selected metropolitan
areas, which form a sample of all such areas and, when
properly combined (weighted), yield employment and
wage estimates at the national and regional levels. The
sample of areas is based on the selection of one area from
a stratum of similar areas. The criteria for stratification
are region, type of industrial activity as measured by per
cent of employment in manufacturing, and major indus
tries. Each area within a stratum is selected with its
probability of selection proportionate to its nonagricultural employment. The larger metropolitan areas are selfrepresenting; i.e., each one forms a stratum by itself and
is certain of inclusion in the area sample.
Estimating Procedures
Estimated average earnings (hourly, weekly, monthly,
or annual) for an industry or an occupation are computed
as the arithmetic mean of individual employee earnings.
All estimates are derived from the sample data. The
averages for occupations, as well as for industries, are
weighted averages of individual earnings and are not com
puted on an establishment basis. Employee benefit pro
visions which apply to a majority of the production or
office workers in an establishment are considered to apply
to all production or office workers in that establishment
and are considered nonexistent when they apply to less
than a majority.
To obtain unbiased estimates, each establishment is
assigned a weight that is the inverse of the sampling ratio
for the stratum from which it was selected; e.g., if a third
of the establishments in one stratum are selected, each
of the sampled establishments is given a weight of 3.
To illustrate the use of weights, suppose the universe
was seven establishments, from which a sample of three
was selected. Assume that establishment A was 1 of 2
establishments in its stratum. It is chosen for the sample
and is given a weight of 2. Establishment B, on the other
hand, was taken with certainty (or a probability of 1) and
is thus given a weight of 1. Establishment C was taken
from the remaining group where 1 of the 4 establishments
was used in the sample, and hence is given a weight of
4. The following calculations are made in estimating
average earnings for a given occupation:
W orkers in occupation
in sam ple
establish m ents
E stablishm ent W eight
A ............
B ..........
C ............
2
1
4
A c tu a l
em ploym en t
in occu pation
40
50
10
E stim ates o f to ta l
A verage
in stratu m
hourly
earnings W orkers
Earnings
$10.40
11.20
10.60
Estimated universe ..............................................
45
2x40
1x50
4x10
2x40x$10.40
lx50x 11.20
4xl0x 10.60
170
$1,816.00
A similar method applies to any characteristic estimated
from the sample. Tb estimate the proportion of employees
in establishments granting paid vacations of 2 weeks after
2 years of service, for instance, the establishments are
classified according to the length of vacation granted after
2 years’ service, establishment weights are applied to
employment, as in the previous example, and the propor
tion of the estimated employment in the 2-week category
is computed. Using the three establishments in the pre
vious example, this can be illustrated as follows:
E stablish m en t
W eight
A c tu a l to ta l
establish m ent
em ploym en t
A ......................
B ......................
C ......................
2
1
4
100
500
75
Estimated universe
W eighted
em ploym en t
Vacation
provisio n s
a fter 2 years
200
500
300
1 week
2 weeks
1 week
.. 1,000
Thus, the estimated percentage of workers in establish
ments granting 2 weeks’ vacation after 2 years of service
. 500
„
is ——— or 50 percent.
1,000
In the area wage survey program, where a sample of
selected metropolitan areas is used to represent the totality
of such areas, a second stage of weighting is used to
expand the individual area estimates to regional and
national levels. Since each area represents a stratum of
similar areas, the total from each area is weighted to the
estimated stratum totals by multiplying by the inverse of
the chance of selection. Summing all such estimated
stratum totals yields the earnings and employment totals
for the regional and the national estimates.
b l s occupational wage surveys have response rates
generally exceeding 80 percent of establishments con
tacted. However, when a sample establishment does not
provide data, the weights of responding sample establish
ments from the same stratum are increased to adjust for
the missing data. Establishments that are out of business
or outside the scope of a survey, and their sampling
weights, are dropped from survey estimates.
Analysis and Presentation
Survey results are published in b l s bulletins, reports,
news releases, and the Bureau’s Monthly Labor Review.
Industry wage and area wage survey reports and bulletins
are issued throughout the year as the surveys are com
pleted. The bulletin on the national survey of pro
fessional, administrative, technical, and clerical pay,
preceded by a news release in July or August, becomes
available each fall. Copies of b l s reports and releases
are available upon request. Bulletins are sold by the
Superintendent of Documents, U.S. Government Printing
Office, Washington, DC 20402; g p o bookstores; and
the b l s Chicago Regional Office, Publications Sales
Center, P.O. Box 2145, Chicago, il 60690. A brief
discussion of some features related to these publications
follows.
Where an industry survey is designed to yield estimates
for selected States or areas, summary reports are published
separately as this local information becomes available.
Industry surveys limited to selected areas do not provide
pay tabulations by type of area, size of establishment,
product, or labor-management agreement coverage that
generally are included in reports on nationwide surveys.
Regardless of geographic scope, industry reports record
the incidence of incentive pay plans and, to the extent
possible, show pay data separately for time and incentive
workers.
Area wage survey reports and bulletins provide percent
age pay increases, adjusted for changes in employment, for
industrial nurses and four broad occupational groups:
Office clerical, electronic data processing, skilled mainte
nance, and unskilled plant workers. These increases are
computed for all industries, manufacturing, and nonmanu
facturing, for each metropolitan area studied. The com
putations include data only from establishments included
in both years of the survey being compared.
Pay relatives for the same broad occupational cate
gories, expressing area average pay as a percentage of the
national average, are published each summer in two
reports: Wage Differences Among Metropolitan Areas and
Wage Differences Among Selected Areas. (The first of
these reports covers the areas in the area wage survey pro
gram; the latter covers areas surveyed for the Employment
Standards Administration.) These reports permit ready
comparisons of average pay levels among areas.
Estimates of labor-management agreement coverage of
plant and office workers are included every third or fourth
year in each area wage survey. Occupational pay relation
ships within individual establishments are summarized in
the individual area bulletins.
The annual bulletins, National Survey o f Professional,
Administrative, Technical, and Clerical Pay, present
occupational salary averages and distributions on an
all-industry basis, nationwide and separately for all metro
politan areas combined, and for establishments employ
ing 2,500 workers or more. Average pay levels by industry
division are shown as percentages of the all-industry aver
ages. Salary trend estimates for the occupations studied
are included as a byproduct of the survey. Prior to 1987,
survey coverage extended fully to all private-sector
industry divisions except services, in which coverage was
limited.6 The 1987 survey, in contrast, was restricted to
services but covered the entire industry division.
6
The 1986 coverage of services was limited to engineering, architec
tural, and surveying services; commercially operated research, develop
ment, and testing laboratories; credit reporting and collection agencies;
computer and data processing services; management, consulting, and
public relations services; noncommercial educational, scientific, and re
search organizations; and accounting, auditing, and bookkeeping services.
46
The Monthly Labor Review regularly publishes articles
on the occupational wage surveys in two forms. Research
summaries alert interested parties to a survey that has
been completed by providing highlights of the findings.
Special topical articles provide in-depth analyses of wages
and related benefits. (See references at the end of this
chapter for specific MLR articles.)
In addition to the survey publications, b l s regularly
makes computer tapes available for sale on the area wage
and p a t c surveys. Requests for computer tapes on
industry wage surveys are considered on an individual
survey basis. Filling such requests primarily depends upon
the Bureau’s ability to protect the identity of respondents
and their data.
Uses and Limitations
Occupational wage data developed in b l s surveys have
a variety of uses. They are used by Federal, State, and
local agencies in wage and salary administration and in
the formulation of public policy on wages, as in minimum
wage legislation. They are of value to Federal and State
mediation and conciliation services and to State unem
ployment compensation agencies in judging the suitability
of job offers. Knowledge of levels, structures, and trends
of pay rates by occupation, industry, locality, and region
is required in the analysis of current economic develop
ments and in studies relating to wage dispersion and
differentials.
Bureau data are used in private wage or salary deter
minations by employers or through the collective bargain
ing process. To the extent that wages are a factor, survey
data also are considered by employers in selecting loca
tions for new facilities and in cost estimating related to
contract work.
Occupational wage survey programs are not designed
to supply mechanical answers to questions of pay policy.
As suggested earlier, limitations are imposed in the selec
tion and definition of industries, of geographic units for
which estimates are developed, of occupations and asso
ciated items studied, and in determination of periodicity
and timing of particular surveys. Depending upon user
needs, it may be necessary to interpolate for occupations
or areas missing from a survey on the basis of knowledge
of pay relationships.
Because of variation among establishments in the pro
portion of workers in the jobs studied and in the general
level of pay, the survey averages do not necessarily reflect
either the absolute or relative relationship found within
the majority of establishments. As mentioned earlier,
however, area wage survey bulletins provide some insights
into pay relationships within establishments through
special analytical tables.
The incidence of incentive pay systems may vary greatly
among the occupations and establishments studied.
Because average hourly earnings of incentive workers
generally exceed those of time-rated workers in the same
job, data are shown separately wherever possible for the
two groups in industry surveys. Incentive plans apply to
only a very small proportion of the workers in the indirect
plant jobs studied in the area wage program.
Although survey-to-survey changes in pay averages for
a job or job group primarily reflect general wage and
salary changes or merit increases received by individuals,
these averages also may be affected by other factors.
Common among these are labor turnover, labor force
expansions and reductions for other reasons, and changes
in the proportion of workers employed in high- and lowpaying establishments. A labor force expansion might
increase the proportion of workers in lower paid, entry
type jobs and thereby tend to lower the average; or the
closing of a relatively high-paying establishment could
cause average earnings in the area to drop.
Much of this problem has been overcome for area wage
survey measures of pay change by holding establishment
employment constant while computing percent increases
in earnings. That is, the previous and current survey earn
ings of each establishment are weighted by that establish
ment’s employment at the time of the previous survey.
Under this system, measurement of change is limited to
establishments included in two consecutive surveys.
The effect of employment shifts among occupations
between survey dates also is eliminated in measuring
average earnings increases for workers covered by the
p a t c survey and by the machinery industry wage survey.
Employment shifts among establishments or turnover of
establishments included in survey samples, however, are
not controlled in these computations, as they are in
calculating area wage survey trends.
In general, the occupational wage survey programs are
designed to measure pay levels and pay structures at
specified points of time, rather than wage trends. For this
reason, users are directed to other b l s series that are
more appropriate indicators of wage change, such as the
Employment Cost Index (see chapter 8 of this bulletin).
Reliability o f surveys. Results of the surveys are sub
ject to both sampling and nonsampling error. Sampling
errors occur because observations come from a sample,
not the entire population or universe defined for a survey.
They will not be uniform for the occupations studied
because the dispersion of earnings among establishments
and the frequency of occurrence of the occupations dif
fer. The sample is designed so that the chances are 9 out
of 10 that the published estimates on average earnings
generally do not differ by more than 5 percent from the
average that would be obtained by studying all
establishments in the survey universe.
The sampling error of the percentage of workers receiv
ing any given employee benefit differs with the size of
the percentage. However, the error is such that rankings
of predominant practices almost always will appear in
47
their true position. Small percentages may be subject to
considerable error but will always remain in the same scale
of magnitude. For instance, the proportion of employees
in establishments providing more than 5 weeks’ paid vaca
tion to long-service employees may be given as 2 percent,
when the percentage for all establishments might be only
1 percent. Such a sampling error, while considerable, does
not affect the essential inference that the practice is a rare
one.
Estimates of the number of workers in a given occupa
tion may have considerable sampling error, due to the wide
variation among establishments in the proportion of
workers found in individual occupations. (It is not unusual
to find sampling errors of as much as 20 percent.) Hence,
the estimated number of workers can be interpreted only
as a rough indicator of the relative importance of various
occupations. The greatest degree of accuracy in these
employment counts is for occupations found principally
in large establishments.
Since completely current and accurate information
regarding establishment products and the creation of new
establishments is not available, the universe from which
the sample is drawn may be incomplete. Sample firms
incorrectly classified are accounted for in the actual field
work, and the universe estimates are revised accordingly.
Those firms which should have been included but were
classified erroneously in other industries cannot be
accounted for.
Since some measure of judgment enters into the
classification of occupations and other characteristics,
there is some reporting variability in the results. A repeti
tion of the survey in an establishment with different
interviewers and respondents would undoubtedly pro
duce slightly different results. Hence, analyses based on
a small number of respondents must be used with care,
even when all eligible establishments are included.
However, when spread over a large number of estab
lishments the differences, being random, would tend to
balance out. No evidence of any consistent error has been
uncovered.
Nonsampling errors can come from a number of other
sources, including inability to obtain information from
some establishments, definitional difficulties, inability of
respondents to provide correct information, and errors
in recording and coding the data obtained or estimating
for missing data. Although not specifically measured, the
surveys’ nonsampling errors are likely to be minimal due
to relatively high response rates, well-trained field
representatives, careful review of the data, and other
survey controls and procedures.
Technical References
Barsky, Carl B., and Personick, Martin E. “ Measuring Wage
Dispersion: Pay Ranges Reflect Industry Traits,”
M onthly Labor Review, April 1981, pp. 35-41.
Morton, John D. “ b ls Prepares to Broaden Scope of Its
White-collar Pay Survey,” Monthly Labor Review,
March 1987, pp. 3-7.
Buckley, John E. “ Wage Differences Among Workers in
the Same Job and Establishment,” M onthly Labor
Review, March 1985, pp. 11-16.
Personick, Martin E. “ White-collar Pay Determination Under
Range-of-rate Systems,” M onthly Labor Review,
December 1984, pp. 25-30.
Carlson, Norma W. “ Time Rates Tighten Their Grip on Man
ufacturing Industries,” M onthly Labor Review, May
1982, pp. 15-22.
Douty, H.M. “ A Century of Wage Statistics: The bls Con
tribution,” Monthly Labor Review, November 1984,
pp. 16-28.
Doyle, Philip M. “ Area Wage Surveys Shed Light on Declines
in Unionization,” Monthly Labor Review, September
1985, pp. 13-20.
Personick, Martin E., and Barsky, Carl B. “ White-collar Pay
Levels Linked to Corporate Work Force Size,” Monthly
Labor Review, May 1982, pp. 23-28.
Scofea, Laura, “ b ls Area Wage Surveys Will Cover More
Areas,” Monthly Labor Review, June 1986, pp. 19-23.
Van Giezen, Robert W. “ A New Look at Occupational Wages
Within Individual Establishments,” M onthly Labor
Review, November 1982, pp. 22-28.
U.S. Department of Labor, Bureau of Labor Statistics,
Measures o f Compensation, Bulletin 2239, 1986.
48
bls
Chapter 7. Negotiated Wage
and Benefit Changes
The Bureau of Labor Statistics prepares information
on current changes in wages and supplementary benefits
agreed to in collective bargaining. The information
includes monthly listings of companies, employer associa
tions, or governmental units in which such changes have
occurred, the unions involved, and the nature of the
change, bl s also prepares quarterly and annual statistical
summaries of negotiated wage changes in all major col
lective bargaining situations in private industry, and
semiannual summaries for State and local government
bargaining units.
Background
began publication of the monthly listing of set
tlements in 1948, when prices and wage rates were rising
rapidly and interest grew in determining the extent to
which settlement patterns spread from industry to
industry. The statistical series summarizing wage changes
was initiated in 1949; regular quarterly publication was
begun in 1954. In 1964, with the increasing importance
of supplementary benefits such as various forms of
premium pay, paid leave, and employer payments for
health, insurance, and pension benefits, the Bureau began
to estimate the size of negotiated changes in total
compensation—the wage and benefit package. Beginning
with 1979, similar data have been published for State and
local government bargaining units.
bls
Description of Statistical Series
Coverage
Private industry agreements. The series summarizes
wage rate changes in major collective bargaining set
tlements (settlements covering 1,000 workers or more) for
production and related workers in manufacturing and
nonsupervisory workers in nonmanufacturing, b l s cur
rently follows about 1,350 bargaining situations, for vir
tually complete coverage of major agreements. Changes
in total compensation are measured for agreements cover
ing 5,000 workers or more in all industries and 1,000
workers or more in construction.
Contracts covering multiplant firms are included if the
agreement as a whole covers 1,000 workers even though
each plant employs fewer workers. Also included are con
tracts with trade associations or with groups of firms that
bargain jointly with a union or unions even though the
firms are not associated formally and each has fewer than
the minimum number of workers within the scope of the
series. When two or more unions, together representing
at least 1,000 workers but individually representing fewer
than 1,000, negotiate essentially identical contracts with
one or more firms, the workers involved are considered
to constitute one bargaining unit.
State and local government agreements. This series sum
marizes wage and benefit changes for workers in State
and local governments where: (1) a labor organization
is recognized as the bargaining agent for a group of
workers; (2) the settlements are embodied in signed,
mutually binding contracts; and (3) wages are determined
by collective bargaining. When introduced in 1979, this
series presented wage and benefit measures for units of
5,000 workers or more. Beginning with 1984 data, the
coverage for wage data was expanded to units of 1,000
workers or more. As of 1986, this series covers 2.3 million
workers in 612 bargaining units. About one-half of all
State and local government employees covered by collec
tive bargaining agreements are included in the series.
Data presented
Wages. Two types of information are presented on
wages. Settlement data measure wage adjustments
specified in the bargaining settlements reached during a
particular time period (e.g., quarter or year). They reflect
decisions to increase, decrease, or not change wages.
These data exclude wage changes that may occur under
cost-of-living adjustment ( c o l a ) clauses which link the
size of future wage adjustments to changes in the Con
sumer Price Index. Lump-sum payments, which are
typically negotiated instead of wage increases or to off
set wage decreases, are also excluded. Both the adjust
ments scheduled during the first 12 months of the contract
(first-year changes) and the total of wage adjustments
scheduled over the life of the contract, expressed as an
annual rate, are presented. There are two measures of
wage changes that are implemented in a reference period.
The effective wage adjustment measure includes
increases, decreases, and no change in wages during
49
the reference period for all workers in the series. The
increases and decreases, which stem from settlements
reached in the period, agreements reached in a prior
period, and c o l a clauses, are reflected in the effective
wage change measure, which relates only to workers
whose wages change.
Compensation. Although at one time the economic
terms of collective bargaining settlements involved wage
rates almost exclusively, today, a wide variety of benefits
are also involved. “ Compensation” refers to the total of
pay and benefits. As with wage data, the Bureau publishes
compensation data for settlements reached during a
period, but limited to settlements covering at least 5,000
workers in all industries and at least 1,000 workers in con
struction. Adjustments scheduled for the first year of the
contract, and those scheduled over the entire contract
term, expressed as an annual rate, are published. The
compensation measures exclude c o l a adjustments and
lump-sum payments.
Data Sources
Calculations of the size of negotiated wage and benefit
changes are based on actual characteristics of the work
force affected by the settlements. These include the
distribution of workers by occupation, earnings, and
length of service. When estimates of compensation
changes are made, data are also obtained on employer
costs for various benefits. The data on work force
characteristics and benefit costs are usually obtained
directly from the companies as part of a variety of b l s
surveys. (Data for these surveys are collected under a
pledge that they will be kept confidential and not released
outside the Bureau.) Other data sources for these calcula
tions include the file of union contracts maintained by
b l s , the file of pension and insurance benefit agreements
and financial information maintained by the Department
of Labor’s Labor-Management Services Administration,
and secondary sources. Secondary sources, including
general circulation newspapers and periodicals and union,
management, and trade publications, are used in produc
ing listings of agreements.
Estimating Procedures
Procedures for pricing settlements center around three
questions: (1) For which items in a collective bargaining
settlement are costs to be determined? (2) How are the
costs to be determined? (3) How are the costs to be
expressed?
Items included in pricing
Many terms of a union-management agreement besides
wage and benefit provisions may affect an employer’s
costs. For example, seniority provisions may influence
costs through their effect on employee efficiency. Such
effects, however, are not measurable. Consequently,
the BLS program is confined to measuring the wage and
benefit components; i.e., to measuring the effect of set
tlements on employer outlays for employee compensa
tion. Included are: Changes in wage rates; modifications
in premium pay, paid leave, and severance pay; and
adjustments in employer payments for pension, health
and welfare, and supplemental unemployment benefits,
excluding the costs of administering these benefits. Also
included are changes in contract provisions specifying
paid time for clothes change, washup, and lunch periods.
Excluded are items which, although related to compensa
tion, are not normally considered part of compensation
such as per diem payments, moving expense reimburse
ments, payments for safety clothing, and provision of
facilities or services such as parking lots and health units.
Indirect effects of settlements are ignored; factors such
as possible extension of settlement terms to nonunion
workers in the same firm or to members of other bargain
ing units are not considered. Similarly, although the cost
of providing lengthened vacations is measured (by the
wages and salaries paid for the additional time off), the
cost of hiring vacation replacements, if necessary, is not
measured. Moreover, effects on unit labor costs, which
involve consideration of employee efficiency as well as
employer payments, are disregarded.
Determination of costs
Since a value is placed on settlements at the time they
are reached, the costs attributed to them are estimates
of outlays to be made in the future; they cannot be taken
from employers’ accounting records. The estimates are
made on the assumption that conditions existing at the
time the contract is negotiated will not change. For exam
ple, analysts assume that methods of financing pensions
will not change, and that expenditures for insurance will
not change except as a result of altered benefit provisions
or modified participation because of changes in company
contributions. They also assume that the composition of
the labor force will not change.
Except for any guaranteed increases, which are treated
as deferred adjustments, possible wage rate changes that
may result from c o l a clauses are excluded because it is
impossible to predict changes in the Consumer Price
Index. Lump-sum payments (e.g., those made instead of
wage increases, performance bonuses, and attendance
bonuses) are also excluded.
Estimates of compensation changes attempt to measure
the costs associated with actual characteristics of the work
force affected by the settlements, not the costs for some
hypothetical employee group. Estimates based on the
actual age, length of service, sex, and skill characteristics
50
Expressing costs
of the workers involved recognize that the choice in incor
porating alternative benefit changes into contracts is
affected by their costs, which, in turn, are affected by
the character of the work force. For example, an extra
week of vacation after 15 years of service will cost very
little when only 10 percent of the workers have that much
service, but will add about 1 percent to the annual cost
of straight-time pay for working time when half of the
workers have been employed for 15 years or more.
Changes in wage rates affect costs for certain bene
fits that are linked to wage rates such as paid leave,
Social Security, and pensions based on earnings. This
effect, variously referred to as “ creep,” “ bulge,” or
“ rollup,” is reflected in estimates of changes in
compensation.
Many items in a collective bargaining agreement are
priced without difficulty. This is particularly true when
settlement terms are expressed as cents-per-hour adjust
ments; e.g., a 20-cent-an-hour general wage increase or
a 5-cent increase in employer contributions to a health
and welfare fund. These stipulated cents-per-hour figures
are used as the costs of the settlement provisions. Per
centage wage adjustments are converted to cents-per-hour
figures on the basis of current average straight-time
hourly earnings in the bargaining unit.
Although less direct, the cost of an additional holiday
is estimated adequately by prorating 8 hours’ average pay
(if the normal workday is 8 hours) over the number of
annual working hours per employee. The cost of an addi
tional week of vacation for 25-year employees is estimated
similarly, but one must know the number of employees
with the required seniority.
Other settlement terms are more difficult to price. For
example, the cost of an unfunded severance pay plan
depends not only on plan provisions but on the frequency
of layoffs, which at best, is hazardous to estimate. Pen
sion improvement costs are particularly difficult to
estimate because employers often have considerable
discretion in funding their obligations, b l s assumes that
a pension benefit change will change existing expenditures
for current service proportionately. Since employer con
tributions for pensions frequently vary widely from year
to year, outlays in several past years are examined to
develop a measure of current payments.
For most provisions, b l s estimates are of actual cash
outlays to be made by employers. In the case of paid leave
provisions, however, an improvement may entail time off
for workers, without additional cash payments by the
employer. Since payment per hour worked will rise, this
change is taken as the cost effect of the settlement provi
sion. For a reduction in the basic workweek, the increase
in hourly rates needed to maintain weekly pay is the major
item priced. A reduced basic workweek may be accom
panied by additional overtime work; unless this overtime
is specified in the agreement, it is ignored in the cost
estimate.
The cost of a given settlement is obtained by summing
the costs (in cents per hour worked) of each wage (and,
if measured, benefit) change. This sum is then expressed
as a percent of wages (or compensation) to facilitate com
parisons of agreements by eliminating the influences of
payroll size and wage level.
Expression of costs as a percent of wages (or compen
sation) requires estimation of an appropriate base (total
wages or total compensation) as well as the cost of the
settlement terms. The base used by the Bureau consists
of current outlays per hour worked for wages (or for all
negotiable items of employee compensation plus employer
expenditures for legally required social insurance). The
overall percentage change generated by each settlement
is weighted by the number of workers affected (the pric
ing of individual settlements is not disclosed). The sum
of the worker-weighted changes is divided by the total
number of workers covered by settlements (including set
tlements that did not change wages or compensation) to
determine the average percent adjustment. Effective wage
adjustment data are handled in similar fashion. Since col
lective bargaining agreements generally are for 2-year
periods or longer, b l s expresses the total percent adjust
ment over the contract term at an annual rate to permit
comparison among agreements for differing time spans
as well as to facilitate the use of the data in conjunction
with other statistical series. The annual rates of adjust
ment take into account the compounding of successive
changes. In addition, the Bureau computes first-year
adjustments as a percent of current hourly wages (or
compensation).
Contracts are considered to run from their effective
dates to their termination dates. However, where there are
wage reopening clauses, the reopening date is taken as the
termination date, and any agreement under the reopen
ing clause is treated as a new settlement. Sometimes, the
parties to a contract agree to an unscheduled contract
reopening. Beginning with full-year data for 1981 (pub
lished in January 1982), compensation changes negotiated
under unscheduled reopenings are included in the data
for new settlements. Their exclusion from earlier data on
settlements made no noticeable difference because, prior
to 1981, they were rare; and, when they occurred, they
usually changed compensation for the balance of the
contract that was already in place, typically no more than
1 year. In 1981, unscheduled reopenings became more
frequent and usually resulted in new contracts that ran
2 to 3 years.
Presentation
The listing of current changes in wages and benefits
is published monthly in the periodical Current Wage
51
Developments ( c w d ). Private industry summaries are
grouped by industry, and government listings by func
tion. The listings include the name of the employer and
(when applicable) the union, the number of workers
involved, the amount and effective date of the change,
details of complex changes, and the reason for the change
(i.e., whether it is a new settlement, a deferred increase,
or a c o l a ).
Statistical summaries of preliminary data on settle
ments and total effective wage and benefit adjustments
in private industry are issued first in news releases in the
month following each quarter and then in c w d . Final
quarterly and annual data are presented in a summary
article published in the Monthly Labor Review and
detailed data are published in c w d each year.
Statistical summaries of State and local government
bargaining settlements are issued in news releases semi
annually in February and August and also appear in
CWD.
Uses and Limitations
The series on wage and compensation adjustments
resulting from collective bargaining is one of the Federal
Government’s principal economic indicators. As such, it
is used by a variety of Federal agencies including the
Council of Economic Advisers, the Federal Reserve
System, and the Congressional Budget Office, for a broad
range of purposes including determining trends in com
pensation and forecasting changes in wage and salary
income and gross national product. The statistics, as well
as the monthly listings, are used by the Federal Media
tion and Conciliation Service; State and local government
agencies; employer and employee organizations; economic
consultants; and researchers and practitioners in industrial
relations, collective bargaining, and economic forecasting.
The user of the compensation data should remember
that the series does not measure all changes in average
hourly expenditures for employee compensation. In
calculating compensation change estimates, a value is put
on the benefit portion of the settlements at the time they
are reached on the assumption that conditions existing
at the time of settlement will not change. The data are
estimates of negotiated change, not total changes in
employer cost.
However, changes in the existing conditions do occur:
In the volume of overtime and shift work, in the com
position of the work force, in the level and stability of
employment, and in factors affecting incentive earnings,
for example. These changes influence outlays for
employee compensation. In some instances, these changes
are introduced by management specifically to offset costs
of new labor agreements. In other cases, changes are the
result of modified production schedules or of technolog
ical developments independent of collective bargaining,
and may influence the cost of the union-management
settlement.
Public and private sector negotiated compensation data
are not strictly comparable because of differences in
bargaining practices and settlement characteristics. Two
differences are the incidence of lump-sum payments and
cost-of-living ( c o l a ) clauses. Lump-sum payments are
rare in government but common in private industry.
c o l a clauses are included in no State and only a few
local government settlements but are in a substantial
number of private industry settlements. Both lump-sum
payments and potential wage changes resulting from
c o l a clauses are excluded from the settlement data. Fur
thermore, State and local government bargaining fre
quently excludes pension benefits, which are often
prescribed by law. In private industry, pensions are
typically a bargaining issue.
52
Chapter 8. Employment
Cost Index
The Employment Cost Index (ECi) measures the rate
of change in employee compensation, which includes
wages, salaries, and employers’ cost for employee bene
fits. The ECi was developed in response to a frequently
expressed need for such a statistical series. Existing meas
ures, while adequate for specific purposes, were found
to be fragmented, limited in industrial and occupational
coverage, insufficiently timely or detailed, or subject to
influences unrelated to the basic trend in employee
compensation.
Several elements distinguish the ECI from other surveys
of employee compensation. It is comprehensive in that
it (1) includes costs incurred by employers for employee
benefits in addition to wages and salaries; and (2) covers
all establishments and occupations in both the private
nonfarm and public sectors.1 It measures the change in
the cost of employing a fixed set of labor inputs, so it
is not affected over time by changes in the composition
of the labor force. The survey is timely in that statistics
are published quarterly, approximately 1 month after
their reference date. The ECI also enables users to com
pare rates of change in detailed occupational, industrial,
geographic, union coverage, and ownership (publicprivate) submeasures.
Background
The ECi survey is being implemented in stages.
Initially, beginning in 1976, published statistics covered
quarterly changes in wages and salaries for the private
nonfarm economy, excluding establishments in Alaska
and Hawaii, and private household workers. In Novem
ber 1978, the survey was expanded to include estab
lishments in Alaska and Hawaii, and an additional 13
statistical series (union/nonunion manufacturing and
nonmanufacturing, for example) were published.
The second major stage was completed in 1980 with
the publication of quarterly changes in total employee
compensation.
The third stage involved expansion of the survey to
State and local government units. With the inclusion of
these government units in November 1981, the overall
1 Coverage of the private sector is limited to the private nonfarm
economy, excluding private household workers. Public sector coverage
includes employees of State and local governments, but excludes workers
in the Federal Government.
series now represents the civilian nonfarm economy,
excluding households and the Federal Government.
The most recent development of the ECI is the publica
tion in 1987 of compensation cost levels. Data collected
for the ECi can be used to calculate cost levels with no
additional burden on survey respondents. The cost levels
use current employment weights derived from b l s ’s Cur
rent Employment Statistics survey and the ECI sample.
The cost levels, with a March reference date, are pub
lished annually during mid summer.
Future development of the ECI will include increases
in the number of published series, especially in the serviceproducing sector of the econom y, and expansion to
include the Federal Government.
Description of the ECI
Major features
The e c i is a measure of change in the price of labor
defined as compensation per employee hour worked. The
self-employed, owner-managers, and unpaid family
workers are excluded from coverage.
The e c i is designed as a Laspeyres, fixed-weight index
at the occupational level, thus eliminating the effects of
employment shifts among occupations. The index weights
are derived from occupational employment for ECI
industries reported in the 1980 Census of Population. The
weights remain fixed from period to period pending a
major index revision, next scheduled to occur when the
results of the 1990 census are incorporated.
The index is computed from data on compensation by
occupation collected from a sample of establishments
and occupations weighted to represent the universe of
establishments and occupations in the economy. The
wage and salary component of the index is represented
by average straight-time hourly earnings in an occupa
tion. Straight-time earnings are defined as total earnings
before deductions, excluding premium payments for over
time, weekend, and late-shift work. Earnings include pro
duction bonuses, commissions, and cost-of-living
allowances but exclude nonproduction bonuses (which are
considered a benefit in the e c i ), payments in kind, room
and board, and tips.
All earnings are computed on an hourly basis, whether
or not this is the actual basis of payment. Earnings of
53
salaried employees and those paid under incentive systems
are converted to an hourly basis. Benefit cost data are
also converted to an hourly basis. Thus, occupational
hourly earnings plus the employer’s cost per hour worked
for employee benefits constitute the price of labor in the
ECl.
Since pay rates generally relate to the job rather than
to the incumbent workers, the basic unit of data collec
tion is a job, as defined by the firm, in an establishment.
Shifts in employment among jobs and establishments are
controlled by measuring wage change for the same jobs
in the same establishments and applying fixed employ
ment weights to the results. The unit of observation is
standardized to a certain extent below the job level by
measuring only selected types of labor within the job;
e.g., full or part time, incentive or time rated, depending
on the predominant type.
The benefit data portion of the ECl encompasses 22
distinct benefit categories, which can be grouped as
follows:
cents-per-hour-worked cost of each benefit provided
employees in an occupation. For example, the data ele
ment for vacations might be expressed as follows: For
an occupation in an establishment, the average worker
received 2.8 weeks of paid vacation. In order to convert
the data element to a cents-per-hour-worked cost, addi
tional information covering workers in the occupation
is needed. Therefore, data are also collected on scheduled
daily and weekly hours and annual weeks. The follow
ing example illustrates the calculation of the cents-perhour-worked cost for a benefit:
CALCULATING THE COST PER HOUR WORKED OF
A BENEFIT—Example:
Data element—2.8 average weeks of vacation
Scheduled weekly hours—40
Straight-time average hourly rate—$6.95
Annual hours worked (computed by data processing
system)—1,950
Paid leave benefits
1.
2.
3.
4.
2.8 weeks/year x 40 hours/week X $6.95/hour
Vacations
Holidays
Sick leave
Other paid leave
= $0.399/hour
1,950 hours/year
THIS EQUATION CAN BE BROKEN INTO THE
FOLLOWING STEPS:
Supplemental pay
5. Premium pay for overtime and work on holidays
and weekends
6. Shift differentials
7. Nonproduction bonuses
2.8 weeks/year x 40 hours/week = 112 (average annual hours of
vacation)
112 hours/year x $6.95/hour = $778.40 (average annual cost of
vacation)
Insurance
($778.40/year) / (1,950 hours/year) = $0,399 (average cost per
hour worked for vacation)
8. Life insurance
9. Health benefits
10. Sickness and accident insurance
Note that average annual hours of vacation are also used by the
data processing system to compute annual hours worked.
Pension and savings plans
11. Pension and retirement benefits
12. Savings and thrift plans
Legally required benefits
13.
14.
15.
16.
17.
18.
19.
20.
Social Security
Railroad retirement
Railroad supplemental retirement
Railroad unemployment insurance
Federal Unemployment Tax Act
State unemployment insurance
Workers’ compensation
Other legally required benefits
Other benefits
21. Severance pay
22. Supplemental unemployment benefit funds
Merchandise discounts in department stores, currently
included as a benefit, will be deleted from the group of
benefits covered in September 1988.
The benefit data supplied by respondents normally
consist of data elements which are used to compute the
The nature of the data collected varies somewhat
depending upon the particular benefit. For paid leave
benefits, the data element is usually expressed in terms
of average number of days, weeks, or hours per year. For
the insurance benefits, the data element may consist of
a rate per thousand dollars of life insurance coverage or
of a rate per month for family medical insurance
coverage. In the case of the legally required benefits, a
tax rate and taxable earnings ceiling are usually collected.
Whatever the form of the data element, the benefit cost
is always converted to cents per hour worked.
Occupational classification
The e c i occupational classification system was
originally based on the classification system used for the
1970 Census of Population. In June 1986, the occupa
tions being surveyed were recoded to the classification
system used in the 1980 census, which is based on the
54
Standard Occupational Classification (soc) system.2 The
Census system classifies all occupations reported into 503
3-digit occupational categories (such as accountant,
stockhandler, etc.) which are then combined into 13
major occupational groups.
For ECi purposes, four of the Census groups are com
bined into two groups (professional and technical workers
are combined, as are two categories of service workers).
Also, the Census groups covering private household
occupations and some farming, forestry, and fishing
occupations include workers outside the scope of the
survey and are, therefore, excluded. As a result of these
modifications, the ECI includes the following nine major
occupational groups ( m o g ’s):
1.
2.
3.
4.
5.
6.
Professional specialty and technical occupations
Executive, administrative, and managerial occupations
Sales occupations
Administrative support, including clerical occupations
Precision production, craft, and repair occupations
Machine operator, assembler, and inspector occupa
tions
7. Transportation and material moving occupations
8. Handler, equipment cleaner, helper, and laborer occu
pations (including forestry and fishing occupations
within the scope of the eci)
9. Service occupations
The Census occupational classification system only lists
occupations to be included under each of the 503 occupa
tional categories. For data collection purposes, definitions
of the Census occupations have been developed.3
Industrial classification
The e c i currently covers all nonfarm establishments
classified in the 1972 edition of the Standard Industrial
Classification Manual (sic), with the exception of private
households and the Federal Government. No minimum
establishment size cutoff is used. The e c i publishes
statistics for all major industry divisions with the excep
tion of mining. Selected industry divisions are presented
in more detail; for example, within manufacturing,
durable and nondurable goods and within services, health
services and hospitals. Statistics are also published for
goods-producing and service-producing industries.
Geographic classification
The geographic coverage o f the ECI includes all States
and the District of Columbia. Rates of change in wages
and salaries are published using the four-region classifica
tion system shown in appendix C. Statistics are also
2 C lassified In dex o f In dustries an d O ccupations, 1980 C ensus o f
P opu lation (Bureau of the Census, 1980).
3 E m p lo y m en t C o st In dex O ccu pation C lassification S ystem
M an u al-1980 (Bureau of Labor Statistics, January 1985).
published for metropolitan areas (establishments located
in a Metropolitan Statistical Area) and for other areas.
Union classification
Occupations surveyed within an establishment are
classified as union if: (1) the majority of workers in the
occupation are represented by a labor organization which
is recognized as their bargaining agent; (2) wages are
determined by collective bargaining; and (3) settlements
are embodied in signed, mutually binding collective
bargaining contracts.
Data Sources and Collection Methods
The wage, salary, and benefit cost data from which the
is computed are obtained from a sample of more
than 3,000 establishments in private industry and about
700 establishments in State and local governments, and
from a sample of jobs within those establishments.
Data collection is initiated by a bls field representative
who visits the reporting unit. The purposes of the initial
visit are to: Introduce the ECI program and obtain
cooperation; determine organizational unit or units for
establishment coverage; select occupations; develop
establishment reporting procedures; and complete the
first schedule. Quarterly reports thereafter are normally
collected by mail or telephone by the bls regional office.
Prior to 1987, a major task in the initial contact by
a bls field representative was to classify all company jobs
into major occupational groups (m o g ’s). The job
matching procedure sought to obtain at least one match
for each of the nine m o g ’s surveyed by the e c i . Wage,
salary, and benefit cost data were then collected for the
selected jobs. In certain cases, data were requested for
two or more company jobs within a single MOG if it
accounted for a significant proportion of employment in
an industry. Beginning with companies visited in 1987,
the e c i uses a Reduced Job Match procedure which
involves a request of data for four to eight company
jobs, with the jobs selected strictly on a probabilityproportionate-to-size basis. There is no longer an attempt
to obtain at least one observation for each of the nine
MOG’s. The number of job matches sought varies with
establishment employment size.
The job-matching process results in the selection of
company jobs which are at the most detailed level recog
nized by that company. Examples would be clerk III and
senior attorney-litigation. During the job-matching proc
ess, characteristics of the company jobs are also deter
mined—whether the groups selected consist of full- or
part-time workers, time or incentive workers, and whether
they are covered by collective bargaining agreements.
The wage data are collected on a “ shuttle” form which
is sent to the respondent each quarter for the addition
eci
55
of new data (see ECi Wage Data Form at the end of this
chapter). The survey months are March, June, Septem
ber, and December; the data relate to the pay period
which includes the 12th day of the month.
Benefit data are initially reported in detail, including
such information as vacation provisions by length-ofservice categories; the length-of-service distribution of
occupational employment (used to compute the cost of
vacations); and employer contributions for pensions,
insurance, and other benefits. Then, each quarter, the
information on benefit provisions is summarized and sent
to the respondents to review and to report any changes
which have occurred since the prior quarter. For exam
ple, in the prior quarter, the respondent might have
reported that 9 of the 10 employees in a surveyed occupa
tion subscribed to a health insurance plan which cost $115
per month. During the quarterly update, the respondent
indicates that the cost of the plan has increased to $129
per month. In both the prior and current quarter, the
employer assumed 50 percent of the plan’s cost.
For ECi purposes, the average cost for workers in the
prior quarter equaled $51.75 per month. (The employer’s
share of the cost for each worker participating in the plan
is $57.50. Ninety percent of the workers participate,
$57.50 X 0.90 = $51.75.) The current quarter’s cost of
the plan would equal $58.05 ($64.50 X 0.90 = $58.05).
Note that the 90-percent participation rate was held con
stant. This would be changed only if the employee con
tribution rate (50 percent of plan cost) increased or
decreased. Holding the participation rate constant
eliminates the effects of forces such as shifts in work force
composition on the measurement of the cost change.
Similarly, when an employer changes an overtime pay
provision, new overtime hours worked are not normally
collected. Instead, the base period overtime hours worked
pertaining to the altered provision are repriced using the
new overtime rate. This practice restricts changes in over
time cost to changes caused only by the adoption of a
new overtime rate and eliminates the effect of changes
in the number of hours of overtime worked.
Survey Design
The ECi sample design has evolved over the 12-year
history of the program. Separate designs have been used
for the public and private sectors of the economy,
although, starting in 1987, the designs of all replacement
samples are similar.
Private sector—respondent universe and sample
design
The original sample design used for the selection of the
e c i sample in 1975 consisted of a two-phase controlled
selection.4 In the first phase, approximately 23 occupa
tions were identified for each 2-digit sic. Using the 1970
Census of Population, the largest five occupations in each
2-digit SIC were selected with certainty. Then one to four,
but generally two, occupations were selected from each
major occupational group within the 2-digit SIC using a
probability-proportionate-to-size method. A sample of
approximately 10,000 establishments5 was selected from
a larger BLS survey of approximately 200,000 establish
ments drawn primarily from the unemployment insurance
universe. The first phase of the ECI survey determined
the occupational employment within each of the 10,000
sampled establishments for each of the 23 selected occu
pations for the 2-digit sic. Imputation was used for par
tial and complete nonrespondents. Using measures of size
designed to enhance the probability of selection of estab
lishments with a large proportion of the employment in
any of the 23 occupations, a subsample of approximately
2,000 establishments was selected, the selection being
done separately within each 2-digit sic. Data were col
lected for the selected occupations within the selected
establishments.
Beginning in 1981, the ECI began replacing the entire
private sector sample using a new sample design. Within
each 2-digit SIC, all detailed Census occupations were
assigned to one of 9 to 15 occupational groups, each con
sisting either of all of the occupations within a major
occupational group ( m o g ), one or more closely related
occupations within a m o g , or the residual occupations
within a MOG. The 2-digit SIC’s were divided into 12
groups that replaced the existing samples over a 4-year
period. The new design was completely implemented in
1986. Allocation of the sample was made proportionate
to the employment of each 2-digit SIC with an initial total
sample size of about 2,000 establishments.6 Within each
establishment, one occupation was selected from each
occupational group with probability of selection propor
tionate to the employment of the occupation within the
group. This was the sample in use at the time of publica
tion in 1987.
Beginning in 1987, the within-establishment occupation
selection methodology was changed to eliminate the initial
classification of all establishment jobs into groups. This
change reduced the collection burden on both respondents
and BLS and improved weight computation. With the new
Reduced Job Match procedure, a sample of four to eight
4 R. Goodman and L. Kish, “ Controlled Selection, a Technique in
Probability Sampling,” Journal o f th e A m erican S tatistical A sso cia
tion, Vol. 45, 1950, pp. 350-72.
5 The term establishment generally indicates a single physical loca
tion. In the public sector, many of the establishments have units at more
than one location. For example, school districts meet the SIC manual’s
criteria for an establishment, but the majority of school districts are
comprised of units in several different locations.
6 The total private sector sample size had grown to about 3,000
establishments by 1987. Allocation to 2-digit SIC’s is now based partly
on generalized variance estimates so as to minimize the variance of
national estimates of annual relatives of total compensation.
56
jobs, the number depending on the size of the establish
ment, is selected. The jobs are selected from either a list
of establishment employees or a list of establishment
employment by job title, using probability proportionate
to employment in the selection of the jobs. Data are col
lected for a homogeneous group of employees, matching
characteristics of either a selected individual or a selected
job. The first data collected using this new methodology
will be introduced into the ECi in 1988. The entire private
sector sample will be replaced using this methodology by
1992.
school. The sampling frame was ordered by region, and
within region, by size of enrollment. When occupational
groups for schools were defined, a phase I survey was
conducted by mail to determine the employment within
each of the groups for the selected schools.
The next stage was to calculate employment estimates
for units not responding to the mail survey. The balance
of the survey design was similar to that of the private sec
tor, with the exception of the subsampling of occupa
tional groups at the time of initiation. The final sample
consisted of 260 establishments.
Public sector—respondent universe and sample
design
Hospitals
Because of the nature of the available sampling frames,
the public sector was divided into four parts: Schools,
hospitals, State and large local governments (all sic ’s
except schools and hospitals), and small local govern
ments. Each has a somewhat different survey design.
As in the first series of private industry replacement
samples, Census occupations were combined into occu
pational groups. When a group is matched in an estab
lishment, a single, detailed job title is selected using
probability-proportionate-to-employment sampling pro
cedures. The use of occupational groups and the sam
pling of a specific job title increase the probability of
finding occupational matches while retaining the advan
tages of surveying narrowly defined occupations.
The procedures described below apply only to the
government sector samples through 1987. Recent im
provements in the public sector coverage of the unem
ployment insurance (ui) universe will permit the selection
of replacement samples for most of this sector directly
from the UI files. The replacement sample for public sec
tor hospitals will be entered into index estimation in 1988,
with the remaining State and local government sectors
following within 2 years. Also, these public sector replace
ment samples will use the Reduced Job Match selection
procedure described above.
Schools
The sampling frame for public elementary and second
ary schools was the 1973-74 National Center for Educa
tion Statistics (n c e s ) listing of all State and local schools.
The frame included most of sic 821 (elementary and
secondary schools); the remainder of sic 821 is covered
in the other parts of the public sector sample. The sam
pling frame for higher education was the 1973-74 n c e s
list of all higher education schools, which covered all of
Sic 822 (colleges, universities, professional schools, and
junior colleges).
Establishments were stratified by 3-digit sic; then, with
a certainty cutoff,7 a sample was selected with proba
bility of selection proportionate to enrollment within the
No survey of occupational employment was undertaken
for public hospitals because of potential nonresponse.
Instead, public hospitals were stratified by ownership and
Census region and selected in a single stage, again with
probability of selection proportionate to size and with a
certainty cutoff. The establishment sampling frame was
the 1976 Department of Health, Education, and Welfare
list of public hospitals. The occupational selection was
essentially a systematic sample within each of the 106
establishments in the final sample.
States and large local governments
No universe listing of establishments was available for
State and local governments; it was, therefore, necessary
to conduct a refinement survey to develop a list of poten
tial sample units. This survey was accomplished through
personal visits and allowed for the selection, based on
the respondent’s criteria, of identifiable units within each
jurisdiction and the assignment of major industry divi
sion designations to each unit.
The local government jurisdictions in the refinement
survey (cities, counties, special districts, etc.) were selected
from the 1972 Census of Governments file provided by
the Bureau of the Census. Only jurisdictions with more
than 100 employees were deemed to need refinement. (See
“ Small local governments” below.)
The 3,729 such local jurisdictions were divided into
three size classes: 100-999, 1,000-29,999, and 30,000
employees and over. The 30,000-and-over units were
picked with certainty and the other two groups were fur
ther classified into four Census regions. This provided
eight probability strata from which a probabilityproportionate-to-size selection of the jurisdictions that
would undergo refinement was made. Fourteen jurisdic
tions were selected from the 100-999 size group, and 26
jurisdictions were selected from the 1,000-29,999 size
group. Six jurisdictions were selected with certainty from
the 30,000-and-over group.
7
Certainty cutoff indicates that all units with a measure of size
greater than a specified figure are automatically selected.
57
The selection of States in the refinement survey was
based on size of public employment as well as the need
to have each of the nine Census regions represented. Five
States were selected with certainty based on the number
of State employees, and 11 others were picked with prob
ability of selection proportionate to size.
The above procedures resulted in the selection of
approximately 780 units for which a phase I survey on
occupational employment was conducted. Occupational
employment was requested for nine occupational groups
within each of these 780 units. These units were stratified
by jurisdiction and industry size. The final sample was
randomly selected in such a way that every jurisdiction
in the original refinement sample had at least one of its
establishments selected.
The final step in the sampling process was the selec
tion of occupations for each selected establishment. The
final sample consisted of about 350 establishments, with
approximately seven occupations per establishment.
Small local governments
Due to their small size, no refinement or phase I survey
was done for small local governments (units with fewer
than 100 employees). Any refinement required was
accomplished by bls field representatives at the time the
data were collected. The universe of small local govern
ments contained about 10,000 units. These units were first
stratified by Census region and ordered by type of local
jurisdiction: Municipalities, counties, townships, and
special districts. A probability-proportionate-to-size sam
ple selection was done in each stratum. Thirty final sam
ple units were selected.
Sample replacement
Beginning in 1981, the existing sample of private sec
tor establishments was gradually replaced by a new sam
ple. A few large establishments were included in both the
old and new sample. Sample replacement is necessary to
ensure that sample sizes remain adequate for publication
and that new establishments are represented in the sam
ple, and to limit the burden on individual establishments.
The entire sample will be replaced every 4 to 5 years.
Replacement will be done in stages, with part of the sam
ple being replaced each quarter.
the adjustment factors are calculated and applied only
once, their effects on the estimates are constant for the
duration of the sample.
For wage change estimation after the base period,
values are imputed when there is a temporary nonre
sponse. The basic assumption is that nonrespondents have,
on the average, the same wage movement that respondents
have. Therefore, for a temporary nonresponse, the priorquarter data for an establishment/occupation are moved
by the average occupational wage change estimated from
similar establishment/occupations. Establishment/occu
pations are considered similar if the establishments are
in the same 2-digit sic and the occupations are in the
same MOG. If there are not sufficient data at this level,
a broader level of aggregation is used. Prior-quarter data
are not adjusted when nonresponse is the result of
seasonal closing of an establishment.
Imputations are also made to fill in any gaps in a
respondent’s benefit data. Imputation for benefits is done
separately for each benefit both in the base period and
on a quarterly basis. A benefit cost is imputed based on
the average cost for the same benefit in similar establish
ment/occupations .
Index Computation
The basic computational framework is the standard
formula for an index number with fixed weights, as
modified by the special statistical conditions that apply
to the ECi.8 This discussion focuses on the ECI measure
of wage changes, but indexes of compensation changes
are calculated in essentially the same fashion.
An index for the eci is simply a weighted average of
the cumulative average wage changes within each
establishment cell, with base-period wage bills as the fixed
weights. The simplified formula is:
where:
Mt,i = Mt-l,i* Rt,i> and
It is the symbol for the index.
The other variables are defined as follows:
W0 j
Adjustments for sample nonresponse
When base-period data collection is completed,
nonresponse adjustment factors are calculated for per
manent refusals and applied to the sample weights of
responding establishment/occupations in the same major
industry division, MOG, and size class. The application
of the nonresponse adjustment factors compensates for
the loss of data due to base-period refusals only. Because
Mt j
Rt j
is the estimated base-period wage bill for the ith cell.
A cell generally is an occupation in a 2-digit SIC
industry, while the wage bill is the average wage of
workers in the cell times the number of workers
represented by the cell.
is the cumulative average wage change in the ith cell
from time o (base period) to time t (current quarter).
is the ratio of the current-quarter weighted average
wage in the cell to the prior-quarter weighted average
8
The actual ECI computational formulas and procedures differ
somewhat from those presented here, which have been simplified for
illustrative purposes.
58
wage in the cell, both calculated in the current quarter
using matched establishment/occupation wage quo
tations. The weights applied are the sample weights
described in the next section.
All wage indexes are computed from the following data:
Average straight-time hourly earnings for 3-digit
Census code occupations, or groups of those occupations,
in those sample establishments for which data are
available for both the current and prior survey periods.
The occupational wage data are identified by major
occupational group, industry, geographic location,
metropolitan area, and union status.
Employment, in 1980, in the 3-digit Census code
occupation or group of occupations in an industry,
obtained from the decennial census.
Sample weights derived from an occupational employ
ment survey or the initial employment reported on the
survey schedule. These weights reflect both employment
in each establishment/occupation surveyed and the prob
ability of selection of that establishment/occupation.
The index computation involves essentially five steps:
1.
2.
Establishment/occupation sample weights are
applied to the occupational earnings to obtain
weighted average earnings for each estimation cell
for the current and prior survey periods. The estima
tion cell is defined on the basis of owner/industry/
occupation. For the private sector, 67 SIC industries
have been identified, most at the 2-digit level. For
the public sector, separate cells are identified for
State and local governments. Industries as broad as
“ public administration” and as narrow as “ colleges
and universities” are treated as separate estimation
cell industries. For example, one estimation cell is
identified as State govern m en t/p u b lic administration/clerical workers.
Each quarter, the ratio of the current-quarter
weighted average wage to the prior-quarter weighted
average wage is, in effect, multiplied by the prior-
quarter cumulative average wage change for the cell.
The product is a measure of the cumulative percent
age wage change in the cell since the base period.
3. This measure of cumulative percentage wage change
is multiplied by the base-period wage bill to generate
an estimate of the current-quarter wage bill for the
cell.
4. Both the current-quarter and the base-period wage
bills are then summed over all cells within the scope
of the index.
5. The summed current-quarter wage bill is divided by
the summed base-period wage bill. The result, when
multiplied by 100, is the current-quarter index. That
index is divided by the prior-quarter index to pro
vide a measure of quarter-to-quarter change, the
link relative.
The following example illustrates the procedures for
a particular industry:
The computations for the occupation and industry
groups follow the same procedures as those for all overall
indexes except for the summation. The wage bills for the
occupational groups are summed across industries and
regions for each group; the wage bills for the industry
division are summed across occupational groups and
regions for each industry division.
Computational procedures for the regional, union/
nonunion, and metropolitan/nonmetropolitan measures
of change differ from those of the national indexes
because the current sample is not large enough to hold
constant the wage bills at that level of detail. For these
nonnational series, each quarter, the prevailing distribu
tion in the sample between, for example, union and non
union within each industry/occupation cell, is used to
apportion the prior-quarter wage bill in that cell between
the union and nonunion series. The portion of the wage
bill assigned to the union sector is then moved by the
percentage change in union wages in the cell, and similarly
for the nonunion sector. Thus, the relative employment
of the union sector in each cell is not held constant over
time. Since the relative weights of the region, the union,
and the metropolitan area subcells are allowed to vary
over time, it is not possible to calculate Laspeyres indexes
for the nonnational series.
(d)
Currentquarter
cumulative
change
(a x d)
(e)
Baseperiod
wage
bill
(0
Currentquarter
wage
bill
(f x e )
(9)
Priorquarter
wage
bill
(f xa )
<h)
$5.25
1.04762
1.29451
$12,613.40
$16,328.17
$15,586.00
7.15
1.00699
1.16242
8,316.37
9,667.11
9,600.00
20,929.77
25,995.28
25,186.00
Priorquarter
cumulative
change
(a)
Currentquarter
weighted
average
earnings
(b)
Priorquarter
weighted
average
earnings
(c)
Relative
(b/c)
Electricians
1.23567
$5.50
Carpenters
1.15435
7.20
Occupation
Total
59
The data are published later in Current Wage Develop
ments and the Monthly Labor Review, monthly bls
periodicals. The data are also available on ib m compatible microcomputer diskettes.
Variance computation
Work on variance estimation has been under way since
1984. Release of the variance estimates is scheduled for
1988. Variance estimates are computed by a method
called “ balanced repeated replication.” Starting with the
replacement samples in 1981, all 2-digit SIC samples have
been divided into a number of variance strata, and the
sample in each variance stratum has been divided into
two half-samples, sometimes referred to as panels. The
computations made for quarterly and annual relatives are
replicated 64 times using the data from one half-sample
from each stratum instead of the data from both half
samples. The variance is then given by:
64
VAR (Rs>t,o)
=
^ (Rs,t,o ~ Rs,t,i)2 /64
i= 1
where:
Rs t 0 is a quarterly or annual relative for some cell
from time s to time t calculated using the full
sample, and
Rs t j is the quarterly or annual relative for the same
cell from time s to time t calculated using ith
balanced half-sample.
The e c i half-sample estimates are approximately nor
mally distributed. Therefore, approximate 95-percent
confidence intervals can be obtained by adding or sub
tracting twice the square root of the variance estimates
to or from the estimate of the relative.
Presentation
e c i statistics are published quarterly in the month after
the survey period. For example, statistics computed from
the survey data for June are published in July. Initially,
the statistics are presented in a news release which includes
descriptions of quarter-to-quarter and year-to-year
trends, tables, and an explanatory note about the survey.
Uses and Limitations
The Employment Cost Index has been designated as
a principal Federal economic indicator by the Office of
Management and Budget. It is the only measure of labor
costs that treats wages and salaries and total compensa
tion consistently, and provides consistent subseries by
occupation and industry. Special wage and salary indexes
are also provided for union status, geographic region,
and metropolitan area status. The ECI is used by the
Federal Reserve Board in monitoring the effects of
monetary and fiscal policies and in formulating those
policies. It enables analysts and policymakers to assess
the impact of labor cost changes on the economy, both
in the aggregate and by sectors. The ECI is particularly
important in studies of the relationships between prices,
productivity, labor costs, and employment. It is also used
as an escalator of labor costs. For example, the Federal
Health Care Financing Administration uses the e c i as
part of an input price index in determining allowable in
creases in hospital charges under Medicare’s Inpatient
Hospital Prospective Payment System.
The limitations of the index must be kept in mind. The
index is not a measure of change in the total cost of em
ploying labor. Not all labor costs (e.g., training expenses,
retroactive pay, etc.) fall under the ECI definition of
compensation. Currently, the ECI does not cover all em
ployers and employees, although it does cover nearly all
workers in the civilian (non-Federal) nonfarm economy.
Finally, the index is not an exact measure of wage or com
pensation change. It is subject to sampling errors which
may cause it to deviate from the results which would be
obtained if the actual records of all establishments could
be used in the index calculation.
Technical References
Nathan, Felicia. “Analyzing Employers’ Costs for Wages, Sal
aries, and Benefits, ” Monthly Labor Review, October 1987.
U.S. Department of Commerce, Bureau of the Census. Clas
sified Index o f Industries and Occupations, 1980 Census
o f Population, 1980.
Schwenk, Albert E. “Introducing New Weights for the Emloyment Cost Index, ” M onthly Labor Review, June 1985.
Sheifer, Victor J. “Employment Cost Index: A Measure of
Change in the ‘Price of Labor’, ” Monthly Labor Review,
July 1975.
Sheifer, Victor J. “How Benefits Will Be Incorporated into the
Employment Cost Index, ” M onthly Labor Review,
January 1978.
U.S. Department of Labor, Bureau of Labor Statistics. Em
ployment Cost Index Occupation Classification System
M anual-1980, 1985.
Wood, G. Donald. “ Estimation Procedures for the Em
ployment Cost Index,” Monthly Labor Review, May
1982.
60
Bureau of Labor Statistics
ECI Wage Data Form
E s t a b lis h m e n t N a m e
P age____
S c h e d u le N u m b e r
L in e
No.
BLS
I d e n t i f i c a t i o n o f S u r v e y O c c u p a t io n s , E s t a b lis h m e n t J o b s , o r I n d iv id u a ls
O cc.
f o r w h o m W a g e I n f o r m a t io n is b e in g r e p o r t e d o n e a c h lin e
. 19
Code
H o u r ly
R a te
H o u r s a n d E a r n in g s
Q
R
(1 )
o f____
R e fe re n c e D a te
........... -...
- ..........-
(2)
<:1)
Num ber of
W o rke rs
P e r L in e
(4 )
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
—
* Please use the back page o f th is fo rm to explain sign ifica nt earnings changes (i.e., decreases o r large increases in the average rate o f pay fo r an o ccup a tio n) fro m one re p o rtin g perio d to the next.
BLS 3038B
(R ev. J u n e 1 9 8 4 )
Page 2
V A C A T IO N
Plan
1.
Company Plan Identification
E lig ibility Requirement
Most Recent Change
D e s c r ip t io n
O c c u p a t io n a l D is t r ib u t io n
Occu. coc e
Len g th
Weeks □
of
or
Service
Percent □
W krs. W t.
W krs. W t.
Wks.
W krs. W t.
Wks.
Wks.
W krs. W t.
W krs. W t.
Wks.
W krs. W t.
Wks.
W krs. W t.
Wks.
W krs. W t.
Wks.
W krs. W t.
Wks.
W krs. W t.
Wks.
Wks.
T otals
Average
B&8888888sj
r
gi?5*ip:s
i
vacation weeks
i
;
W
'
;.v
.
\
.
.... .L.
2 . D a ta E n t r ie s
(1-9)
C ontrol Inform ation
Occupational
Code
Benefit
Code
(10-13)
(14-15) (16-18)
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
02
02
02
02
02
02
02
02
02
02
Status Code
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
Value Entry
Conversion
Code
Average Vacation Weeks
(19-27)
(28-29)
(30-35)
I I I I 'ill
I
I I I I -III
I
l I I l 'ill
I
till -III
I
I
I I I I [ill
I I I I • I I I
I
I I I I -III
I
I
I I I I ['ll
I I I I 'ill
I
I I I I
_' l l ! _I
__
j
I ■ I I I
I
I I I
t i l l
I [ I I I
I [ I I I
I ' I I I
I
I I I
I
I II
I
I I I
_L j _i
_^_
07
E n d o f c a rd *
62
80
Chapter 9. The Employee
Benefits Survey
Background
By 1940, many employees received paid vacations, but
relatively few—especially blue-collar workers—received
other employee benefits, such as paid holidays or
employer-provided protection against the financial con
sequences of accident, sickness, death, and old age. The
adoption of pension and welfare plans was encouraged
through favorable tax treatment offered by the Federal
Government as early as 1921. However, the phenomenal
increase in plans after 1940 resulted chiefly from three
factors: (1) wage controls during World War II and the
early postwar period that permitted supplementary
benefit improvements while denying wage increases, (2)
National Labor Relations Board decisions bringing pen
sions and other benefits within the scope of compulsory
collective bargaining, and (3) the 1949 report of the Steel
Industry Fact Finding Board which maintained that
industry had an obligation to provide workers with social
insurance and pensions. Before 1940, key developments
in the growth of benefit plans occurred in nonunion
environments; after the outbreak of World War II, many
initiatives in employee benefits emerged through collec
tive bargaining.
As early as the 1920’s, the Bureau reviewed employee
benefit plans in its analyses of collective bargaining
agreements and trade union activities.1 By the mid1940’s, occupational wage studies yielded data on the
incidence and provisions of paid vacation and sick leave
plans and the incidence of insurance and pension plans
for plant and office workers.2
Analysis of employee benefit plans continued to
expand. One group of studies emphasized provisions of
individual plans. Based on small samples, these analyses
were designed to provide information about the particular
benefit plans studied rather than to report on the overall
incidence of plans or plan provisions.3 For example,
1 Trade A greem ents in 1923 and 1924, Bulletin 393 (Bureau of Labor
Statistics, 1925); and B eneficial A c tiv ities o f A m erican Trade Unions,
Bulletin 465 (Bureau of Labor Statistics, 1928). Some detail on con
tract provisions for supplementary benefits is contained in Union A gree
m en t P rovisio n s, Bulletin 686 (Bureau of Labor Statistics, 1942).
2 Wage Structure o f the M achinery Industries, January 1945, Series
2, No. 1 (Bureau of Labor Statistics, 1946).
3 H ealth-B enefit P rogram s E stablished Through C ollective Bargain
ing, Bulletin 841 (Bureau of Labor Statistics, 1945).
major provisions of selected health, insurance, and pen
sion plans were summarized periodically and published
in Digest o f Selected Health and Insurance Plans and
Digest o f Selected Pension Plans every 3 or 4 years until
1978, when the program was discontinued.
After the universe of welfare and pension plans became
known (as a result of the Welfare and Pension Plans
Disclosure Act of 1958),4 studies were made based on
samples representative of all plans filed under the act.5
In 1959, the Bureau initiated a series of surveys of
employer expenditures for employee compensation. This
program, which continued until 1977, measured outlays
for individual elements of compensation, including pay
for leave and contributions to private and public welfare
and retirement plans.6 The occupational wage studies,
which include limited information on benefit plan pro
visions, have added incidence of such benefits as dental
insurance and health maintenance organization coverage
in recent years.7
The most recent development in the Bureau’s analysis
of employee benefit plans occurred in the late 1970’s at
the request of the U.S. Civil Service Commission (now
the Office of Personnel Management). The Federal Salary
Reform Act of 1962 and its successor, the Federal Pay
Comparability Act of 1970, provided for annual adjust
ments in salaries of Federal white-collar employees to
achieve comparability with pay rates in private enterprise
for the same levels of work. The Bureau’s National
Survey of Professional, Administrative, Technical, and
Clerical Pay ( p a t c ) provides the data on private industry
salaries used in administering this legislation.
The rapid growth of employee benefits has raised ques
tions about the validity of a comparability process limited
4 In accordance with the act, administrators of welfare and pension
plans, excluding those for government workers and employees of non
profit organizations, having at least 26 participants, filed with the
Department of Labor detailed descriptions of their plans, including all
amendments. Administrators of plans having at least 100 participants
also had to file annual statistical reports on the financial status of their
plans. In general, similar filings are now required under the Employee
Retirement Income Security Act of 1974 (ERISA), which replaced the
Welfare and Pension Plans Disclosure Act.
5 E arly R etirem en t P rovision s o f Pension Plans, 1971, Report 429
(Bureau of Labor Statistics, 1974).
6 E m ployee C om pen sation in the P rivate N on farm E conom y, 1977,
Summary 80-5 (Bureau of Labor Statistics, 1980).
7 For a description of the occupational wage studies, see chapter 6.
63
to wages and salaries alone. In the 1970’s, the General
Accounting Office and two Presidential review groups
recommended that the Federal pay comparability system
be expanded to include both pay and benefits.
In response to these recommendations, the Office of
Personnel Management ( o p m ) initiated its Total Com
pensation Comparability ( t c c ) project. Computer
models were developed which determined the annual cost
per employee to the Federal Government—given the
characteristics of its work force—if it were to adopt the
various benefit plans in private industry. These costs
could then be compared with costs for Federal plans, as
determined by the same computer models.
Because of the Bureau’s long experience in studying
employee benefits, o p m asked the Bureau to participate
in the gathering of data on plan provisions and charac
teristics. The Employee Benefits Survey (EBS) was then
developed. In 1979, a test survey was conducted in con
junction with the patc survey. A full-scale survey in
medium and large private sector firms was conducted
annually from 1980 to 1986. In 1987, the survey examined
employee benefit plans in State and local governments.
Although OPM no longer uses the cost estimating models
developed for the TCC project, survey findings are
examined by o p m , other government agencies, Congress,
and the private sector as a key source of information on
the provisions of employee benefit plans.
Description of the Survey
Private sector surveys conducted between 1979 and 1986
covered establishments8 in the United States, excluding
Alaska and Hawaii, employing at least 50, 100, or 250
workers, depending on the industry. Industrial coverage
included: Mining; construction; manufacturing; transpor
tation, communications, electric, gas, and sanitary serv
ices; wholesale trade; retail trade; finance, insurance, and
real estate; and selected services. The 1987 public sector
survey covered State and local governments with 50 or
more employees in the contiguous 48 States. (The 1988
private sector survey will cover establishments with 100
or more employees in all nonfarm industries.)
Excluded from the survey are executive management
employees (defined as those whose decisions have direct
and substantial effects on an organization’s policymak
ing); part-time, temporary, and seasonal employees; and
operating employees in continuous travel status, such as
airline flight crews and long-distance truckdrivers.
8 For this survey, an establishment is an economic unit(s) which pro
duces goods or services, a central administrative office, or an auxiliary
unit providing support services to a company. In manufacturing
industries, the establishment is usually a single physical location. In non
manufacturing industries, all locations of an individual company within
a Metropolitan Statistical Area (MSA), a Primary Metropolitan
Statistical Area (PMSA), or a nonmetropolitan county are usually con
sidered a single establishment.
Sampled establishments are requested to provide data
on work schedules and details of plans financed wholly
or partly by employers in each of the following benefit
areas: Flexible benefit plans and reimbursement ac
counts; paid lunch periods; paid rest periods; paid
holidays; paid vacations; personal leave; funeral leave;
military leave; jury-duty leave; sick leave; sickness and
accident insurance; long-term disability insurance; health
insurance; life insurance; defined benefit pension plans;
and defined contribution retirement and capital
accumulation plans.
b l s also collects limited data on the incidence of
several additional benefits such as: Severance pay;
employee discounts; nonproduction bonuses; gifts;
relocation allowances; recreation facilities; subsidized
meals; educational assistance; parking; personal use of
company-owned car; in-house infirmaries; and child care.
Data Sources and Collection Methods
Data for the survey are collected primarily by visits of
Bureau field representatives to a sample of establishments
within the scope of the survey. To reduce the reporting
burden, respondents are asked to provide documents
describing their retirement plans, capital accumulation
plans, and plans covering the four insured benefit areas
surveyed (sickness and accident, long-term disability,
health, and life insurance). These documents are analyzed
by b l s staff in Washington to obtain the required data
on plan provisions. With a few exceptions, plans which
are fully employee paid are not reported. Data on paid
leave and other paid time off generally are obtained
directly from the employer at the time of the visit. Since
the survey does not develop information on the cost to
the employer of providing benefits, respondents are
seldom asked to provide cost data. However, employer
contribution rates are requested for certain collectively
bargained multiemployer health, welfare, and pension
plans where benefit amounts are tied directly to the
negotiated contribution level, such as in the construction
and trucking industries. Data are collected separately for
three occupational groups—professional-administrative,
technical-clerical, and production workers in the private
sector, and regular employees, teachers, and police and
firefighters in the public sector.
Information obtained from respondents and plan
documents is entered on computer files. Three data bases
are created—one for establishment control data, another
for paid leave plan provisions, and a third for retirement
and insurance plan provisions. The control data base con
tains information on the establishments surveyed, includ
ing: Number of employees, number of plan participants,
industry, geographic location, and sampling weight.
The plan data bases contain the provisions of each plan
for which information was obtained. Plan identification
64
codes are such that a plan, once analyzed, need not be
analyzed again regardless of how many establishments
report it (e.g., a companywide health insurance program
or a multiemployer pension plan).
Survey Design
The scope of the private sector survey is similar to that
of the Bureau’s p a t c survey. The list of establishments
from which the sample is selected (called the sampling
frame) is, therefore, related to that developed for the
p a t c . This sampling frame is developed by refining data
from the most recently available State unemployment
insurance (ui) reports for the 48 States covered by the
survey and the District of Columbia. The refinement pro
cedures include an effort to ensure that sampling frame
units correspond to the definition of an establishment
adopted for this survey.
The private sector sample for this survey primarily is
a subsample of the p a t c sample to reduce the costs and
resources required for data collection. The sample of
establishments is selected by first stratifying the sampling
frame by broad industry group and establishment size
group based on the total employment in the establishment.
The sample size is allocated to each stratum (defined
by industry and size) approximately proportional to the
total employment of all sampling frame establishments
in the stratum. Thus, a stratum which contains 1 percent
of the total employment within the scope of the survey
receives approximately 1 percent of the total sample. The
result of this allocation procedure is that each stratum
has a sampling fraction (the ratio of the number of units
in the sample to the number in the sampling frame) which
is proportionate to the average employment of the units
in the stratum.
Within each stratum, a random sample is selected using
a probability technique to maximize the probability of
retaining establishments which were selected in the
previous survey. This method of selection reduces col
lection costs by decreasing the number of new estab
lishments in the sample.
Similar procedures are used to select the 24 States and
850 local governments in the public sector sample. The
48 States within the scope of the survey are stratified into
the four Census Economic Regions, and a sample is then
selected with probability of inclusion being proportional
to employment. Local governments are selected in a threestage procedure. First, a sample of Metropolitan Statis
tical Areas and nonmetropolitan counties is selected.
Next, within each area or county selected, governmental
units are stratified into four broad industrial groups—
general government, schools, health services, and special
districts. Finally, within each of these industrial groups,
a sample of governments is selected, with probability pro
portional to employment.
Each of the combinations of occupational groups and
work schedule or benefit areas (e.g., health insurance for
production employees) is treated as an individual survey,
and separate estimates are developed for each. This treat
ment facilitates the use of partially completed establish
ment reports in the survey. Therefore, the actual number
of responses for the survey varies for each of the
combinations.
Two procedures are used to adjust for missing data
from partial reports and total refusals. First, imputations
are made for the number of plan participants or plan
details when the data are not reported. These imputations
are made by randomly selecting a similar plan from
another establishment in a similar industry and size class.
For other forms of missing data (or nonresponse), an
adjustment is made using a weight adjustment technique
based on sample unit employment. Establishments are
grouped together in cells similar to those used in sample
selection. Using the assumption that, on the average,
nonrespondents’ data would be similar to that reported
by respondents in each cell, the weight of each respond
ent is multiplied by a factor equal to the total employ
ment in the cell divided by the employment of the
responding units. The weight adjustments for missing
data used in this survey are calculated in four stages for
each occupational group and work schedule or benefit
area combination. This allows a maximum amount of
data from partially completed establishment schedules to
be incorporated into survey estimates.
The survey design uses an unbiased estimator (the
Horvitz-Thompson) which assigns the inverse of each
sample unit’s probability of selection as a weight to the
unit’s data. The estimator is modified to account for a
weight adjustment factor, fj, developed during the ad
justment for nonresponse. The estimator is:
Y =
n
f; Yi
E
-----i = 1 Pi
where:
n = number of responding units
fj = weight adjustment factor for the ith unit
Yj = value for the characteristic of the ith unit
Pj = the probability of including the ith unit in the
sample
Sampling and estimating procedures are designed to
yield national data for all studied industries combined.
Survey findings do not yield reliable estimates for
individual industries or geographic regions. Data are,
however, reported separately for three occupational
groups.
65
Presentation
Uses and Limitations
Summary survey results are published in a news release
in the spring following the survey year, and an annual
b l s bulletin9 includes major survey findings. Estimates
in these publications show the percent of employees that
are covered by paid leave plans; participate in insurance,
defined contribution, and pension plans; or are eligible
for other benefits. Counts of participants in benefit plans
include those who have not met possible minimum lengthof-service requirements at the time of the survey. Workers
are counted as participants in employee benefit plans that
require the employee to pay part of the cost only if they
elect the plan and pay their share. With a few exceptions
(such as retiree health insurance), plans for which the
employee pays the full premium are outside the scope of
the survey, even if the employer pays administrative costs.
Tabulations show the percent of workers covered by
individual benefit plans or plan provisions. Percentages
are calculated in three ways. One technique shows the
number of covered workers as a percent of all workers
within the scope of the survey. It is designed to show the
incidence of the individual employee benefits.
The second approach shows the number of workers
covered by specific features in a benefit area as a percent
of all employees who participate in that general benefit
area. These tables answer questions concerning typical
coverages provided to persons with a given insurance
benefit or a private pension plan; for example, what per
cent of all employees with health insurance receive den
tal coverage?
The third approach provides a closeup look at an
important feature of the plan; for example, what percent
of all employees with dental coverage in their health
insurance are covered for orthodontic work?
The benefit plan data bases contain additional detailed
information for which estimates and tabulations are not
routinely developed by b l s . These data are used for indepth analyses of specific aspects of employee benefits
that are occasionally published in Monthly Labor Review
articles (see references).
Employee benefit data collected during the annual
survey, including detailed provisions of plans and the
number of participants, are available on magnetic
tapes.10 In accordance with a pledge of confidentiality to
survey respondents, all information that could identify
a specific reporting establishment is removed. The tapes
may be used to derive national estimates, similar to those
presented in the bulletin, for those provisions in the data
base that are not regularly tabulated by BLS.
The extensive body of information on employee
benefits generated in this survey provides a unique data
resource. It is a major source of information for labor
and management representatives involved in contract
negotiations. Employers frequently seek information
permitting comparison of their benefit plans with pre
vailing practices. Labor unions also use benefit data to
assess potential areas for increasing nonwage compensa
tion. Other users of the data are State and Federal con
ciliators and mediators, public and private arbitrators,
Members of Congress and congressional staff consider
ing legislation affecting the welfare of workers, and
government officials responsible for recommending legis
lation and reviewing proposed legislation. For example,
Congress and the Administration need data to evaluate
the revenue implications of the favorable tax treatment
accorded many types of benefits. Also, social welfare
planners use data on private benefit plans to assess the
ability of employees to provide for the current and future
health and welfare needs of themselves and their
dependents.
In addition, b l s tabulations and analyses of employee
benefits can also be of use to teachers, students, and
others in the academic field; private consultants; research
ers; writers; and those not directly involved in legislation
or collective bargaining but concerned with the develop
ment, status, and trends in employee benefits.
A limitation of the data is that, since data collection
is restricted to provisions of formal plans, the extent of
such benefits as rest periods and personal leave may be
understated. Furthermore, the data show the coverage
of benefit plans but not the actual use of these benefits;
for example, that part of paid sick leave actually taken.
Users of the e b s data should keep in mind that the
survey does not measure employer costs for benefits. In
addition, the current scope of the survey excludes small
firms. Studies of employee benefits that include all firms
typically report lower participation rates for most
benefits. Also, reliable estimates can be produced only
at the national level, with no geographic or industry
detail. Data for selected industries are available from
industry wage surveys and, for selected geographic
regions, from area wage surveys. These occupational
wage surveys, which include some small firms, provide
data on paid holiday and vacation practices and the
incidence of welfare and pension plans, but not detailed
provisions of the benefits (see chapter 6).
Since data gathered in Bureau surveys are confiden
tial, specific plan provisions cannot be published.11
9 The most recent bulletin is E m p lo yee B en efits in M edium and
L arge Firm s, 1986, Bulletin 2281, June 1987.
10 The tapes may be purchased from the Office of Wages and
Industrial Relations, Bureau of Labor Statistics, Washington, DC
20212. They are available in 6250 BPI. Lists of data items on the com
puter files are available upon request.
11
The D igest o f Selected H ealth an d Insurance P lans and the D igest
o f S elected Pension P lans, mentioned earlier, identified the plans
analyzed. The information, however, was obtained with the plan spon
sor’s consent for the specific purpose of the publications.
66
Chart 1. Employee Benefits Survey generalized standard errors
Standard error
However, details of benefits provided under large
negotiated agreements are published in Current Wage
Developments (see chapter 7) and are available in the
Bureau’s public-use file of negotiated contracts. Besides
small firms, the survey also excludes executive manage
ment and operating employees in constant travel status
(such as airline pilots), as well as part-time, temporary,
and seasonal employees. Alaska and Hawaii are not
surveyed; neither are the Federal Government, agri
culture, and private households. The data, therefore, do
not statistically represent all employees in the United
States, or even all employees in private industry. Never
theless, the survey provides the most extensive informa
tion available on the provisions of employee benefits.
The EBS is designed to yield estimates of the percent of
employees with specific benefit provisions in the survey
year, not the change in plan provisions over time. Some
plan provisions are found mainly in one or two industries.
When employment changes do not occur evenly across
industries, shifts in survey findings regarding relative inci
dence of types of benefit plans may stem, not from changes
in plans, but from disproportionate changes in the number
of employees covered by different types of plans.
Reliability of Estimates
The statistics in the ebs bulletins are estimates from
samples rather than tabulations based on all units within
scope of the surveys. Consequently, the data are subject
to sampling errors, as well as nonsampling errors.
Sampling errors are the differences that can arise be
tween results derived from a sample and those computed
from observations of all units in the population being
studied. When probability techniques are used to select
a sample, as in the Employee Benefits Survey, statistical
measures called “ standard errors” can be calculated to
measure possible sampling errors.
This evaluation of survey results involves the forma
tion of confidence intervals that can be interpreted in the
following manner: Assume that repeated random samples
of the same size were drawn from a given population and
an estimate of some value, such as a mean or percentage,
was made from each sample. Then, the intervals described
by one standard error below each sample’s estimate and
one standard error above would include the population’s
value for 68 percent of the samples. Confidence rises to
90 percent if the intervals surrounding the sample esti
mates are widened to plus and minus 1.6 standard errors,
and to 95 percent if the intervals are increased to plus
and minus 2 standard errors.
Chart 1 provides standard errors for use in evaluating
the private sector estimates in tables of the bulletins con
taining percentage estimates. Standard errors for tables
containing nonpercentage estimates could not be general
ized into graphic representation. They are presented as
tables of standard errors in the bulletins, starting in 1986.
Nonsampling errors also affect survey results. They can
be attributed to many sources: Inability to obtain infor
mation about all establishments in the sample; defini
tional difficulties; differences in the interpretation of
67
questions; inability or unwillingness of respondents to
provide correct information; mistakes in recording or
coding the data; and other errors of collection, response,
processing, coverage, and estimation for missing data.
Through the use of computer edits of the data and pro
fessional review of both individual and summarized data,
efforts are made to reduce the nonsampling errors in
recording, coding, and processing the data. However, to
the extent that the characteristics of nonrespondents are
not the same as those of respondents, nonsampling errors
are introduced in the development of estimates. Because
the impact of these limitations on the EBS estimates is
unknown, reliability measurements are incomplete.
For those readers interested in further mathematical
details, the next section describes how chart 1 was derived
from 1982 survey data.
Mathematical details on estimates and generalized
standard errors chart
Each estimator used in the production of the tables in
the bulletins is approximately normally distributed.
Standard errors for the percentage estimates were
originally computed from the 1982 survey data, using a
random group method. To simplify their presentation,
a curve was fitted to the standard error estimates, by
regression techniques.
The curve’s equation is:
s __ e[a + b {ln(P)}2 + c (ln(lOO-P)}2 + d ln(P) ln(lOO-P)]
where:
S
P
e
In
=
=
=
=
standard error
percentage estimate from the bulletin
exponential function
natural logarithm function
For the 1982 Employee Benefits Survey,
a = -0.64683,
b = -0.02603,
c = -0.017458, and
d = 0.123726
These are regression coefficients. The curve fits the data
with R2 = 0.85 and no pattern in the residuals.
Moreover, differences between curves using 1982 and
1983 data are negligible.
The equation of the curve was obtained empirically, by
starting with the equation:
S = a Pb (100 - P)c.
Technical References
Employee b enefits— general
Insurance plans
Bureau of Labor Statistics. Employee Benefits in Medium
and Large Firms, 1986, Bulletin 2281, June 1987.
Bell, Donald R. “Dental and Vision Care Benefits in Health
Insurance Plans, ” M onthly Labor Review, June 1980.
Frumkin, Robert N., and Wiatrowski, William J. “Bureau of
Labor Statistics Thkes a New Look at Employee Benefits, ”
M onthly Labor Review, August 1982.
Blostin, Allan P. “Is Employer-sponsored Life Insurance
Declining Relative to Other Benefits?” M onthly Labor
Review, September 1981.
Pension plans
Blostin, Allan P. “Mental Health Benefits Financed by Em
ployers, ” Monthly Labor Review, July 1987.
Bell, Donald R., and Graham, Avy. “Surviving Spouse’s Bene
fits in Private Pension Plans, ” M onthly Labor Review,
April 1984.
Blostin, Allan P., and Marclay, William, “ m m o s and Other
Health Plans: Coverage and Employee Premiums,”
Monthly Labor Review, June 1983.
Bell, Donald R., and Hill, Diane. “How Social Security Pay
ments Affect Private Pensions, ” M onthly Labor Review,
May 1984.
Frumkin, Robert N. “Health Insurance Tfends in Cost Control
and Coverage, ” Monthly Labor Review, September 1986.
Bell, Donald R., and Wiatrowski, William J. “Disability Bene
fits for Employees in Private Pension Plans, ” Monthly
Labor Review, August 1982.
Bell, Donald R., and Marclay, William. “Trends in Retirement
Eligibility and Pension Benefits, 1974-1983,” Monthly
Labor Review, April 1987.
Schmitt, Donald G. “Today’s Pension Plans: How Much
Do They Pay?” Monthly Labor Review, December 1985.
Schmitt, Donald G. “Postretirement Increases Under Private
Pension Plans, ” Monthly Labor Review, September 1984.
Hedger, Douglas, and Schmitt, Donald G. “Tfends in Major
Medical Coverage During a Period of Rising C osts,”
Monthly Labor Review, July 1983.
Hill, Diane B. “Employer-sponsored Long-term Disability
Insurance, ” Monthly Labor Review, July 1987.
Miller, Michael A. “Age-related Reductions in Workers’ Life
Insurance, ” Monthly Labor Review, September 1985.
Wiatrowski, William J. “Employee Income Protection Against
Short-term Disabilities, ” Monthly Labor Review, February
1985.
68
Chapter 10. Productivity
Measures: Business Sector
and Major Subsectors
Background
Indexes of labor productivity, multifactor productivity,
and related measures for broad economic sectors and
manufacturing industries are published by the Bureau of
Labor Statistics. Measures of output per hour have been
developed for the business sector, and nonfarm and farm
subsectors, from 1909 to the present. For the period after
1947, these data have been supplemented with comparable
measures of compensation and costs and corresponding
series for manufacturing (total, durable, and nondurable)
and nonfinancial corporations. For the latter period,
indexes are available quarterly as well as annually (table
1). These productivity measures, first published in 1959,
represent the culmination of a long series of developments
in productivity measurement in the Bureau.1
The multifactor productivity indexes for major sectors
measure output per combined unit of labor and capital
input in private business, private nonfarm business, and
manufacturing. Multifactor productivity indexes for 20
manufacturing industries measure output per combined
unit of capital, labor, energy, materials, and purchased
services inputs—KLEMS inputs.
Productivity measures, related inputs, and how often
the indexes are available are shown in table 1 for selected
sectors of the U.S. economy.
Description of Measures
The Bureau’s productivity measures are constructed as
the ratio of real gross product originating ( g p o ) in a sec
tor to the corresponding inputs engaged in the sector. The
changes through time in these ratios reflect changes in
output per unit of the corresponding inputs. The changes
in productivity and related measures during the course
of the business cycle typically show patterns which dif
fer substantially from long-term movements and, there
fore, are the objects of special analytic studies.
The Bureau’s multifactor productivity measures
supplement the labor productivity series and provide
1 Trends in O u tpu t p e r M an-H our in the P rivate E conom y, 1909-58,
Bulletin 1249, 1959 (Bureau of Labor Statistics, 1959).
Table 1. Availability of productivity measures for major
sectors and subsectors of the economy
P r o d u c t iv it y m e a s u r e
In p u t
In d e x
a v a ila b le
Output per hour of all persons:
Business1 ...........................................
Nonfarm business ........................
Nonfinancial corporations___
Manufacturing ......................
Durable ..............................
Nondurable ........................
Labor
Labor
Labor
Labor
Labor
Labor
Quarterly
Quarterly
Quarterly
Quarterly
Quarterly
Quarterly
Multifactor productivity:
Private business ................................ Labor, capital
Private nonfarm business............ Labor, capital
Manufacturing .......................... Labor, capital
Annually
Annually
Annually
KLEMS2 multifactor productivity:
Manufacturing and 20 2-digit SIC
manufacturing industries ............ Labor, capital,
energy,
materials,
services
Annually
1 Includes government enterprises; multifactor productivity measures exclude
such enterprises.
2Capital (K), labor (L), energy (E), materials (M), and purchased services (S)
inputs.
additional insights into labor productivity growth and
economic change.2 The GPO-based multifactor produc
tivity measures for the private business, private nonfarm
business, and manufacturing sectors are based on capital
and labor inputs. Measures for 2-digit Standard Industrial
Classification (SIC) manufacturing industries are based
on the real value of production and capital, labor, energy,
materials, and services inputs. For all these multifactor
productivity measures, labor input is the same as in the
labor productivity measures.
No single productivity ratio can be regarded as best for
all purposes. The b l s approach makes available a
number of alternative measures together with detailed
descriptions of the methods used.
2
J. R. Norsworthy, Michael Harper, and Kent Kunze, “The
Slowdown in Productivity Growth: Analysis of Some Contributing Fac
tors,” B rookin gs P apers on E con om ic A c tiv ity , Fall, 1979.
69
rather than persons are counted, so that multiple job
holders are counted more than once. Weekly hours are
measured as hours paid rather than hours at work. A
survey has been introduced to develop a set of labor input
measures based on hours at work and will be used to
extend the present series.4 (See b ls form. 2000P at end
of this chapter.)
The ces data are based on payroll records from a sam
ple of establishments in which the probability of sample
selection is related to the establishment size: Large
establishments (relative to the sector) fall into the sam
ple with certainty, whereas smaller establishments are
sampled on a probability basis. Data on employment,
hours, and earnings are collected monthly; the reference
period for these data is the payroll period including the
12th of the month. (The c es methods are described in
chapter 2.) Establishment data are published monthly in
Data Sources and Estimating Procedures
Output Per Hour Measures
Output
Real gross domestic product originating in the business
sector and subsectors is the basis of the output com
ponents of the major sector labor productivity measures.
These output components are based on and are consist
ent with the National Income and Product Accounts
( n i p a ), including the gross national product (GNP)
accounts, prepared by the Bureau of Economic Analysis
( b e a ) of the U.S. Department of Commerce.
Computation of business sector GPO begins with GNP,
which equals income received by labor and property
owners for services rendered in the current production
of goods and services, in addition to capital consump
tion allowances, indirect business taxes, and several other
minor items. Gross domestic product (g d p ) is simply g n p
less “ rest-of-world” output, that is, less net factor pay
ments to domestic owners of factors of production
located outside the United States.
Business sector output is equal to g d p less general
government, output of nonprofit institutions, output
of paid employees of private households, rental value
of owner-occupied dwellings, and the statistical
discrepancy.
Gross domestic product in current dollars cannot be
used directly as the output measure because it reflects
price changes as well as changes in physical volume. The
Bureau of Economic Analysis prepares measures of
constant-dollar g d p for the business economy and its
major sectors. These measures exclude changes in the
value of production resulting from price changes. They
reflect only changes in real product, which is the basis
for productivity measures.3
Output data for the manufacturing sector based on
gross product are computed by the b e a on an annual
basis only. In order to achieve quarterly estimates of
manufacturing output consistent with the b e a ’s gross
product concept, b l s uses the quarterly changes in the
Federal Reserve Board index of manufacturing produc
tion to move the gross product data. The results are
benchmarked annually to the published b e a output
levels. Thus, the output data used for all major sectors
are consistent with the output concepts embodied in the
National Income and Product Accounts.
Labor input
The primary source of hours and employment data is
the b l s Current Employment Statistics (CES) program,
which provides monthly survey data on total employment
and average weekly hours of production and nonsupervisory workers in nonagricultural establishments. Jobs
Employment and Earnings.
Since c es data include only nonfarm wage and salary
workers, data from the Current Population Survey (CPS)
are used for farm employment; in the nonfarm sector,
the National Income and Product Accounts and the c p s
are used for government enterprise employment, pro
prietors, unpaid family workers, and paid employees of
private households.
Separate estimates for employment and hours paid are
developed for each major sector and are aggregated to
business and nonfarm business levels. Hours of labor
input are treated as homogeneous units; no distinction
is made among workers with different skill levels or wages.
For nonmanufacturing sectors, employment and
average weekly hours are computed from the CES, CPS,
and n ipa sources. Although ces data on average weekly
hours refer only to nonsupervisory workers, it is assumed
for hours computation that the length of the workweek
in each nonmanufacturing industry is the same for all
wage and salary workers.
In manufacturing, separate measures for production
and nonproduction workers’ hours are derived and
aggregated to the manufacturing total. Employment and
average weekly hours for production workers and employ
ment for nonproduction workers are taken directly from
CES data. Average weekly hours for nonproduction
workers are developed from bls studies o f wages and sup
plements in manufacturing which provide data on the
regularly scheduled workweek of white-collar employees.
3 A detailed description of the methods and procedures for
estimating GNP and GDP in current and constant dollars is given in
Carol S. Carson, “ GNP: An Overview of Sources Data and Estimating
Methods,” S u rvey o f C urrent Business, July 1987, pg. 103-26. Also
see Methodology Paper No. 1 “ Introduction to National Income
Accounting” (Bureau of Economic Analysis, 1985). Further informa
tion on estimates for major industry sectors is presented in the October
1962 issue of the Su rvey o f C urrent Business.
4 Kent Kunze, “A New BLS Survey Measures the Ratio of Hours
Worked to Hours Paid,” M o n th ly L a b o r R eview , June 1984, pp. 3-7.
70
Compensation and labor costs
Indexes of compensation per hour measure the hourly
cost to employers of wages and salaries, as well as sup
plemental payments, which include employers’ contribu
tions to Social Security, unemployment insurance taxes,
and payments for private health insurance and pension
plans. Measures of real compensation per hour reflect
the adjustment of hourly compensation for changes in
the Consumer Price Index for All Urban Consumers
(CPI-U).
Unit labor costs measure the cost of labor input re
quired to produce one unit of output and are derived by
dividing compensation in current dollars by output in
constant dollars. Unit nonlabor payments measure the
cost of nonlabor items such as depreciation, rent, interest,
and indirect business taxes, in addition to corporate
profits and profit-type income of proprietorships and
partnerships.
Multifactor Productivity Measures for Major
Sectors
The multifactor productivity ( m f p ) indexes for major
sectors measure output per combined unit of labor and
capital input in private business, private nonfarm busi
ness, and manufacturing.5 The output measure is iden
tical to the output used in the measures of output per hour
with the exception that the output of government enter
prises is eliminated from the m f p measures because of
inadequate information on capital used in these enter
prises. Labor input is also identical to that used in the
output per hour measures, with the same exception.
Capital inputs for the m f p measures are computed
using the service-flow concept, b l s measures of capital
services inputs are prepared using national accounts data
on real gross investment in depreciable assets and inven
tories. Capital stocks for 43 types of depreciable assets
are constructed using the perpetual inventory method and
by assuming that services decline as a function of age.
These “ age/efficiency” schedules, which are based in part
on empirical evidence on capital depreciation, are applied
to real investment by type of asset.
These measures of depreciable stocks are combined
with stock estimates for inventories and land using the
Tornquist aggregation method. Asset weights are based
on estimates of implicit rental prices which are indicative
of marginal products. Rental prices are estimated with
a user-cost-of-capital formula consisting of the real rate
of return plus the rate of economic depreciation adjusted
for tax incentives.
Total input is computed by weighting capital and labor
using the Tornquist weighting formula. For each input,
the weight is the income share of the input in total
income. The major sector m f p measures are available for
the years 1948 to the present.
Multifactor Productivity Measures for Manufac
turing Industries
Multifactor productivity indexes for 20 manufactur
ing industries measure output per unit of weighted and
combined capital, labor, energy, nonenergy materials,
and purchased business services inputs.6
For the manufacturing industries m f p measures, out
put is the deflated value of production of an industry;
hence it differs from the g p o output measures used for
the major sector output per hour and m f p indexes. The
value of production is shipments, adjusted for inventory
change, to purchasers outside the industry. Capital is
measured as it is for the major sector m f p indexes; ren
tal prices of capital are computed for each industry.
Labor is also measured as it is for major sector m f p .
The inclusion in the industry m f p measures of all
intermediate inputs—energy, nonenergy materials, and
purchased business services—is consistent with the use
of total value of production as the output measure.
Energy input is constructed using data on price and quan
tity of fuels purchased for use as heat or power.
Nonenergy materials input includes all commodity inputs
exclusive of fuels but inclusive of fuel-type inputs used
as raw materials in manufacturing. The purchased
business services input series is constructed using price
and value data on services purchased by manufacturing
industries from service industries. Data sources used in
constructing these three inputs include input-output
tables, surveys of establishments in manufacturing and
other industries, and price indexes.
Total input is computed by weighting all inputs using
the Tornquist formula. For each input, the weight is the
cost share of the input in total costs. The industry m f p
measures are available for 1949 to the present.
Analysis and Presentation
Indexes of labor productivity show changes in the ratio
of output to hours of labor input. Similarly, indexes of
multifactor productivity show changes in the ratio of out
put to combined inputs. However, these indexes should
not be interpreted as presenting the contribution of a
single input, or a combination of inputs, to production.
Changes over time in these indexes reflect many
influences, including variations in output (especially in
the short term since most inputs are partially fixed), the
utilization of capacity, changes in the characteristics
6
An explanation of the methods and some results are found in an
5
The measures themselves and the underlying methods and pro article by William Gullickson and Michael J. Harper, “Multifactor Pro
cedures are explained in Trends in M u ltifa cto r P ro d u ctivity, 1948-81,
ductivity in 20 U.S. Manufacturing Industries, 1949-83,” M o n th ly
Bulletin 2178 (Bureau of Labor Statistics, 1983).
L a b o r R eview , October 1987, pp. 18-28.
71
and efforts of the work force, changes in managerial skill,
and technological developments.
Indexes of output per hour, compensation per hour,
and related cost data are published quarterly in the bls
news release, “ Productivity and Costs.” In addition,
quarterly and annual analyses are published from time
to time in the Monthly Labor Review. Historical indexes
of these and related data are available on request, as are
detailed descriptions of data sources and computational
procedures.
Multifactor productivity measures are announced each
October for the preceding calendar year in the “ Multifac
tor Productivity Measures” news release. Included are
annual indexes of multifactor productivity and related
measures for private business, private nonfarm business,
and manufacturing.
Indexes of productivity and related cost data are
available monthly in Employment and Earnings and the
Monthly Labor Review, in the Handbook o f Labor
Statistics, on l a b st a t data tapes, and bls data diskettes.
Compensation and labor costs
b e a develops employee compensation data as part of
the national income accounts. These quarterly data
include direct payments to labor—wages and salaries
(including executive compensation), commissions, tips,
bonuses, and payments in kind representing income to
the recipients—and supplements to these direct payments.
Supplements consist of employer contributions to funds
for social insurance, private pension and health and
welfare plans, compensation for injuries, etc.
The compensation measures taken from establishment
payrolls refer exclusively to wage and salary workers.
Labor cost would be seriously understated by this
measure of employee compensation alone in sectors such
as farm and retail trade, where hours worked by propri
etors represent a substantial portion of total labor input.
b l s , therefore, imputes a compensation cost for labor
services of proprietors and includes the hours of unpaid
family workers in the hours of all employees engaged in
a sector. Labor compensation per hour for proprietors
is assumed to be the same as that of the average employee
in that sector.
Calculation Procedures
Labor productivity
Labor productivity, or output per hour, is computed
as:
Constant-dollar output
Labor productivity = “
777
:
Hours of labor input
Unit labor and nonlabor costs
The Bureau also prepares data on labor and nonlabor
costs per unit of output for the business sector and its
major components. Unit labor costs relate hourly com
pensation of all persons to output per hour and is de
fined as compensation per unit of constant-dollar out
put. Nonlabor payments are the excess of gross product
originating in an economic sector over corresponding
labor compensation, and include corporate profits and
the profit-type income of proprietors. Nonlabor costs
include interest, depreciation, rent, and indirect business
taxes.
In aggregate sectors, productivity changes through time
reflect movements within the various component
industries as well as shifts in the relative importance of
each of the industries. For example, changes in labor pro
ductivity and multifactor productivity are influenced by
the relative shift of inputs (labor and capital) from lowto high-productivity industries and by productivity
changes in the component subsectors.7
Short-term movements in productivity and unit labor
costs often result from cyclical variaton in output; this
tends to distort the long-term relationship between out
put and labor input, as noted below, or output and
multifactor input.
P = O /H
In instances where several sectors are involved, labor
productivity can be computed equivalently as:
P = (EjO jI/E-H j
or as
P = EiWi(Oi/H i)
where:
Oj
is constant-dollar output in sector i
H.
is hours of labor input in sector i
Wj = Hj/EjHj is the hours-based weighting factor for
sector i
P
is average labor productivity for the aggregate
sector
The computation of labor compensation per hour paral
lels the computation of output per hour. Unit labor costs
(ULC) are computed as labor compensation (c ) per unit
of (constant-dollar) output, but is often represented as:
ULC = (C/H) - (O/H)
7
The farm-nonfarm shift is examined in some detail by J. R. Norsworthy and L. J. Fulco in “ Productivity and Costs in the Private
Economy,” M o n th ly L a b o r R eview , June 1974, pp. 3-9.
This form highlights the relationships between unit labor
costs, hourly compensation, and labor productivity.
Real compensation per hour (RC) is computed as
hourly compensation deflated by the seasonally adjusted
Consumer Price Index for All Urban Consumers (CPi-U):
72
RC = (C/H) - CPI-U
Unit nonlabor payments ( u n l p ) include all nonlabor
components of gross product originating in a given
sector—depreciation, rent, interest, and indirect business
taxes as well as profits and profit-type income—whereas
unit nonlabor cost excludes profit. These measures are
computed as:
UNLP = (CU - C )/0
and
UNLC = (CU - C - P R )/0
where:
CU
C
O
PR
is
is
is
is
current-dollar gross product originating
current-dollar compensation
constant-dollar output
current-dollar profits
Multifactor productivity
The computational method used by b l s for its
multifactor productivity measures is the Tornquist index.
Some of the basic properties of this index are: It is
calculated as a weighted average of growth rates of the
components; the weights are allowed to vary for each time
period; and, for productivity measures, the weights are
defined as the mean of the relative compensation shares
of the components in two adjacent time periods. Hence,
the growth rate of the index (I/I) is the proportional
change over time (the dot notation refers to continuous
change with respect to time), such that:
i/i
=
E iW.tr x it/ x it)
Labor’s share in gross product originating in a given
sector is simply the ratio of labor compensation paid in
that sector to the gross product, both measured in cur
rent dollars:
where xit/x it is the growth rate of the ith input calculated
as:
LS = C/CU
The weights (wit) are defined as the means of the
relative compensation shares of all the inputs:
and, analogously, the nonlabor or capital share is defined
as:
V xk = ln xit - ln xu-i
wu = (sit + V i) /2
NLS = (CU - C)/CU = 1 - LS
Most of the measures noted above are prepared quart
erly in index form for the major sectors of the private
sector. In addition, quarterly percentage changes at a
compound annual rate and percentage changes from the
same quarter in the previous year are computed:8
Qt = 100 (Vt/V t_,)4 - 100
Yt = 100 (Vt/V t_4) - 100
s it
pit = price or wage of input
in period t.
Multifactor productivity growth is defined as the growth
rate in output (O/O) less the growth rate in aggregated
inputs:
MFP = 6 /0 - i/I
For m f p measures of output per combined unit of
labor and capital inputs, this formula is implemented as:
l/ l = wkK/K + Wj l / L
where:
i/i = wkk /K + w, L/ l
wk = relative compensation share of capital
In order to achieve greater precision, all computations
are made from the measures themselves rather than from
their corresponding indexes.
8 The estimation of quarterly (or subannual) changes at compound
annual rates as the differences between movements in the underlying
series involves approximations. For changes in the neighborhood of
1 or 2 percent, these approximations are good; however, the inexact
ness of these approximations is amplified by relatively large changes
in the economic measures such as those experienced during periods of
inflation, sharp recession, and rapid recovery.
Since most of the productivity and costs measures are reported as
percentages to one decimal place, e.g., 2.6 percent, questions some
times arise because the greater precision carried in the automated com
putation results in differences in related measures in the final decimal
place.
Pitx it
-----------------
E iPitx it
where:
t is a time subscript denoting the quarter
V is a series described above
Qt is the quarterly percentage change in series V from
quarter t-1 to quarter t, measured at a compound
annual rate
Yt is the percentage change in series V from quarter t-4
(the same quarter 1 year before) to quarter t
_
w, = relative compensation share of labor
K/K = growth in capital services
L/L = growth in hours
Uses and Limitations
Measures of output per hour, output per unit of capi
tal, and output per combined unit of multifactor input
(multifactor productivity) and related measures of costs
are designed for use in economic analysis and public and
private policy planning. The data are used in forecasting
and analysis of prices, wages, and technological change.
The labor productivity, multifactor productivity, and
related cost measures are useful in investigating the rela
tionships between productivity, wages, prices, profits,
73
and costs of production. As noted above, gross domestic
product represents the sum of all production costs: Labor
compensation, profits, depreciation, interest, rent, indi
rect business taxes, and other minor items. Unit labor
costs, or compensation per unit of output, represent a
major portion of total unit costs and reflect the combined
effect of changes in output per hour and compensation
per hour; thus, an increase in compensation per hour
tends to increase unit labor costs while an increase in out
put per hour tends to reduce it, other things being equal.
Therefore, through its impact on unit labor costs, out
put per hour is an important element in the wage-price
relationship because it is an indicator of the extent to
which compensation gains can occur without putting
pressure on prices or reducing payments to other input
factors.
Certain characteristics of the productivity and related
cost data should be recognized in order to apply them
appropriately to specific situations. First, the data for
aggregate sectors reflect changes within various constit
uent industries as well as shifts in the relative importance
of these industries: a portion of labor productivity growth
from 1947 to the present is attributable to the shift of
workers from farm to nonfarm occupations. Second, the
measures are often linked by lead or lag relationships,
particularly during the business cycle when inventories,
overtime hours, and the rate of capital utilization are used
to buffer the effects of short-term swings in product
demand. Third, data and other resources available for
their preparation somewhat limit the productivity, out
put, compensation, and employment measures which can
be constructed. In several sectors where output is difficult
to define in a satisfactory way, productivity measures are
correspondingly weak. Examples are the construction
industry and the financial services sector, where output
is an imputed value of labor and other inputs. The pro
ductivity and costs measures for these sectors should be
interpreted with caution.
Technical References
Bureau of Labor Statistics
Trends in Multifactor Productivity, 1948-81, Bulletin 2178,
1983.
Presents bls annual indexes of multifactor produc
tivity for private business, private nonfarm business, and
manufacturing for the period 1948 through 1981. Also
presents bls annual measures of output per unit of
capital services input for the three sectors.
Fulco, L. J. “ The Decline in Productivity During the
First Half of 1985,” M onthly Labor Review, December
1985.
A summary of movements in published major sectors—
business, nonfarm business, manufacturing, and nonfinancial corporations—during the first two quarters of
1985.
Productivity: A Selected Annotated Bibliography 1979-1982,
Bulletin 2212, 1984.
Over 1,400 references concerning productivity and pro
ductivity measurement. Each reference includes a brief
annotation.
Fulco, L. J. “ Productivity and Costs During 1984,”
Monthly Labor Review, June 1985.
Annual review article for 1984. Examines productivity
movements during the year and charts changes in pro
ductivity, hourly compensation, and unit labor costs from
1973 forward.
Dean, Edwin R., and Kunze, Kent. “ United States Multifactor
Productivity Growth, 1948-86,” Monthly Labor Review,
forthcoming.
Presents growth rates of multifactor productivity for
the periods 1948-73, 1973-79, and 1979-86 for private
business, nonfarm business, and manufacturing. Analyzes
multifactor measures for recent years and describes recent
data revisions and methodological improvements that
have been incorporated into these measures.
Gullickson, William, and Harper, Michael J. “ Multifactor
Productivity in 20 U.S. Manufacturing Industries, 194983,” Monthly Labor Review, October 1987.
Presents multifactor productivity measures for 20 man
ufacturing industries and for total manufacturing, based
on annual measures of output and inputs of capital, labor,
energy, materials, and purchased business services.
Analyzes multi factor growth rates in manufacturing
industries.
Fulco, L. J. “ U.S. Productivity Growth Since 1982: The
Post-Recession Experience,” Monthly Labor Review,
December 1986.
A review of current developments in major sectors
of the economy focusing on the first 14 quarters of the
recovery phase of the business cycle. Contrasts experience
during the recovery which began in the fourth quarter
of 1982 with all previous post-World War II cycles.
Harper, Michael J.; Berndt, Ernst R.; and Wood, David O.
“ Rates of Return and Capital Aggregation Using Alter
native Rental Prices,” bls working paper, 1987.
Examines the theoretical rationale for and empirical
implementation of rental price formulas for use in
weighting capital assets for multifactor productivity
measurement.
74
Technical References—Continued
Harper, Michael J., and Gullickson, William. “ Cost Function
Models and Accounting for Growth in U.S. Manufac
turing, 1949-83,” prepared for American Economic
Association Meetings, December 1986.
The effects of factor substitution induced by relative
price changes on labor productivity are assessed using an
econometric cost function model.
Sveikauskas, Leo. “ The Contribution of R&D to Productivity
Growth,” Monthly Labor Review, March 1986.
Results of a bls study suggest that the direct contribu
tion of research and development to postwar productivity
growth was between 0.1 and 0.2 percent annually in the
nonfarm business sector; r &d had no substantial effect
on the post-1973 productivity slowdown.
Harper, Michael J. “ The Measurement of Productive Capital
Stock, Capital Wealth and Capital Services,” Working
Paper No. 128, 1982.
Analysis of the computation of capital depreciation for
productivity measurement.
Waldorf, William H.; Kunze, Kent; Rosenblum, Larry S.; and
Tannen, Michael B. “ New Measures of the Contribu
tion of Education and Experience to U.S. Productivity
Growth,” prepared for American Economic Association
Meetings, December 1986.
Estimates of education and experience (based on a new
model) show little contribution to productivity growth
(0.2 percent annually) and very little explanation of the
productivity slowdown.
Hulten, Charles R.; Robertson, James W.; and Wykoff, Frank
C. “ Energy, Obsolescence, and the Productivity Slow
down,” prepared for Western Economic Association
Meetings, July 1986.
An empirical examination of the hypothesis that high
energy prices contributed to the post-1973 productivity
slowdown by inducing capital obsolesence.
Kunze, Kent. “ Hours at Work Increase Relative to Hours
Paid,” Monthly Labor Review, June 1985.
The ratio of hours at work to hours paid in nonagricultural establishments increased slightly in 1983.
Kunze, Kent. “ A New bls Survey Measures the Ratio of
Hours Worked to Hours Paid,” Monthly Labor Review,
June 1984.
Hours at work accounted for about 93 percent of hours
paid for production and nonsupervisory workers in 1982,
according to an annual survey which includes only the
time required to be on the job site, thereby excluding paid
holidays, sick leave, and vacations.
Mark, Jerome A. “ Problems Encountered in Measuring
Single-Factor and Multifactor Productivity,” Monthly
Labor Review, December 1986.
Development of new data sources, better utilization of
existing sources, and broader coverage are some of the
ways in which the bls has improved its productivity
measures; progress has been made, but inadequacies
remain.
Mark, Jerome A., and Waldorf, William H. “ Multifactor
Productivity: A New bls Measure,” Monthly Labor
Review, December 1983.
Annual indexes for private business show that advances
in multifactor productivity account for most of the
growth of output per hour of all persons during
1948-81.
Other publications
Caves, Douglas W.; Christensen, Laurits R.; and Diewert, W.
Erwin. ‘‘The Economic Theory of Index Numbers and
the Measurement of Input, Output, and Productivity,”
Econometrica, Vol. 50, No.6, 1983, pp. 1393-1414.
Denison, Edward F. Trends in American Economic Growth,
1929-1982. The Brookings Institution, Washington, DC,
1985.
Gollop, Frank M.; Fraumeni, Barbara M.; and Jorgenson,
Dale W. Productivity and U.S. Economic Growth.
Cambridge, m a , The H arvard University Press,
forthcoming.
Kendrick, John W., and Vaccara, Beatrice N., editors. New
Developments in Productivity Measurement and Analysis.
Chicago, The University of Chicago Press, 1980.
National Research Council’s Panel to Review Productivity
Statistics. Measurement and Interpretation o f Produc
tivity. Washington, DC, The National Academy of
Sciences, 1979.
Norsworthy, J.R.; Harper, Michael J.; and Kunze, Kent.
“ The Slowdown in Productivity Growth: Analysis of
Some Contributing Factors,” Brookings Papers on
Economic Activity. Washington, d c . The Brookings
Institution, Fall 1979.
Usher, Dan, ed. The Measurement o f Capital. Chicago, The
University of Chicago Press, 1980.
75
U.S. Department of Labor
Bureau of Labor Statistics
Hours At Work
Production Workers
This report is authorized by law, 29 U.S.C. 2. Your voluntary cooperation is needed to make the results of this survey
comprehensive, accurate, and timely. The information collected on this form by the Bureau of Labor Statistics will be
held in confidence and will be used for statistical purposes only.
Form Approved
O.M.B. No. 1220-0076
Aoproval Expires 8-31-87
RETURN TO:
r
BUREAU OF LABOR STATISTICS
Room 2068, MAIL CODE 13
441 G STREET, N.W.
WASHINGTON, D C. 20212
n
BLS Use Only
Telephone No. 202-523-5931
Call collect if you need any help
in completing this form.
L
:
(Change name & mailing address if incorrect)
:
PLEASE READ INSTRUCTIONS ON THE BACK BEFORE ENTERING DATA
•
ALL EMPLOYEES/PRODUCTION WORKERS-Enter the total number of All Employees and Production Workers during payroll period which
includes March 12, 1986.
• HOURS PAID—Enter into the table below, the total number of hours that Production Workers were paid during each quarter. This includes all
hours actually worked plus hours for paid vacations, paid sick leave, paid holidays and other paid personal or administrative leave.
• HOURS AT WORK—Subtract from the “HOURS PAID” entry for each quarter the paid leave time (vacation, sick leave, holidays, and other per
sonal or administrative leave) and enter the difference in this column.
A l l E m p lo y e e s
Number of Employees During Payroll Period which
wmm
Includes March 12, 1986
P r o d u c t io n W o r k e r s
B L S U s e O n ly
mm
Hours Information for Production Workers Only
Q u a r te r ly P e r io d
H o u r s P a id (o m it fra c tio n s)
H o u r s a t W o r k (o m it fra c tio n s )
First Q u a rte r 1 9 8 6
J a n u a ry — M a rc h
mm
S e c o n d Q u a rte r 1 9 8 6
A p ril— J u n e
67-74
T h ird Q u a rte r 1 9 8 6
J u ly — S e p te m b e r
Fo u rth Q u a rte r 1 9 8 6
O c to b e r— D e c e m b e r
194.112:
Annual Total 1986
January—December
■
1a. What records did you use to compile the above information?
1 □ Payroll
2 □ Personnel
3 □ Other.
Are these records computerized?
2. What types of paid leave do you offer?
1 □ Vacation
2 □ Sick
3 □ Holiday
1 □ Yes
1 *3-114:
mm
2 D No
117121
4 □ Personal/Administrative
5 □ Other
6 □ None
3. If unable to complete for the establishment(s) identified above, please call collect.
4. Enter below any unusual factors responsible for significant differences from normal hours worked during any quarter. Please indicate in which
quarter these factors occurred. Examples are: more business, layoffs, strikes, fire, weather, seasonal, etc.
If questions arise concerning this report, whom should we contact? (Please Print)
Name
Date
Title
BLS 2000P (Rev. Nov. 1986)
1-BLS COPY
Area Code
Telephone
INSTRUCTIONS FOR COMPLETING REPORT (BLS 2000P)
DATA REQUEST IS FOR CALENDAR YEAR 1986
ALL EMPLOYEES
E n te r th e to tal n u m b e r of p e rs o n s on th e pa yroll(s) w h o w o rk e d full or p a rt-tim e or re c e iv e d p a y fo r a n y part of th e p a y pe rio d w h ich
in c lu d e s M a rc h 1 2 , 1 9 8 6 . If th e e s ta b lis h m e n t(s ) c lo s e d do w n (e ith e r te m p o ra rily or p e rm a n e n tly ) du ring 1 9 8 6 , p le a s e pro v id e d a ta
for th e tim e th e e s ta b lis h m e n t(s ) w a s o p e ra tio n a l. If th e e s ta b lis h m e n t(s ) w a s c lo sed do w n d u rin g 1 9 8 6 , p le a s e in d ic a te be low , th e
d a te s of clo su re.
T o ____________ L____________/
M o n th
D ay
F r o m ____________/____________ L_________
M o n th
Day
Year
Year
PRODUCTION WORKERS
E n te r th e to tal n u m b e r o f p ro d u c tio n w o rk e rs , both full a n d p a rt-tim e , on yo u r payroll(s), w h e th e r w a g e or s a la rie d , w h o w o rk e d d u r
ing or re c e iv e d p a y for a n y p a rt of th e p a y p e rio d w h ich in c lu d e s M a rc h 12.
T h e te rm “ p ro d u c tio n w o rk e r” re fe rs to all o c c u p a tio n a l g ro u p s w h o s e w o rk is not p rim arily a d m in is tra tiv e or m a n a g e ria l, re g a rd le s s
of skill level w ith in th e fo llo w in g in dustries: M in in g a n d Q u a rry in g , C ru d e P e tro le u m , N a tu ra l G a s a n d G a s o lin e P ro d u c tio n , a n d th e
C o n s tru c tio n a n d M a n u fa c tu rin g in d u s tries. T h e s e o c c u p a tio n a l g ro u p s includ e: W o rk in g su p ervis o rs a n d all n o n s u p e rv iso ry w o rke rs ,
(In c lu d in g g ro u p load ers a n d train ees) e n g a g e d in e x c a v a tio n , h a u lin g , tru c kin g , hoisting, v e n tila tio n , d ra in a g e , p u m p in g , drilling,
b la s tin g , lo a d in g , c ru s h in g , p ro c e s s in g , in s p e c tio n , s to ra g e , h a n d lin g , w a re h o u s in g , s h ip p in g , m a in te n a n c e , re p a ir, ja n ito ria l, reco rd
k e e p in g , fa b ric a tin g a n d a s s e m b ly , a s w ell as c ra ft w o rk e rs , m e c h a n ic s , a p p re n tic e s , h e lp e rs , la b o re rs , p lu m b e rs , p a in te rs , p la s
te re rs , c a rp e n te rs , m a s o n s , w e ld e rs or a n y of th e s p e c ia l tra d e s . A lso in c lu d e all o th e r n o n s u p e rv iso ry e m p lo y e e s w h o s e se rv ic e s
a r e c lo sely a s s o c ia te d w ith th o s e e m p lo y e e s a b o v e .
T h e te rm “ p ro d u c tio n w o rk e r ”
excludes
e m p lo y e e s e n g a g e d in th e fo llow ing activities: E x e c u tiv e , p u rc h a s in g , fin a n c e , a c c o u n tin g ,
le g a l, p e rs o n n e l, c a fe te ria , m e d ic a l, p ro fe s s io n a l a n d te c h n ic a l a c tivities, s a le s , a d v e rtis in g , c re d it c o llec tio n , a n d in th e in stallatio n
a n d s e rvicin g of o w n p ro d u c ts, ro u tin e o ffic e fu n c tio n s a n d fa c to ry su p ervis io n (a b o ve w orking s u p e rv is o r’s level). (E m p loyees In
th e ab o v e activ ities , h o w e v e r, s h o u ld b e In c lu d e d In th e ALL EM PLOYEES fig u re .)
Period:
N o rm a lly , d a ta w ill re fe r to c a le n d a r q u a rte r, i.e ., fro m J a n u a ry 1 th ro u g h M a rc h 31 ; A pril 1 th ro u g h J u n e 30 ; J u ly 1 th ro u g h
S e p te m b e r 30 ; a n d O c to b e r 1 th ro u g h D e c e m b e r 3 1 . If yo u r reco rd s re la te to a pe rio d o th e r th a n th e c a le n d a r q u a rte r, p le a s e in
d ic a te b e g in n in g a n d clo sin g d a te s .
Hours Paid: Include
all ho urs fo r w h ic h p a y is re c e iv e d d ire c tly fro m th e e m p lo y e r.
Include
pa id v a c a tio n tim e , p a id sick le a v e ,
p a id ho lida ys a n d o th e r p a id p e rs o n a l or a d m in is tra tiv e le a v e . If p a y m e n ts a re m a d e in lieu of tim e off, rep o rt th e hours e q u iv a le n t
to th e p a y m e n ts m a d e . F o r e x a m p le , th re e ho urs le a v e tim e a t tw o -th ird s th e re g u la r ra te sh ou ld be rep o rted as tw o hours pa id.
Exclude
ho urs a s s o c ia te d w ith u n p a id le a v e , n o rm a l tra v e l tim e fro m h o m e to w o rk, u n p a id w a s h u p tim e , a n d u n p a id m e a l tim e .
Hours at Work: Include
Include, b e s id e s n o rm a l
all tim e a n e m p lo y e e is
required
to b e on th e e m p lo y e r’s p re m is e s , on d u ty, or a t a p re s c rib e d w o rk p la c e .
w o rk in g h o urs, rest p e rio d s , s ta n d -b y tim e , d o w n tim e , tra v e l tim e from jo b site d u rin g w o rk in g d a y an d
tra v e l tim e a w a y fro m h o m e if it c u ts a c ro s s w o rk in g d a y . D o not c o n v e rt o v e rtim e or p re m iu m pa id hours to s tra ig h t-tim e e q u iv a le n t
hours.
NOTE:
F o r s u rv e y p u rp o s e s H O U R S A T W O R K e q u a ls H O U R S P A ID less pa id le a v e tim e (v a c a tio n , sic k le a v e , ho lida ys, a n d o th e r
pa id p e rs o n a l or a d m in is tra tiv e le a v e ).
Annual Total:
T h e su m o f H O U R S P A ID fo r e a c h q u a rte r sh ou ld e q u a l th e A N N U A L T O T A L , H O U R S P A ID fig u re . L ikew is e, th e
su m of H O U R S A T W O R K fo r e a c h q u a rte r s h o u ld e q u a l th e A N N U A L T O T A L H O U R S A T W O R K fig u re.
77
Chapter 11. Productivity
Measures: industries and
Government
Background
Studies of output per employee hour in individual
industries have long been a part of the b l s program. A
study of 60 manufacturing industries in 1898, prompted
by congressional concern that human labor was being
displaced by machinery, was presented in the report Hand
and Machine Labor, this provided striking evidence of
the savings in labor resulting from mechanization in the
last half of the 19th century. The impact of productivity
advance upon employment remained an important focus
of b l s throughout the 1920’s and 1930’s. Also during this
period, the Bureau began the preparation and publica
tion of industry indexes of output per employee hour,
which were based on available production data from the
periodic Census o f Manufactures and employment
statistics collected by b l s .
In 1940, Congress authorized the Bureau of Labor
Statistics to undertake continuing studies of productivity
and technological changes. The Bureau extended earlier
indexes of output per employee hour developed by the
National Research Project of the Works Progress
Administration, and published measures for selected
industries. This work, however, was reduced in volume
during World War II, owing to the lack of meaningful
production and employee hour data for many manufac
turing industries.
The advent of World War II also caused a change in
the emphasis of the program from problems of unem
ployment to concern with the most efficient utilization
of scarce labor resources, b l s undertook a number of
studies of labor requirements for defense industries, such
as synthetic rubber and shipbuilding. After the war, the
industry studies program resumed on a regular basis, and
was supplemented by a number of industry studies based
on the direct collection of data from employers. Budget
restrictions after 1952 prevented the continuation of direct
collection of data. Consequently, the preparation of
industry measures is largely limited to those industries
where data are readily available.
In recent years, public interest in productivity has
grown, and increases in output per employee hour have
been recognized as important indicators of economic
progress and a means to higher income levels, rather than
merely a threat to job opportunities.
The industry studies cover a variety of manufacturing
and nonmanufacturing industries at the 2-, 3-, and 4-digit
Standard Industrial Classification level. Measures for
these industries are published on an annual basis and are
provided for most years beginning in 1947 or 1958 and
continuing through the most recent year for which data
are available.
Coverage includes industries in the manufacturing,
mining, trade, transportation, communication, public
utilities, finance, and business and personal services sec
tors. In addition, productivity measures for various func
tional areas in the Federal Government are published
annually. Recently, studies were initiated to develop pro
ductivity measures for State and local governments, which
employ 80 percent of all civilian government employees.
The Bureau has been expanding its industry produc
tivity measurement program by developing measures
which include other inputs besides labor. The industry
multifactor productivity indexes measure changes in an
industry’s output in relation to changes in labor, capital,
and intermediate purchases. In addition to providing
indicators of productivity change useful for analysis in
their own right, such measures also are helpful in analyz
ing the causes of change in output per hour.
Labor Productivity Measures
Concepts
Indexes of output per employee hour measure changes
in the relationship between the physical volume of an
industry’s output and the employee hours expended in
that output. BLS computes an index of output per
employee hour by dividing an output index by an index
of aggregate employee hours. For most industries,
measures are prepared separately relating output to (a) all
employee hours, (b) production worker hours, and
(c) nonproduction worker hours. (The standard defini
tions of production workers and nonproduction workers
are used.) Three corresponding measures also are com
puted relating output to the number of employees. For
industries in trade and services, measures are prepared
relating output to the hours of all persons involved in pro
ducing that output, including self-employed and unpaid
family workers.
78
The output per employee hour measures relate output
to one input—labor time; they do not measure the specific
contribution of labor, capital, or any other factor of pro
duction. Rather, they reflect the joint effect of a number
of interrelated influences such as changes in technology,
capital investment per worker, utilization of capacity,
layout and flow of material, skill and effort of the work
force, managerial skill, and labor-management relations.
Also, indexes which relate output to one group of employ
ees represent the total output of the industry resulting
from all employees; they are not representative of the
specific contribution of that group of employees.
For an industry producing a single uniform output, this
index of productivity may be expressed as follows:
P =
that were expended in the base year to produce the baseyear composite with the hours that would have been
required in the current year to produce the same com
posite. Thus, these composite indexes eliminate the effects
of shifts, over time, in the relative importance of prod
ucts or services on output per hour.
In either form, an index of output per employee hour
expressed as the quotient of an index of weighted output
and an index of employee hours becomes:
Output index + Employee hours = Output per employee hour
(Laspeyres)
index
index (Paasche)
goji
1
^ _ h_ = qj(q°I°) =
1 ~ q0
Lo
^ cM )
h
Eloqo
_
= £1oq.
' Elodo
Output index ■+■Employee hours = Output per employee hour
(Paasche)
index
index (Laspeyres)
where:
n .^
Pj = the index of productivity or output per employee
hour in the current year
qj and q0 = the output quantities in the current and base years
respectively
L = the aggregate employee hours in the current year
L q = the aggregate employee hours in the base year
lj = unit labor requirement in current year
10 = unit labor requirement in base year
Thus, for an industry producing a single uniform out
put, the index of productivity turns out to be simply the
ratio of the unit labor requirement in the base year to
the unit labor requirement in the current year. If 10 is
greater (less) than li} the ratio (or productivity) is said to
have increased (decreased) over the time period studied.
For an industry producing a number of different prod
ucts or services (the more typical case), the output per
employee hour index is the ratio for two periods of the
total hours required to produce a given composite of
products or services. Indexes of such industries vary with
the composite and can take many forms. Two of these
forms are:
1"
b. Using a base-period composite
P. =
An index constructed according to (a) compares the
employee hours that would have been required in the base
year to produce the current composite with the hours
actually expended in their production. An index con
structed according to (b) compares the employee hours
:a A
” a iOo
The employee hours index measures the change in
aggregate employee hours between the base and current
periods. The employee hours data are the total hours
expended by employees in establishments classified in the
industry to produce the base-period and current-period
composites.
As can be seen in the formulas, the appropriate out
put index is one which compares the quantities of the
various products or services in the current and the base
periods, each weighted by the employee hours expended
per unit produced in a given period. A current-period
weighted output per hour index uses a base-period
weighted output index divided by the employee hours
index. Conversely, a base-period weighted output per
employee hour index is consistent with an output index
which utilizes current-period weights.
Methods and Sources
Industries
a. Using a current-period composite
Eqjl
a ,q .
n-qj
Output
bls industry output indexes are based on quantifiable
units of products or services of the industry combined
with fixed-period weights. Whenever possible, physical
quantities are used as the unit of measurement. For those
industries lacking quantity data, constant-dollar value of
shipments, sales, or revenue data are used to develop the
output series. Quantity data on physical output are
usually most comprehensive for years covered by an
economic census. To make maximum use of the com
prehensive census data, output indexes are derived from
data for two consecutive censuses; these indexes are
79
referred to as benchmark indexes. For intercensal years,
annual indexes are based on either physical output data
(generally in less detail than for census years) or, if such
data are not available, value of output adjusted for price
change (the value of output in constant dollars). The
annual series subsequently are adjusted to the benchmark
levels for the census years.
Weights. In order to derive a labor productivity index
for an industry that is a mean of the productivity move
ments of the component outputs, the various products
are combined with unit employee hour weights. Such
weights are derived from special surveys or from data for
specialized establishments published in the Census o f
Manufactures. In some industries, however, unit employ
ee hour information is not available for individual prod
ucts. In these cases, bls uses substitute weights when it
is believed that they are proportional to unit employee
hour weights; these are usually unit value weights. Unit
value weights are computed from census or survey data
on the quantity and value of shipments of the primary
products of the industry. The introduction of these sub
stitute weights results in an industry output per employee
hour index which reflects shifts in value per employee hour
of the various products in the industry. Thus, a change
can occur in the index without any change in the output
per employee hour for any product of the industry.
The extent to which error or bias may be introduced
by the use of unit value weights is not known. The index
is equivalent to one weighted with unit employee hours
if the unit employee hours and unit values among the
products are proportional, or if there is no correlation
between the relative change in quantity and value per
employee hour.1 There is evidence that unit values are
fairly reliable approximations for individual products in
instances where wages constitute a large proportion of
total value of output. An error generated in the output
index by an error in the weights usually is considerably
smaller than the error in the weights themselves.
In some industries, unit value weights for specific prod
ucts and unit employee hour weights for product groups
are used at different stages in constructing the industry
output indexes. When this procedure is used, the individ
ual products are first aggregated into primary product
group indexes with unit value weights. These indexes are
then combined into an industry output index with primary
product group employee hours. The primary product
group employee hours relate to a base period, as do the
value weights.
To obtain primary product group employee hour
weights, total employee hours for plants specializing in
each primary product class, derived from published
1 Irving H. Siegel, “ Further Notes on the Difference Between Index
Number Formulas, ” Journal o f the A m erican S tatistical A sso cia tio n ,
December 1941, pp. 519-24.
census data on production worker hours and nonproduc
tion worker employment, are supplemented by unpub
lished bls estimates of nonproduction worker hours. (See
section on employee hours later in this chapter for the
procedures used to estimate nonproduction worker
employee hours.) Ratios of employee hours to value of
shipments are multiplied by the corresponding value of
primary products shipped by the entire industry to pro
vide the estimated primary product group employee hour
weights. This procedure assumes that employee hours per
dollar for each product class shipped by the entire industry
are the same as those for plants specializing in the prod
uct group. This procedure is used only when the “speciali
zation” and “coverage” ratios of the industry are high,
and specialization data for all or most of the product
groups are available.2
Most published industry indexes have used: 1947
weights for the 1947-58 period, 1958 weights for 1959-63,
1963 weights for 1964-67, 1967 weights for 1968-72,
1972 weights for 1973-77, 1977 weights for 1978-82,
and 1982 weights for years after 1982. The Bureau updates
the weights as data become available from the periodic
censuses.
Benchmark indexes. For many manufacturing, mining,
trade, and service industries, indexes reflecting changes
in output between census years are constructed. These are
called benchmark indexes and are generally available for
the nonregulated industries for which census data are
collected.
For manufacturing industries, benchmark indexes are
developed through the use of the following procedure:
Price indexes for each primary product class are generated
from data on the value of each individual product within
the class, whether made in the industry or elsewhere. Pro
ducer price indexes are used wherever possible to convert
the product values to constant-dollar estimates. If a pro
ducer price index is not available, a price index is devel
oped using both the quantity and value data reported for
the product in the Census o f Manufactures. The primary
product class price indexes are derived from the sum of
the current-dollar values and the sum of the constantdollar values.
These “wherever made” primary product class price in
dexes are used to deflate the value of primary products
produced only by the industry. This procedure assumes
that the price movements of the primary products within
the industry are the same as the price movements for all
primary products wherever made. These constant-dollar
values are related to corresponding base-year values in
2
The “ specialization ratio” is the value of shipments of primary
products of plants in the industry expressed as a percent of total
shipments of all products (primary plus secondary) made by these same
establishments. The “ coverage ratio” is the value of shipments of the
primary products made by plants classified in the industry as a percent
of the total shipments of the industry’s primary products made by all
producers, both in and out of the specified industry.
80
order to derive separate primary product indexes within
the industry.
These separate primary product indexes are then com
bined with employee hour weights to derive the total
industry primary product output index. The index of
primary products of the industry is multiplied by a
“ coverage” adjustment to represent the total output of
the industry. This adjustment is the ratio of the index of
value of industry shipments (after inclusion of net addi
tions to inventories) to the index of value of shipments
of primary products. The final industry output index thus
reflects inventory buildups and changing proportions of
secondary products.
Benchmark indexes for the mining industries are com
puted from data as reported in the Census o f Mineral
Industries. For trade and service industries, benchmark
indexes are computed from sales data reported in the
Census o f Business.
Annual indexes. Annual output indexes are constructed
by the following procedures. The annual indexes are
adjusted, if necessary, to the levels of the benchmark
indexes previously described. The adjustment factors for
2 census years are used to determine the adjustment fac
tors for the intervening years by linear interpolation.
1. Physical output. Most annual output indexes are
based on physical quantities of products combined with
fixed-period unit employee hour or unit value weights.
The basic quantity data are generally primary products
of an industry classified into product groups; the finest
level of detail available is used. The quantity data relate
to primary products “ wherever made” and, in some
cases, to shipments of the products.
The Bureau’s annual measures of production are con
structed from data on physical quantities of products
which comprise a high percent of the total value of an
industry’s output. Coverage varies between 80 and 100
percent.
2. Deflated value. When adequate annual physical
quantity data are not available, indexes are derived from
data on the value of industry output adjusted for price
change. Since the adjustment for price change is most
often downward, the indexes usually are called “ deflated
value” indexes. Such indexes are conceptually equivalent
to indexes which use data on physical quantities of prod
ucts combined with unit value weights. This index is
derived by dividing the value of the industry’s output by
an industry price index. An index of these deflated values
shows the change in the real value of output between the
past and current periods.3
For manufacturing industries, data on value of pro
duction are often not directly available, and data on value
of shipments must be used. In this case, data on value
of shipments for each year are divided by an industry
price index representing the average annual price for the
year. Beginning and end-of-year finished goods and
work-in-process inventories are also deflated. The
estimated value of shipments in constant dollars is then
adjusted by the net change in inventories—also in con
stant dollars—to yield an estimate of the constant-dollar
value of production. For industries in trade and services,
data on the value of sales for each year are divided by
a specially constructed industry price index to derive a
measure of the change in the industries’ real output.
Sources. Industry output indexes are prepared from
basic data published by various public and private agen
cies, using the greatest level of detail available.
Data from the Bureau of the Census, U.S. Department
of Commerce, are used extensively in developing output
statistics for manufacturing, trade, and service industries.
The Bureau of Mines, U.S. Department of the Interior,
compiles most of the information for the mining and
cement industries. Other important Government sources
include the U.S. Department of Energy, the U.S. Depart
ment of Agriculture, the Fish and Wildlife Service, U.S.
Department of the Interior, the Interstate Commerce
Commission, the Internal Revenue Service, and the
Department of Transportation. Important sources of
trade association data include the Textile Economics
Bureau, Inc., National Association of Hosiery Manufac
turers, Inc., National Canners Association, Rubber
Manufacturers Association, and the American Iron and
Steel Institute.
For deflated value series, industry price indexes are
derived from producer and consumer price indexes
developed by the Bureau of Labor Statistics.
Employee hours
An index of employee hours is computed by dividing
the aggregate employee hours for each year by the baseperiod aggregate. Because of data limitations, employee
hours are treated as homogeneous and additive with no
distinction made between hours of different groups of
employees. For industries in which the self-employed are
important, indexes are constructed for the hours of all
3 For example:
Value index/Price index (Paasche) = Output index (Laspeyres)
EPi Qj
EPj Qj
Epoqo
' Epoqi
Ep^
”
Ep0q0
where p( and po represent prices of products in the industry in the cur
rent and base periods, respectively. This index requires quantities of
all items produced in each year. These data are not available for the
particular industries where this measure is used, and quantity data are
usually available for the base year only. Accordingly, the deflated value
indexes employed usually take the following form:
Value index/Price index (Laspeyres) = Output index (Paasche)
81
Epj q(
Ep; qQ
Ep; ^
Epoqo
' Epoqo “
EPiqo
persons, which includes paid employees, partners, pro
prietors, and unpaid family workers.
Sources. Industry employment and employee hour
indexes are developed from basic data compiled by the
Bureau of Labor Statistics, the Bureau of the Census, and
other sources. For most private nonagricultural indus
tries, BLS publishes employment and average weekly
hours data for production or nonsupervisory workers and
employment data for all employees. The Bureau of the
Census publishes employment and aggregate hours data
for production workers and employment data for all
employees.
BLS and the Bureau of the Census differ in their defi
nition of employee hours and in their sampling and
reporting methods. In general, BLS data are the pre
ferred source for measuring industry employment and
hours. Census employment is the average of production
workers plus the number of other employees in midMarch. The number of production workers is the average
for the payroll periods for the 12th of March, May,
August, and November. In contrast, the b l s (790)
employment statistics program is a cooperative Federal
and State project. Employment and hours are collected
monthly and are benchmarked each year to compre
hensive data from the State unemployment insurance
programs.
Nonproduction worker hours. Only employment data
are available for nonproduction workers. The average
annual hours of these workers must be estimated. The
estimates of aggregate nonproduction worker employee
hours for the manufacturing industries are derived from
published employment data, and estimates of average
annual hours worked or paid per nonproduction worker.
For years prior to 1968, the estimates of average annual
hours worked were calculated by multiplying the number
of workweeks in the year times the scheduled weekly
hours. This produced an estimate of annual hours paid.
Estimated hours for vacations, holidays, disability, and
personal time off were subtracted from average annual
hours paid to obtain an estimate for average annual hours
worked.
Estimated hours for vacations, holidays, and disabili
ties were based on data from various b l s surveys and
studies of the U.S. Department of Health and Human
Services. Personal time off was estimated as a constant
from references in relevant publications.
From 1968 to 1977, the estimates of average annual
hours paid and hours worked were based on data col
lected in the b l s biennial surveys of employee compen
sation in the private nonfarm economy. Since these
surveys are no longer conducted, the 1977 levels are being
carried forward until other data become available.
For the mining industries, estimates for the hours of
nonproduction workers are based on data collected by
the Mine Safety and Health Administration. For the trade
and service industries, estimates are made for the hours
of partners, proprietors, and unpaid family workers using
unpublished data collected in the Current Population
Survey, and for supervisory workers using data from the
Census o f Population.
All-employee hours estimates for manufacturing indus
tries are derived by summing the aggregate hours for pro
duction workers and the estimated aggregate hours for
nonproduction workers. For trade and service industries,
all-person hours estimates are derived by summing the
aggregate hours for paid employees and the estimated
aggregate hours for partners, proprietors, and unpaid
family workers.
Comparability of output and employee hours data
For manufacturing industries, employee hours data are
based on total employee hours of establishments classified
in an industry. Annual physical output data, on the other
hand, usually include the products which are primary to
an industry that are reported on a “ wherever made”
basis. Thus, there can be some discrepancy in the
coverage of output and employee hours measures. This
is not a serious problem unless there is considerable varia
tion from year to year in the proportion of primary prod
ucts to total products of an industry, or if there is a
change in the proportion of primary products which are
made in other industries. The comparability of the
employee hours and output data is indicated by the
specialization and coverage ratios which the Bureau of
the Census publishes. All industries in the b l s industry
measurement program have high and stable specializa
tion and coverage ratios.
In selecting industries for the measurement program,
attention is also given to changes in the degree of ver
tical integration. Employee hours relate to all operations
performed by establishments of an industry, while out
put usually is measured in terms of the final product. If
establishments undertake additional operations (such as
the manufacture of components which had previously
been purchased from suppliers) employee hours will
increase but there will be no corresponding increase in
final output. Thus, output per employee hour indexes
would be biased. In developing industry indexes, b l s
examines data such as the ratio of cost of materials to
value of shipments for any indication of a change in the
degree of vertical integration.
Government
Federal
Indexes of output per employee year, output, and em
ployee years for selected functional areas of Government
82
activity4 and for approximately 400 participating
organizations are constructed in a manner similar to that
described for industries. At the present time, these
measures cover about 69 percent (2.1 million employee
years) of the Federal civilian work force.
Ideally, a productivity index should relate final out
puts to their associated direct and indirect input(s), and,
in fact, the output data are final from the perspective of
the functional areas within which these data are classified.
However, since the outputs of one organization may be
consumed wholly or partially by another Federal
organization in the production of its final outputs, all out
put indicators in the Federal sample may not be final from
the perspective of a higher level of organization; for
example, the entire Federal Government. Therefore, the
overall statistics do not represent “ Federal productivity”
but rather, the weighted average of the productivity
changes of the measured Federal organizations included
in the sample.
Through an annual collection process, most data are
submitted directly by agencies to BLS, but in some cases
data are obtained from secondary sources such as agency
budgets and annual reports. In the Federal sample, more
than 3,000 products and services are aggregated into out
put indexes by combining the quantities of each type of
output by their respective base-year labor requirements.
These unit employee year weights are constructed from
the detailed output and input data provided by each
organization. Although the weights relate to fixed periods,
they are updated every 5 years. The output segments are
linked and referenced to a fiscal year 1977 base.
The organizational indexes are grouped into 28 func
tional categories, based on similarity of activity. Some
of these categories, such as printing and duplication, are
more homogeneous than others, such as general support
services, which include many diverse activities. Nonethe
less, these categories provide insight into the trends for
the major functional areas underlying the overall sam
ple. Although productivity, output, and input indexes are
also constructed for each participating organization, these
are not published but are returned to each organization
for its own use (for example, to stimulate further exami
nation of the causes of productivity change within each
organization). This is one method used by BLS to validate
the basic data (that is, by examining the reasonableness
of the derived trends).
4 The 28 functions are: Audit of operations; buildings and grounds;
communications; education and training; electric power production and
distribution; equipment maintenance; finance and accounting; general
support services; information services; legal and judicial activities; library
services; loans and grants; medical services; military base services;
natural resources and environmental management; personnel investiga
tions; personnel management; postal services; printing and duplication;
procurement; records management; regulation—compliance and
enforcement; regulation—rulemaking and licensing; social services and
benefits; specialized manufacturing; supply and inventory control; traffic
management; and transportation.
Employee year indexes are developed from agency
data submissions and secondary sources. As in all labor
input measures used by the Bureau to develop produc
tivity indexes, employee years are considered homoge
neous and additive. Each employee year reflects the
regularly scheduled time, overtime, and leave time of all
full-time, part-time, or intermittent employees. An
employee year is equivalent to one individual paid for
2,087 hours.
State and local
Government measurements are being extended to
include indexes of output per employee year, output, and
employee years for selected State and local services. A
cross-section of services is being examined and indexes
computed using concepts and methods similar to those
for industries and the Federal Government. This research,
which uses published and readily available secondary
data, is keyed to the Standard Industrial Classification
system.
Output indexes reflect final services to the public. The
consequences or effectiveness of government service are
not measured. The focus is on government production.
Prison system outputs, for example, are the number of
prisoners housed and to whom treatment is administered,
not the effectiveness of rehabilitation or community
safety. Output data are taken from Federal agencies such
as the Departments of Labor, Justice, and Transporta
tion and industry groups such as the American Public
Power Association, American Public Transit Association,
and Distilled Spirits Council of the U.S.
Employee year index computations use the same
general concepts as those used in the Federal measure
ments. The primary source of employment data is the
Bureau of the Census’ annual public employment survey.
This is supplemented by data from Federal agencies and
industry associations when available.
Multifactor Productivity Measures
Concepts
The industry multifactor productivity indexes measure
productivity growth by measuring changes in the relation
ship between the quantity of an industry’s physical out
put and the quantity of inputs consumed in producing
that output, where measured inputs include capital and
intermediate purchases (including raw materials, pur
chased services, and purchased energy) as well as labor
input.
The index used to calculate multifactor productivity
is the Tornquist index and is of the form:
83
In A
In _Qt - w k I !n
Qt-i
A
+
K t -l
+ w ip | In
WL
IP .
IP t-i
where:
physical quantity of output for that category to obtain
the unit value (or price) estimates. The price of each out
put category is multiplied by its corresponding quantity
and then expressed as a share of the total value of out
put. These value shares are averaged at time t and t-1.
Sources. Data sources are the same as those used in
calculating indexes for the labor productivity measures.
In = the natural logarithm of the variable
A = multifactor productivity
Q = output
K = capital input
L = labor input
IP = intermediate purchases input
WK, WL, Wip = compensation share weights
The weights are the means of the compensation shares
in two adjoining time periods.
(S,‘ + S,-)
‘
Employee hours
Employee hour indexes are calculated in the same way
as those used in measuring industry labor productivity.
As with the labor productivity measure, employee hours
are treated as homogeneous and additive with no distinc
tion made between hours of different groups of employ
ees. The index is computed by simply dividing the
aggregate employee hours for each year by the baseperiod total.
Sources. Data sources are the same as those used in
calculating employee hour indexes for the labor produc
tivity measures.
2
where:
Capital
1
S(Pil X /)
P|l = price of input Xj in period t
The Tornquist formula yields growth rates which are
differences in logarithms. The antilogs of these rates are
taken to get the percent changes in multifactor produc
tivity. These percent changes are chained together to form
the index.
Methods and Sources
Output
The multifactor productivity output measures are
calculated using, whenever possible, the same units of
products or services of the industry as are used in measur
ing output for the industry labor productivity measures.
Whenever possible, physical quantities are used as the unit
of measurement; when physical quantities are unavail
able, constant-dollar value of shipments, sales, or revenue
data are used. The multifactor productivity output meas
ures differ from the labor productivity output measures
primarily in the method used for weighting together the
various categories of output, as explained below.
The measure of capital input is based on the flow of
services derived from the stock of physical assets. Physical
capital is composed of equipment, structures, land, and
inventories. Financial capital is excluded. Capital services
are estimated by calculating capital stocks; changes in the
stocks are assumed proportional to changes in capital
services for each asset. Stocks of different asset types are
Tornquist-aggregated, using estimated rental prices to
construct the weights for assets of different types.
Capital stocks are calculated using the perpetual inven
tory method, which takes into account the continual addi
tions to and subtractions from the stock of capital as new
investment and retirement of old capital take place. The
perpetual inventory method measures stocks at the end
of a year equal to a weighted sum of all past investments,
where the weights are the asset’s efficiency relative to a
new asset. A hyperbolic age-efficiency function is used
to calculate the relative efficiency of an asset at different
ages. The hyperbolic age-efficiency function can be
expressed:
St = (L - t) / (L - (B)t)
where:
St = the relative efficiency of a t-year-old asset
L = the service life
t
Weights. The multifactor productivity output measure
utilizes price weights for combining the various categories
of output of an industry. Generally, census value data
for each detailed output category are divided by the
= the age of the asset
B = the parameter of efficiency decline
The parameter of efficiency decline is assumed to
be 0.5 for equipment and 0.75 for structures. These
84
parameters yield a function in which assets lose efficiency
more slowly at first, then rapidly later in life.
Stocks of equipment, structures, inventories, and land
are estimated separately. Individual price deflators for
each asset category are constructed and used to convert
the current-dollar investment to constant dollars.
Industry-specific service lives are computed for each type
of equipment asset for use in the perpetual inventory
method.
Current-dollar values of inventory stocks are calculated
for three separate categories of manufacturers’ inven
tories: Finished goods, work in process, and materials and
supplies. Inventory stocks for each year are calculated
as the average of the end-of-year stocks in years t and
t-1 to represent the average utilized during the year as
a whole. This is also done with equipment, structures,
and land. Current-dollar inventory values for the three
categories of inventories are deflated with appropriate
price indexes.
Land stocks are estimated as a function of the move
ment in constant-dollar gross structures stocks for the
given industry.
Weights. The various equipment, structure, inventory,
and land stock series in constant dollars are aggregated
into one capital input measure using estimated rental
prices as weights. Rental prices are calculated for each
asset as:
RP = [(P x R) + (P x D) - (Pl - P11)] x (l-u z -k )/(l-u )
where:
RP = the rental price
Sources. Industry capital indexes are developed from
basic data published and maintained by the Bureau of
the Census, U.S. Department of Commerce; the Bureau
of Economic Analysis, U.S. Department of Commerce;
and the Office of Economic Growth, Bureau of Labor
Statistics. Price indexes are derived from producer price
indexes developed by the Bureau of Labor Statistics.
Intermediate purchases
The index of intermediate purchases input is a Torn
quist aggregate of separate indexes of change in real
materials, services, fuels, and electricity consumed by an
industry. With the exception of electricity, for which both
price and quantity data are available, the above indexes
are calculated by dividing annual current-dollar values
by appropriate price indexes to obtain constant-dollar
annual estimates. Separate price deflators for materials
and fuels for each industry are constructed using detailed
price and value data for individual subcomponents of
each group. The aggregate deflators are divided into the
current-dollar values to derive constant-dollar estimates.
The constant-dollar series for each component are
indexed by dividing each year’s estimate by the baseperiod aggregate.
Weights. The indexes of change in real materials, serv
ices, fuels, and electricity are weighted together with value
share weights to derive an aggregate intermediate pur
chases index. These weights are derived by dividing the
current-dollar values of each by the total combined value
of intermediate purchases, and averaging these weights
at times t and t-1.
P = the deflator for the given asset type
R = the internal rate of return
D = the rate of depreciation for a given asset type
Sources. Industry intermediate purchases indexes are
developed from basic data published by the Bureau of
the Census and the Bureau of Economic Analysis.
pt-p t-1 = the capital gain term for the asset
( l - u z - k ) / ( l - u ) reflects the effects o f taxation
and:
u = the corporate tax rate
z = the present value of $1 of depreciation
deductions
k = the effective investment tax credit rate
This method of calculating rental prices is similar to
that used in calculating multifactor productivity for major
sectors of the economy except that no attempt is made
to incorporate the effects of indirect business taxes, for
which data are lacking at the industry level.
The rental prices are expressed in rates per constant
dollar of productive capital stocks. Each rental price is
multiplied by its constant-dollar capital stock to obtain
current-dollar capital costs which are then converted to
value shares for Tornquist aggregation.
Weights for major input components
The indexes representing quantity change for each of
the three major inputs are weighted together to compute
the index of combined inputs. The relative weights for
each year are derived from total costs for each input. All
employee labor costs from census data are used for the
labor weight. The sum of current-dollar values for
materials, services, fuels, and electricity constitute the
weight for intermediate purchases. The weight for capital
is derived by subtracting labor costs and an estimate of
purchased services from Census value-added data. These
compensation shares are averaged at time t and t-1.
Presentation
BLS industry and governm ent indexes are published an
nually in the bulletin, Productivity Measures fo r Selected
85
Industries and Government Services. A limited amount
of the most current data is provided in an annual news
release. As new industry indexes are developed, they are
presented as articles in the Monthly Labor Review. The
articles contain an analysis of productivity, output, and
employment trends in the industry. Technical notes
describing the methodology used to develop the indexes
are available on request. Unpublished indexes for all 4-,
3-, and 2-digit Sic manufacturing industries are available
for analytical purposes upon request.
Indexes of output per employee hour also are published
in the Statistical Abstract o f the United States and in the
Handbook o f Labor Statistics, on l a b s t a t and on b l s
data diskettes. Some indexes for earlier years are
published in Historical Statistics o f the United States.
b l s Federal Government indexes are also available in
the Handbook o f Labor Statistics, on l a b s t a t data
tapes and on b l s data diskettes. More detailed Federal
data and State and local government data are available
from b l s .
Uses and Limitations
Measures of output per employee hour are particularly
useful for studying changes in labor utilization, project
ing future employment requirements, analyzing trends in
labor costs, comparing productivity progress among
countries, examining the effects of technological
improvements on employment and unemployment, and
analyzing related economic and industrial activities. Such
analysis usually requires that indexes of output per
employee hour be used in conjunction with other data.
Specifically, related data on production and employment
are useful in studying technological effects; to study
trends in labor costs, data on earnings and other labor
expenditures are necessary.
These productivity measures of output per employee
hour are subject to certain qualifications. First, existing
techniques cannot fully take into account changes in the
quality of goods and services produced. Second, although
efforts have been made to maintain consistency of
coverage between the output and labor input estimates,
some statistical differences may remain. Third, changes
in the degree of plant integration and specialization often
are not reflected adequately in the production statistics.
This may result in overstatement of productivity gains
in some years and understatement in others. Fourth,
indexes involving nonproduction worker hours are sub
ject to a wider margin of error than are the indexes using
production worker hours because of the technique for
estimating average employee hours of nonproduction
workers. Errors in estimating hours of nonproduction
workers, however, have a relatively insignificant effect
on the estimates of hours for all employees. Fifth,
industries in which all person hours are used as the
denominator are subject to a wider margin of error
because of the limited data available for unpaid family
workers, the self-employed, and paid managers. Finally,
year-to-year changes in output per employee hour are
irregular, and, therefore, are not necessarily indicative
of basic changes in long-term trends. Conversely, long
term trends are not necessarily applicable to any one year
or to any period in the future. Because of these and other
statistical limitations, these indexes cannot be considered
precise measures; instead they should be interpreted as
general indicators of movements of output per employee
hour.
Indexes of multifactor productivity are subject to many
of the same limitations previously mentioned with the
exception of the effects of changes in the ratio of other
factor inputs to labor. Since the multifactor indexes relate
output to inputs of labor, capital, and intermediate pur
chases, increases in one input relative to others are not
reflected as productivity change. Indeed, one of the uses
of multifactor productivity measures is in analyzing the
effects of movements of capital and intermediate pur
chases, relative to labor, versus other influences on labor
productivity change.
86
Technical References
of government productivity; stresses the need for detailed
product data and discusses examples of current efforts
in the collection of, and the improvement in, pertinent
data; and surveys problems of concepts, methods, and
data adequacy in measuring productivity at the State and
local government levels.
Bureau of Labor Statistics
Fisk, Donald M. Measuring Productivity in State and Local
Government, bls Bulletin 2166, December 1983.
Reports on a study of ways that national labor pro
ductivity trends might be calculated for State and local
government. Reviews past research and studies, examines
available national data, and outlines a strategy for fur
ther work.
Kutscher, Ronald E., and Mark, Jerome A. “ The ServiceProducing Sector: Some Common Perceptions Re
viewed,” M onthly Labor Review, April 1983.
Compares the growth in output per hour in the serviceproducing industries to the goods-producing industries.
Also examines the level of capital intensity in each sec
tor and the underlying employment shifts between the two
sectors.
Mark, Jerome A. “ Industry Indexes of Output Per ManH our,” Monthly Labor Review, November 1962.
Describes the methods used in constructing bls
indexes of output per employee hour. Covers methods
and sources, construction of production and employee
hour indexes, and limitations.
Mark, Jerome A. “ Measuring Productivity in Service Indus
tries,” Monthly Labor Review, June 1982.
Surveys problems of measurement of output and input.
Discusses bls measures for such industries as retail food
stores, eating and drinking places, intercity trucking and
bus transportation, communications, banking, hotels and
motels, and others.
Sherwood, Mark. “ Multifactor Productivity in Steel and
Motor Vehicle Industries,” Monthly Labor Review,
August, 1987.
Describes new multifactor productivity measures for
two industries. Explains the relationship of multifactor
productivity to labor productivity and discusses underly
ing trends in output and inputs of labor, capital, and
intermediate purchases.
Other publications
Kendrick, John W., and Vaccara, Beatrice N., eds. New
Developments in Productivity Measurement and Analysis.
Studies in Income and Wealth, Vol. 44. Chicago, The
University of Chicago Press, 1980.
Collection of papers on such subjects as labor and
multifactor productivity by industry; productivity in
selected service sectors; and international comparisons of
productivity. Includes a study of high and low produc
tivity establishments; current efforts to measure produc
tivity in the public sector; effects of research and develop
ment on industry productivity growth; and energy and
pollution effects on productivity and international com
parisons of economic growth.
National Academy of Sciences. Measurement and Interpreta
tion o f Productivity. Washington, 1979.
Collection of papers on such topics as the concepts and
measurement of productivity; the limitations of produc
tivity statistics; the measurement of outputs and inputs;
the sources of economic growth; measures of company
productivity; and international comparisons of produc
tivity.
Mark, Jerome A. “ Measuring Productivity in Government—
Federal, State, and Local,” Public Productivity Review,
March 1981.
Differentiates between measures of efficiency, interme
diate work activity, and effectiveness; describes concepts,
methods, and problems relevant to the measurement
87
Chapter 12.
Change
Technological
economy and include those where the pace may be slow
as well as those where rapid change is expected.
The emphasis of these studies is on technological devel
opments within each industry in an early stage of the
innovation’s commercial use; i.e., the period after intro
duction on the market but before widespread adoption.
Inventions and discoveries still in the “ drawing board”
stage are considered unlikely to have as much impact over
the next decade as those already tested and are generally
not discussed.
The report briefly describes recent technological devel
opments, indicating insofar as practicable some economic
advantages of various types of new equipment, processes,
or products; their importance in terms of the employee
hours engaged in the operations affected; estimated extent
of use currently and in 5 to 10 years; and some factors
affecting adoption such as the volume of investment and
expenditures for research and development. The advan
tages described include not only labor savings per unit,
but also quality improvements, fuel and material econo
mies, greater accuracy, new markets, etc.
In assessing the employment implications of techno
logical changes, account is taken of the possible rate of
growth in output per employee hour and in the industry’s
total output. Appraisal also is made of the changes in
occupational structure and of some issues and examples
of adjustment of workers to technological change.
Background
Studies of technological changes and their labor impli
cations have been undertaken by bls over the years for
a variety of purposes. During the 1930’s, public interest
focused on the unemployed, and reports were prepared
on displacement of workers resulting from technological
change in various industries. During World War II, emerg
ing technologies were studied for purposes of improving
work force utilization.
Beginning in the mid-1950’s, nationwide attention was
focused on the implications of new developments classi
fied under the general term “ autom ation.” BLS made a
series of studies on a plant basis, in the insurance, petro
leum refining, bakery, air lines, and electronics industries,
to explore the labor implications of various changes.
Later, broader studies were undertaken, including a
survey of the labor impact of changeover to electronic
computers in 20 large companies and intensive studies of
technological change in the coal and paper industries.
These studies formed the basis, beginning in the early
1960’s, for a more systematic investigation of likely future
changes. The first report, entitled Technological Trends
in 36 Major American Industries, was issued by the Presi
dent’s Advisory Committee on Labor-Management
Policy in 1964. A revised edition covering 40 industries
was published in 1966. More recent industry studies have
been published as they are completed.
Technological innovation studies
Description of Studies
Some technological innovations have applicability in
many industries. Among these are such developments as
computers, numerical control of machine tools, material
handling equipment, and control instruments. Because
of their far-reaching impact, special studies have been
made of the nature, status, prospects for adoption, and
implications for unit labor requirements, occupational
change, training needs, and problems of industrial rela
tions. In analyzing their impact in different industries,
differences as well as similarities are revealed.
The Bureau’s research program on technological
change involves preparing two basic types of studies:
Summary reports surveying trends in major industries,
and studies undertaken periodically of major techno
logical innovations, such as computers, that affect
workers in different industries.
Summary reports on major industries
To provide a broad overview of significant trends in
the economy, the Bureau prepares a summary report,
applying to key industries, on new types of machinery,
processes, and products which are believed likely to have
an important effect over the next 5 to 10 years. The
industries covered comprise a cross-section of the
Data Sources and Collection Methods
A variety of data sources and collection methods are
utilized in making studies of technological change and
its impact.
88
Personal interviews
In making studies, analysts personally conduct inten
sive interviews with plant managers, personnel directors,
and other officials who have direct knowledge of changes
at their plant. Union officials at the plant and, in some
cases, individual workers are interviewed. The analyst
uses a checklist of questions in conducting informal inter
views in order to elicit the maximum amount of data.
Plants and offices included in these studies are selected
on the basis of having recently made a major change in
their equipment, products, or methods of production.
Personal interviews also are used to help determine
industry trends. Informal interviews are conducted with
engineers, scientists, economists, and other experts in
companies which produce and use new technology, and
with unions, trade associations, government agencies,
universities, etc., who have specialized knowledge of a
particular technological development or industry trend.
One objective in these cases is to obtain their expert judg
ment about the nature, pace of introduction, and possi
ble impact of developments with which few plants have
had any experience. The emphasis in these interviews is
on the technological change rather than on experiences
in adjusting.
Trade and t^jhnical publications
Important sources of information concerning techno
logical trends are trade journals, technical magazines and
books, conference proceedings, government hearings, and
company reports. Annual reports of leading corporations
and company house organs often contain useful infor
mation on current technological developments in some
industries. These publications are reviewed to obtain
information about the status and prospects of important
developments and to ascertain which companies and
plants merit more intensive field visiting. Reports and
publications of firms that produce particular types of
equipment often are found useful in studies of industries
that use such equipment.
Statistical data sources
Quantitative information about the status of specific
technological developments is fragmentary and scarce.
The Bureau makes use of available data from many
public and private sources. These sources include: General
Services Administration, annual inventory of computers
in the Federal Government; International Trade Admini
stration, U.S. Industrial Outlook (annual); International
Data Corporation, E D P Industry Report; American
Bankers’ Association, survey of banking automation;
and American Machinist, inventory of metalworking
machinery.
Statistical information on industrywide trends is useful
in analyzing the economic implications of technological
change. Among the important sources used are the
Bureau’s indexes of output per employee hour and related
series on production, employment, and hours; the Bureau
of the Census’ data on expenditures on plant and equip
ment; and the National Science Foundation’s estimates
of research and development.
Plant records
In making detailed studies of the impact of techno
logical change on individual workers within a plant,
analysts sometimes can obtain from employers’ files data
on such aspects as the age, sex, and related personal
characteristics of employees whose jobs are eliminated
and the jobs in the plant held by each individual affected
before and after the change. Similar data are collected
on individuals who are selected for the positions created
in connection with automated equipment.
Expert review
In preparing forecasts of technological trends, a critical
step is the review of preliminary reports by outstanding
experts in each industry. Drafts of industry reports are
mailed to company executives, union research directors,
trade association officials, technical journal editors, and
university and government specialists for their assessment
of the validity and adequacy of projected trends. Over
450 persons were contacted in this way in the prepara
tion of a report on technological trends in major indus
tries. Some experts are visited personally to review draft
statements in detail. Reports on technological prospects
are designed to reflect, as much as possible, the authori
tative views of a number of persons who have expert,
firsthand knowledge of each industry.
Analysis and Interpretation
For a better understanding of research results in this
field, it is important to keep in mind the meaning of cer
tain key ideas and concepts. Some of the problems of
interpretation and analysis, therefore, are set forth
briefly.
Definition of technological change
Technological change is defined broadly in the bls
studies as encompassing significant changes in processes
and equipment, products and services produced, and
materials, fuels, and energy used. The term “ automa
tion,” which is sometimes popularly used as a synonym
for “ technological change,” designates, strictly speak
ing, a particular type of current development. It has
been variously defined, for example, as “ automatic
operation,” “ the mechanization of sensory control and
89
thought processes,” and ‘‘a concern with production
processes as a system.”
While bls studies have been concerned with develop
ments in automation, particularly in anticipating long
term trends, they are not the only technological changes
taking place that affect labor requirements and labor
relations. For example, new ways of generating power,
piggybacking in transportation, use of synthetic mate
rials in manufacturing, mechanized methods of material
handling, and faster steelmaking processes are impor
tant technological developments not usually covered by
technical definitions of automation, but having signifi
cant employment implications.
Impact on productivity
Since one of the principal consequences of techno
logical change, so far as work force utilization is con
cerned, is an increase in productivity (output per
employee hour), special attention is given in bls studies
to analyzing changes in industrial productivity. Such
trend analysis is a useful method of measuring the pace
of technological change. Changes in productivity, how
ever, also reflect changes in capacity utilization and many
other nontechnical factors. It is important to recognize
that the productivity trend is only a partial measure of
the rate of technological change.
In determining the impact of a specific technology, bls
studies try to indicate the reduction in unit labor
requirements that the new process is designed to achieve.
In some cases, estimates of labor savings are derived on
the basis of comparisons with the estimated average tech
nology of the industry under study; in others, with the
best equipment that is available; or in actual plant studies,
with the technology that is actually displaced.
It is also important to distinguish between the impact
on productivity of the operation directly affected and on
productivity of the plant as a whole. An advanced
machine tool, for example, may result in a relatively large
reduction in unit labor requirements in the machining
operation, but would have little impact on finishing and
assembling, and may even require additional labor in
engineering and maintenance work. The impact on plant
productivity, therefore, would be considerably less than
the effect on productivity of any department or opera
tion directly affected.
Impact on employment
In assessing the impact of technological change on em
ployment, it is necessary to consider the implications of
plant policies and the effects of economic changes with
which technological changes interact. Analysis of the
impact of technological change purely in terms of
machinery is incomplete.
At the plant level, for example, the substitution of
machinery for labor may substantially reduce job oppor
tunities in operations directly affected. If efforts are
made, however, to eliminate these jobs by not filling
vacancies or by transfer of affected workers to other posi
tions in the plant or office, labor savings could be
achieved without displacing the workers affected.
Moreover, the employment impact of technological
change is also interrelated with the effects of the business
cycle. Thus, workers whose jobs are eliminated by
technological changes may not be displaced from a plant
until a decline in demand results in layoffs—a long time
after the change has been made in some cases. In the
subsequent recovery, however, they may not be hired
back because their jobs no longer exist.
The employment trend for the industry as a whole must
also be examined. The plant which reduces its unit costs
through technological improvement may be able to gain
a larger share of the market and increase its employment,
but at the expense of the less technically advanced com
peting plants, which may be forced to shut down, displac
ing workers far from the location of the change.
Because of the complex of economic factors that oper
ate through the market, including changes in demand,
location, foreign competition, corporate organization,
and consumer taste, it is very difficult to isolate the effects
of technological change.
Impact on occupations
Two aspects of occupational change resulting from
technological changes are examined. Changes in job
structure—the distribution of the plant or office work
force by function or broad skill grouping—are studied
to determine the extent of upgrading or downgrading.
Since the content of jobs may be altered as a result of
changes in equipment or processes, attention also is
directed to intensive before-and-after analysis of job
duties and the knowledge and abilities required to per
form these duties as indicated by job descriptions and
observation. The content of newly created jobs also is
studied, and the qualifications required and personal
characteristics of individuals selected for these new posi
tions are described, so far as possible.
Adjustment to technological change
Technological change has important implications for
personnel management and collective bargaining within
plants. The introduction of new machinery, products, or
processes often requires movement of workers among
jobs. Often the adjustment proceeds according to rules
established in advance through collective bargaining. Pro
visions to assist workers whose jobs are eliminated include
severance pay, retraining, and early retirement. Besides
analyzing the operation of formal provisions under col
lective bargaining, Bureau studies describe informal
90
efforts to provide training, to utilize attrition, and to
obtain jobs for displaced workers elsewhere. The limita
tions of these measures as well as their advantages are
important matters studied.
Uses and Limitations
BLS studies and reports of technological change are
useful to managers, union leaders, educators, economists,
government officials, and others in planning policies to
cushion the impact of change. The study of emerging
technological trends and possible implications, moreover,
provides a basis for more valid projections of produc
tivity and economic growth. They also are useful in
pinpointing employment problems and determining the
most productive direction of future research to obtain
possible solutions.
Some limitations of the Bureau’s studies of technolog
ical change must be kept in mind in assessing their
appropriateness for particular uses. In general, it is
important to recognize that judgments about the future
direction and pace of technological change and its
implications are necessarily complex and difficult. The
rate of introduction of new technology depends not only
on technical advantages but also on many economic fac
tors, such as the volume of investment, market prospects,
and the availability of trained workers, all of which are
subject to significant variations. Moreover, since the
period of introduction generally spans a number of years,
the outlook must be reappraised from time to time in the
light of new information.
Finally, studies of the impact of technological change
deal primarily with changes within individual industries.
But these changes often involve changes in the type and
amount of goods and services purchased from other
industries and could, therefore, have important implica
tions for production and employment in industries sup
plying inputs. The accumulation of information on
interindustry relationships, through the Bureau’s eco
nomic growth studies, provides a quantitative basis for
analyzing this aspect of technological change.
Technical References
Bureau of Labor Statistics
Technology and Labor in Four Industries (meat products,
foundries, metalworking machinery, electrical and elec
tronic equipment), bls Bulletin 2104, 1982.
Appraises major technological changes emerging in key
industries and discusses their impact on productivity and
occupations over the next 5 to 10 years.
Mark, Jerome A. “Technology and Employment: Some Results
of bls Research,” Monthly Labor Review, April 1987.
Highlights findings from the bls research program on
technological change including implications of new tech
nology for employment and productivity.
Technology, Productivity, and Labor in the Bituminous Coal
Industry, 1950-79, bls Bulletin 2072, 1981.
Appraises some of the major structural and techno
logical changes in the bituminous coal industry and their
impact on labor in the industry.
Outlook fo r Computer Process Control, bls Bulletin 1658, 1970.
Describes the impact of computer process control on
employment, occupations, skills, training, and labormanagement relations in six process industries. Discusses
outlook for future and implications for production and
productivity.
The Impact o f Technology on Labor in Five Industries
(printing and publishing, water transportation, copper ore
mining, fabricated structural metal, intercity trucking),
bls Bulletin 2137, 1982.
Appraises major technological changes emerging in key
industries and discusses their impact on productivity and
occupations over the next 5 to 10 years.
Technological Change and Its Labor Impact in Four Industries
(hosiery, folding paperboard boxes, metal cans, laundry
and cleaning), bls Bulletin 2182, 1984.
Appraises major technological changes emerging in key
industries and discusses their impact on productivity and
occupations over the next 5 to 10 years.
The Impact o f Technology on Labor in Four Industries
(textiles, paper and paperboard, steel, motor vehicles), bls
Bulletin, 2228, 1985.
Appraises major technological changes emerging in key
industries and discusses their impact on productivity and
occupations over the next 5 to 10 years.
Technology and Its Impact on Labor in Four Industries (tires,
aluminum, aerospace, banking), bls Bulletin 2242, 1986.
Appraises major technological changes emerging in key
industries and discusses their impact on productivity and
occupations over the next 5 to 10 years.
The Revised Workweek: Results o f a Pilot Study o f 16 Firms,
bls Bulletin 1846, 1975.
Explores the impact of changes in the workweek sched
ule to determine objectives and methods for introducing
workweek changes and to assess the availability of data
for further research on the implications for productivity
and employment.
Technology and Labor Developments in Four Industries
(lumber and wood products, footwear, hydraulic cement,
wholesale trade), bls Bulletin 2263, 1986.
Appraises major technological changes emerging in key
industries and discusses their impact on productivity and
occupations over the next 5 to 10 years.
91
Chapter 13.
Statistics
Foreign Labor
Background
From its inception, the Bureau has collected and pub
lished statistical information on labor conditions and
developments abroad. Foreign labor research and statis
tical analyses have been undertaken because (1) infor
mation on labor conditions published by a majority of
foreign countries is not readily available to U.S. labor
representatives, employers, Government officials, and
others, and is often not available in English; (2) often,
only an expert can judge the quality of foreign statistical
sources; (3) comparisons between U.S. and foreign labor
conditions shed light on U.S. economic performance
relative to other industrial nations; and (4) comparisons
provide information on the competitive position of the
United States in foreign trade, which has an important
influence on the U.S. economy and employment.
Description of Measures
The emphasis of the current program is on the develop
ment of international comparisons of the labor force,
employment, and unemployment; productivity and labor
costs; hourly compensation costs of manufacturing pro
duction workers; and trends in consumer prices and real
compensation of manufacturing employees. The meas
ures compiled relate primarily to the major industrial
countries, but other countries of importance to U.S.
foreign trade are included in the hourly compensation cost
measures. Most of the series are prepared on an annual
average basis; comparative figures on unemployment and
consumer prices are prepared monthly.
Labor force, employment, and unemployment. Com
parative measures of the labor force, employment, unem
ployment, and related indicators are prepared regularly
for the United States, Canada, Japan, Australia, France,
Germany, Italy, the Netherlands, Sweden, and the United
Kingdom. For most of the countries, the series begin with
1959. Unemployment rates, approximating U.S. con
cepts, are prepared monthly for most of the countries;
the other measures are calculated annually. Current
research is directed toward the development of a series
of comparative unemployment measures ranging from
relatively narrow measures to measures encompassing
employed persons working part time for economic
reasons and discouraged workers.
Productivity and labor costs. Comparative trends in
manufacturing labor productivity (output per hour),
hourly compensation, unit labor costs (labor compensa
tion per unit of output), and related measures are com
piled on an annual average basis for the United States,
Canada, Japan, Korea, Belgium, Denmark, France, Ger
many, Italy, the Netherlands, Norway, Sweden, and the
United Kingdom. The comparisons are limited to trend
measures only; reliable comparisons of levels of manufac
turing productivity and unit labor costs are not available.
Trends are expressed in index form and as percentage
changes at annual rates. For most countries, the series
begin with 1950. Indexes of unit labor costs for foreign
countries are calculated on a U.S. dollar basis as well as
in national currency terms to take account of relative
changes in currency exchange rates. Comparative
measures by broad industry within manufacturing have
also been developed for Japan, France, and Germany.
Current research is directed toward the expansion of the
measures for several countries, including Japan and Ger
many, to include capital as well as labor inputs in the pro
ductivity measures.
Comparative levels and trends in productivity and labor
costs in the iron and steel industry in the United States,
Japan, France, Germany, and the United Kingdom have
been compiled annually beginning with 1964. The
measures express levels of foreign output per hour, hourly
compensation, and unit labor costs relative to the U.S.
level (United States = 100). They also show trends in index
form and at annual rates of change.
Comparative levels (United States = 100) and trends in
gross domestic product ( g d p ), g d p per capita, and g d p
per employed person are calculated on an annual aver
age basis for the United States, Canada, Japan, Korea,
and nine European countries beginning with 1950. The
g d p level comparisons, which are based on estimated
purchasing-power-parity ( p p p ) exchange rates, are benchmarked to data from the United Nations International
Comparison Project. Purchasing-power-parity exchange
rates represent the number of foreign currency units
required to buy goods and services equivalent to what can
be purchased with one unit of U.S. currency. A common
practice has been to base such comparisons on official
92
market exchange rates. However, market exchange rates
seldom reflect the relative purchasing power of different
currencies.
Hourly compensation costs. Measures of total compen
sation per hour worked for production workers in all
manufacturing and in 39 component manufacturing
industries are computed for about 35 countries. The series
are annual and begin with 1975. For all manufacturing,
measures of hourly direct pay and pay for time worked
are also computed. The measures are computed in
national currency and converted into U.S. dollars at
prevailing commercial currency exchange rates. Hourly
compensation converted into U.S. dollars at commercial
exchange rates provides a measure of comparative levels
of employer labor costs. It does not indicate relative liv
ing standards of workers or the purchasing power of their
income. Prices of goods and services vary greatly among
countries, and commercial exchange rates are not reliable
indicators of relative differences in prices.
Consumer prices. Indexes of consumer prices are
compiled regularly for the United States and 14 foreign
countries on a common base year. Annual indexes since
1950 and monthly or quarterly indexes since 1970 are
available for most of the countries. Annual indexes for
selected component series are also compiled for 12
countries.
Other measures. Other comparative measures, generally
available on an annual basis, include indexes of real
hourly and weekly compensation of manufacturing
employees for 13 countries; the number of work stop
pages resulting from industrial disputes and their severity
rates, as measured by days lost per thousand employees
in nonagricultural industries, for 15 countries; and
selected producer price indexes for 8 countries.
Data Sources
Research on comparative labor statistics is based
upon statistical data and other source materials from
(a) statistical agencies of foreign countries; (b) interna
tional and supranational bodies such as the United
Nations, International Labour Office ( il o ), Organization
for Economic Cooperation and Development (OECD),
and the Statistical Office of the European Communities
( e u r o s t a t ); and (c) private agencies such as banks,
industry associations, and research institutions. All data
are drawn from secondary sources; the Bureau does not
initiate surveys or data collection programs abroad. The
U.S. Department of State provides many of the foreign
periodicals and publications used and provides assistance
in obtaining answers to many technical questions about
foreign data series.
Estimating Procedures
Because statistical concepts and methods vary from
country to country, international comparisons of statis
tical data can be misleading. The Bureau attempts to
derive meaningful comparisons by selecting a conceptual
framework for comparative purposes; analyzing foreign
statistical series and selecting those which most nearly
match the desired concepts; and adjusting statistical
series, where necessary and feasible, for greater intercoun
try comparability.
Labor force, employment, and unemployment. For
these comparisons, the Bureau adjusts each country’s
published data, if necessary, to provide measures approx
imately consistent with U.S. definitions and standards.
Although precise comparability may not be achieved,
these adjusted figures provide a better basis for interna
tional comparisons than the figures regularly published
by each country. The statistics for 6 of the 10 countries
regularly studied—the United States, Canada, Australia,
Japan, Italy, and Sweden—are obtained from monthly
or quarterly household surveys. No adjustments are made
to the published data for Canada and Australia, since
their concepts and methods are virtually identical to those
in the United States. Slight adjustments are made to the
data for Japan and Sweden; a substantial adjustment is
made to the Italian data.
Current unemployment measures for the other four
countries studied—France, Germany, the Netherlands,
and the United Kingdom—are derived from monthly
administrative data on the number of registrants at public
employment offices. These four countries also conduct
periodic household surveys of the labor force (the United
Kingdom conducts a monthly survey; however, because
of the small sample size, figures are only published on
an annual basis; France and Germany conduct annual
surveys; and the Netherlands conducts a survey bien
nially) which contain benchmark data that are used to
adjust the levels of the labor force, employment, and
unemployment for greater comparability with U.S. con
cepts. Measures of current labor force, employment, and
unemployment are obtained by applying adjustment fac
tors from the most recent year’s labor force surveys to
published figures.
Productivity and labor costs. Indexes of manufactur
ing labor productivity, hourly compensation, and unit
labor costs are constructed from three basic aggregative
measures: Output, total hours, and total compensation.
The hours and compensation measures refer to all
employed persons including self-employed persons in the
United States and Canada, and to all employees in the
other countries. Hours refer to hours paid in the United
States and to hours worked in the other countries. In
general, the measures relate to total manufacturing as
93
defined by the International Standard Industrial Classi
fication. However, the measures for France (beginning
1959), Italy (beginning 1970), and the United Kingdom
(beginning 1971) refer to mining and manufacturing less
energy-related products, and the figures for the Nether
lands exclude petroleum refining from 1969 to 1976.
The long-term output measures are gross product
originating in manufacturing (value added) in constant
prices from the national accounts of each country—ex
cept those for Japan prior to 1970 and the Netherlands
prior to 1960 and from 1969 to 1977, which are indexes
of industrial production. While methods of deriving
national accounts measures of manufacturing output dif
fer substantially from country to country, and the British
national accounts measures are essentially identical to
their indexes of industrial production, the use of different
procedures does not, in itself, connote lack of compara
bility—rather, it reflects differences among countries in
the availability and reliability of underlying data series.
For current measures, indexes of industrial production
are used until national accounts data become available.
The aggregate hours measures are developed from
statistics of manufacturing employment and average
hours. The series used for Canada and Sweden are official
series from their statistical agencies. For the other coun
tries, the total hours measures are developed by the
Bureau using employment data either published with the
national accounts or from other comprehensive employ
ment series, and estimates of annual hours worked.
The compensation (labor cost) measures are from
national accounts—except those for Belgium, which are
developed by the Bureau using statistics of employment,
average hours, and hourly compensation. Compensation
includes all payments in cash or kind made directly to
employees plus employer expenditures for legally required
insurance programs and contractual and private benefit
plans. In addition, for some countries, compensation is
adjusted for other significant taxes on payrolls or employ
ment (or reduced to reflect subsidies), even if they are not
for the direct benefit of workers, because such taxes are
regarded as labor costs. However, compensation does not
include all items of labor cost. The costs of recruitment,
employee training, and plant facilities and services—such
as cafeterias and medical clinics—are not covered because
data are not available for most countries. Self-employed
workers are included in the U.S. and Canadian figures by
assuming that their hourly compensation is equal to the
average for wage and salary employees.
For all countries, preliminary estimates of hours and
compensation for recent years are generally based on cur
rent indicators of manufacturing employment, average
hours, and hourly compensation until national accounts
and other statistics used for the long-term measures be
come available.
The Bureau’s 1964 and 1972 measures of comparative
productivity and labor costs in the iron and steel industry,
with the exception of the exclusion of wire products for
Japan, wheels and axles for Germany, and wire and wire
products for the United Kingdom, are based on the U.S.
definition of the industry, which covers blast furnaces,
steelworks, and rolling and finishing mills (sic 331). In
addition, each country’s output has been measured using
a common set of weights (U.S. 1977 labor requirements
for about 70 products), and the labor input data have
been carefully matched with the output figures. Measures
for years subsequent to the latest benchmark are obtained
by applying trend indexes to the benchmark measures.
Except for the United States, the trend indexes are based
on different output weights and less comprehensive data
sources than those used for the benchmark years. The
level comparisons for the four countries are presented in
ranges, showing minimum and maximum estimates for
each country relative to the United States, rather than
as single best estimates, because of gaps in the available
data.
The Bureau’s measures of comparative levels and
trends of gross domestic product per capita and per
employed person are based on benchmark levels of GDP
extrapolated or interpolated to other years, and on annual
population and employment estimates. The g d p level
comparisons are based on estimated purchasing-powerparity exchange rates. The employment figures for some
countries have been adjusted for greater comparability
with U.S. concepts.
The benchmark (currently 1985) level comparisons of
g d p for all countries except Korea were derived from
data produced for Phase V of the United Nations Inter
national Comparison Project ( u n i c p ) by the OECD and
EUROSTAT. The benchmark figures were derived by com
paring relative prices at detailed levels of expenditure
( p p p exchange rates by item of expenditure) and
aggregating these price relatives to derive overall p p p
exchange rates for g d p . Each country’s current 1985
value of g d p was then converted to a real volume
measure in U.S. dollars using the p p p exchange rate for
total g d p . These original volume measures of g d p have
been modified by b l s , where applicable, to take account
of subsequent revisions by countries of their national
accounts by applying the 1985 p p p exchange rates for
total g d p to the revised country measures of 1985 g d p .
The figures for Korea are based on a comparable set of
1980 benchmark comparisons for a group of Asian coun
tries, including Japan, which were linked to the OECD
countries at the level of total g d p using Japan as a
“ bridge” country. The benchmark-year comparisons of
real g d p were extrapolated by b l s to other years using
relative changes in g d p at constant market prices, as
measured by each country.
Hourly compensation costs. Measures of hourly com
pensation costs are prepared because hourly compensa
tion provides a better basis for international comparisons
94
of labor costs than the earnings statistics which are
regularly published by most countries. Average earnings
do not include all items of labor compensation, nor do
they include the same items of compensation in each
country. Hourly compensation is defined as all direct
payments made to the worker (pay for time worked, pay
for time not worked, all bonuses, and pay in kind) before
payroll deductions of any kind, plus employer expendi
tures for legally required insurance programs and con
tractual and private benefit plans. In addition, for some
countries, total compensation is adjusted for other taxes
on payroll or employment (or reduced to reflect sub
sidies), even if they are not for the direct benefit of
workers, because such taxes are regarded as labor costs.
For consistency, compensation is measured on an hoursworked basis for every country.
The total compensation measures are derived by
adjusting each country’s published earnings series for
items of direct pay not included in earnings and for
employer expenditures for Social Security, contractual
and private insurance programs, and other labor taxes
or subsidies. For the United States and other countries
that measure earnings on an hours-paid basis, the figures
are also adjusted in order to approximate compensation
per hour worked. Adjustment factors are obtained
primarily from periodic labor cost surveys and inter
polated or projected to nonsurvey years on the basis of
other available information, or they are obtained from
censuses of manufactures or reports on Social Security
and fringe benefit systems. The underlying earnings
statistics for some countries are also adjusted, where
possible, to account for major differences in worker
coverage; differences in industrial classification systems;
and changes over time in survey coverage, sample bench
marks, or frequency of surveys. Compensation is con
verted to U.S. dollars using average daily exchange rates
for the reference period, as published by the Federal
Reserve Board or the International Monetary Fund.
Consumer prices. No adjustments are made to the
overall consumer price indexes as published by each
country except to convert them to a uniform base year.
Indexes for selected component series are adjusted, where
possible, for consistency of item coverage among
countries.
Other measures. Indexes of real hourly or weekly com
pensation are constructed by deflating indexes of nominal
compensation by each country’s consumer price index.
Work stoppages usually refer to strikes and lockouts, but
the exact definition differs from country to country. The
statistics are not adjusted for comparability. No
adjustments are made to country producer price indexes
except to link indexes published on different base years
and to convert them to a common index base.
Analysis and Presentation
Analyses of international labor statistics focus upon
comparisons with U.S. data. Wherever possible, foreign
data are adjusted to U.S. definitions and concepts to
facilitate comparisons; for example, the adjustment of
foreign unemployment rates to approximate U.S. con
cepts and the adjustment of production worker earnings
to total hourly compensation.
Labor force, employment, and unemployment data
are analyzed to determine the sources or components of
differences and changes in labor force measures. Shifts
in labor force composition are analyzed by age, sex, and
industrial sector. Productivity and unit labor cost data
are analyzed to explain the relative contributions of
changes in output, employment, average hours, com
pensation, and exchange rates upon changes in the
measures. Changes in employee compensation are
analyzed to determine the relative contributions of
direct pay and other elements of compensation.
The presentation of foreign labor statistics varies with
the degree of analysis and major use of the data. Com
prehensive bulletins have been published, covering
manufacturing productivity and labor cost trends, steel
productivity and costs, unemployment and labor force
comparisons, and youth unemployment comparisons.
For more current developments, articles are published
periodically in the Monthly Labor Review. Some series
are published regularly in the statistical section of the
Monthly Labor Review; an annual news release is issued
on comparative trends in manufacturing productivity and
labor costs; and the hourly compensation cost measures
for total manufacturing are issued in b l s reports. The
b l s ’ s Handbook o f Labor Statistics and the Bureau of
the Census’ Statistical Abstract o f the United States pub
lish many of the principal foreign data series, and some
series are published in the annual Economic Report o f
the President. Many unpublished tabulations of current
comparative data are available on request.
Uses and Limitations
The principal uses of information on foreign labor
statistics are (a) to assess U.S. economic performance
relative to other industrial countries; (b) to inform
Government and private officials of foreign economic
developments that may affect U.S. international
economic policy; (c) to evaluate the competitive posi
tion of the United States in international trade; (d) to
review foreign experience for possible application
domestically; and (e) to provide labor statistics and
related information to individuals, corporations, labor
unions, and others concerned with foreign investment and
development.
Although considerable progress has been made in
95
making international economic statistics more uniform
among countries, e.g., through the work of international
agencies such as the United Nations and the International
Labour Office, international statistical comparisons
should be used cautiously. Nevertheless, through careful
analysis of each country’s data, valid statistical com
parisons can be made. Whenever possible, b l s adjusts
foreign data, if necessary, for greater consistency with
U.S. measures; in some cases, data are sufficiently simi
lar in definition and concept for valid comparisons
without adjustment. Moreover, when conceptual dif
ferences are substantial, the Bureau attempts to describe
the differences in sufficient detail to provide guidance in
the interpretation of the data.
Technical References
Labor force, employment, and unemployment
International Labour Office. International Recommendations
on Labour Statistics. Geneva, 1976.
Presents recommendations on standardizing labor
statistics, including recommendations on employment and
unemployment statistics and statistics of labor costs.
International Labour Office. “ Thirteenth International Con
ference of Labour Statisticians,” Bulletin o f Labour
Statistics, 1983-3.
Presents revisions to the recommendations on employ
ment and unemployment statistics.
Moy, Joyanna, and Sorrentino, Constance. “ Unemployment,
Labor Force Trends, and Layoff Practices in 10 Coun
tries,” Monthly Labor Review, December 1981.
Sorrentino, Constance. “ Japan’s Low Unemployment: An
In-Depth Analysis,” Monthly Labor Review, March
1984, and “ Japanese Unemployment: bls Updates Its
Analysis,” Monthly Labor Review, June 1987.
Sorrentino, Constance. “ The Uses of the European Commu
nity Labour Force Surveys for International Unemploy
ment Comparisons,” Statistical Office of the European
Communities, Conference on the European Community
Labour Force Surveys in the Next Decade. Luxembourg,
October 1987.
Investigates the uses of the European Community (ec )
labor force surveys for purposes of international com
parisons of unemployment and provides comparisons of
alternative measures of unemployment for the ec coun
tries, the United States, Canada, and Japan.
U.S. Department of Labor, Bureau of Labor Statistics. Inter
national Comparisons o f Unemployment, Bulletin 1979,
August 1978, and subsequent unpublished country
supplements.
Provides the conceptual framework and a comprehen
sive description of the Bureau’s work on international
unemployment comparisons, describes in detail the
methods of adjusting foreign unemployment rates to U.S.
concepts, and analyzes various factors contributing to dif
ferences in unemployment levels.
Bureau of Labor Statistics. Youth Unemployment: A n Inter
national Perspective, Bulletin 2098, September 1981.
Also, Sorrentino, Constance. “ Youth Unemployment:
An International Perspective,” Monthly Labor Review,
July 1981.
Examines the labor market experience of youth in the
United States and eight other industrial countries from
the early 1960’s to the late 1970’s.
Productivity and labor costs
Dean, Edwin; Darrough, Masako; and Neef, Arthur. “ Altern
ative Measures of Capital Inputs for Computation of
Multifactor Productivity Growth in Japanese Manufac
turing.” National Bureau of Economic Research, Confer
ence on Research in Income and Wealth, U .S.-Japan
Productivity Conference. Cambridge, Massachusetts,
August 1985.
Examines and assesses the various data sources and
methods available for measuring capital stock in Japanese
manufacturing.
Hill, T.P. The Measurement o f Real Product. Paris, Organ
ization for Economic Cooperation and Development,
February 1971.
A theoretical and empirical analysis of the growth rates
for different industries and countries.
Hill, Peter. “ International Price Levels and Purchasing Power
Parities.” Paris, Organization for Economic Cooperation
and Development, Economic Studies No. 6, Spring
1986.
An examination of relative price levels, purchasing
power parities, and exchange rates based on data derived
from the o e c d project to calculate real gross domestic
product and associated purchasing power parities for
1980.
Kravis, Irving B. “ A Survey of International Comparisons of
Productivity,” The Economic Journal, Vol. 86, March
1976.
Provides a survey of the wide variety of research on
international comparisons of levels of productivity.
96
Technical References—Continued
Kravis, Irving B. “Comparative Studies of National Incomes
and Prices,” Journal o f Economic Literature, Vol. XXII,
March 1984.
Summarizes developments in the study of compara
tive national incomes and price levels with special
attention to the United Nations International Compari
son Project.
Kravis, Irving B. “ The Three Faces of the International
Comparison Project,” The World Bank, Research Ob
server, Vol. 1, No. 1, January 1986.
Neef, Arthur, and Dean, Edwin. “Comparative Changes in
Labor Productivity and Unit Labor Costs by Manu
facturing Industry: United States and Western Europe,”
American Enterprise Institute Conference on Inter
industry Differences in Productivity Growth. Washing
ton, DC, October 1984.
Analyzes labor productivity growth in component
manufacturing industries in four countries—the United
States, France, Germany, and the United Kingdom—and
the effect of industrial composition on total manu
facturing productivity growth and the labor productivity
slowdown each country has experienced since about
1973.
Nelson, Richard R. “ Research on Productivity Growth and
Productivity Differences: Dead Ends and New Depar
tures,” The Journal o f Economic Literature, Vol. 19,
September 1981.
Reviews the research on productivity growth over time
and across countries.
Shelton, William C., and Chandler, John H. “Technical Note
—International Comparisons of Unit Labor Cost: Con
cepts and Methods,” Monthly Labor Review, May 1963.
Ward, Michael. Purchasing Power Parities and Real Expend
itures in the o e c d . Paris, Organization for Economic
Cooperation and Development, 1985.
Report on the o e c d project to calculate real gross
domestic product and associated purchasing power
parities for 1980.
United Nations.
World Comparisons o f Purchasing Power
and Real Product fo r 1980. Phase IV o f the International
Comparison Project. New York, 1986.
U.S. Department of Labor, Bureau of Labor Statistics. A n
International Comparison o f Unit Labor Cost in the
Iron and Steel Industry, 1964: United States, France,
Germany, United Kingdom, Bulletin 1580, 1968.
Provides the conceptual framework and 1964 results
of the Bureau’s comparisons of absolute levels of pro
ductivity and labor costs in a major industry.
97
Chapter 14. Occupational
Safety and Health Statistics
Part I.
Annual Survey of Occupational
Injuries and Illnesses
Background
The Bureau of Labor Statistics has long been interested
in statistics on safety and health conditions for workers
on the job and issued its first report on work injuries as
early as 1893. Subsequent BLS publications reflected a
growing concern for the worker disabled on the job and
were helpful in the development of the present workers’
compensation system
The Occupational Safety and Health Act of 1970 made
recordkeeping and reporting of occupational safety and
health data mandatory. In 1971, the Secretary of Labor
delegated to the Commissioner of the Bureau of Labor
Statistics the responsibility for “ furthering the purposes
of the Occupational Safety and Health Act by develop
ing and maintaining an effective program of collection,
compilation, analysis and publication of occupational
safety and health statistics.’’ The Secretary further
directed the Commissioner to coordinate the above func
tions with the Assistant Secretary for Occupational Safety
and Health.
The recordkeeping system, which is the foundation of
the Bureau’s statistical program in this field, was
developed to aid the Occupational Safety and Health
Administration ( o s h a ) in setting standards, to assist
safety and health officers in identifying hazardous opera
tions, to provide b l s and State agencies with uniform
and reliable safety and health statistics, to provide
employers and employees with information about con
ditions at their workplace, and to aid the National
Institute for Occupational Safety and Health ( n i o s h ) in
its research. The records must contain information
suitable for use by Federal and State safety and health
officers, and include sufficient data to help management
and employees pinpoint problem areas.
Recordkeeping and Reporting
Requirements
Several major changes in the recordkeeping system
have taken place since it was first implemented in July
1971. First, in an effort to reduce the recordkeeping
burden on employers with small-sized establishments,
those with fewer than eight employees were administra
tively exempted from the recordkeeping requirements in
January 1973. An exemption from recordkeeping require
ments for employers with fewer than 11 employees was
made a permanent part of the regulations in July 1977.
In December 1982, the exemption from routine keeping
of the log and supplementary record was extended to all
low-risk industries in retail trade; finance, insurance, and
real estate; and services (except SIC’s 52-54, 70, 75-76,
and 79-80). In January 1975, the classification of lost
workdays was modified to include both days away from
work and days of restricted work activity.
Recordkeeping form o s h a N o. 200, Log and Summary
of Occupational Injuries and Illnesses, designed to stream
line OSHA recordkeeping and reporting, was implemented
in January 1978. This form made it easier for employers,
employees, and safety and health officers to identify the
major injury and illness problems. (A facsimile of o s h a
No. 200 is included at the end of the chapter.)
On April 24, 1986, OMB approved revised recordkeep
ing guidelines for occupational injuries and illnesses.
These guidelines provide supplemental instructions for
the recordkeeping forms and represent the Department
of Labor’s interpretation of employer recordkeeping
requirements under the Occupational Safety and Health
Act of 1970 and 29 CFR Part 1904.
The cases which must be recorded include all workrelated deaths, illnesses, and those injuries which result
98
in: Loss of consciousness, restriction of work or motion,
transfer to another job, or medical treatment beyond first
aid. Employers must record each case as either a fatality,
an injury or illness with lost workdays, or an injury or
illness without lost workdays in a one-line entry on the
form. A case is recorded as a lost workday case if it
involves 1 or more days following the day of injury or
onset of illness on which the employee was away from
work or unable to perform all or any part of his or her
normal assignment during all or any part of the workday
or shift. The number of such days is recorded in two
categories: Days away from work or days of restricted
work activity. Days of restricted work activity are days
when the employee is assigned to another job on a tem
porary basis, works at a permanent job less than full time,
or works at a permanently assigned job but cannot per
form all duties normally connected with it. Chart 1 is a
guide to the recordability of cases under the act.
Each case must also be described in detail on a sup
plementary record ( o s h a N o. 101) or equivalent, such as
a State’s workers’ compensation form if it includes all nec
essary information. The annual summary of occupational
injuries and illnesses must be posted at each establishment
where notices to employees are customarily posted no later
than February 1 and remain in place until March 1. (A
copy of o s h a No. 101 is included at the end of the
chapter.)
Concepts
Definitions used in the annual survey are the same as
those used in the o s h a recordkeeping system. Reports for
all recordable injuries and illnesses occurring during the
year include information on the number of fatalities, inju
ries and illnesses with workdays lost, the number of work
days lost, and injuries and illnesses without workdays lost.
To determine priorities in the development of safety
standards and in o s h a compliance activities, data must
be collected and presented in a manner that allows for
comparison among industries and establishments of vary
ing sizes. Therefore, incidence rates are produced for each
type of case reported under OSHA definitions. Incidence
rates express various measures of injuries and illnesses in
terms of a constant, i.e., exposure hours in the work envi
ronment (200,000 employee hours or the equivalent of 100
full-time employees working for 1 year), thus allowing for
a common statistical base across industries regardless of
employment size of establishments. In this way, the injury
and illness experience of a firm with 5 cases recorded for
70 employees may be shown on the same base as that of
an entire industry with 12,000 cases for 150,000 employees.
(The method of calculating incidence rates is discussed
in a later section.)
Comparisons may also be made to evaluate the per
formance of a particular industry over a period of time,
similar establishments in the same industry, or establish
ments in the same industry but in different geographic
areas. Further comparisons are possible using the different
types of rates computed for each industry—rates for total
cases, cases that involve lost workdays, cases that do not
involve lost workdays, and the number of workdays lost.
These measures are available for injuries, illnesses, and
injuries and illnesses combined.
Scope of the Survey
The survey sample selected by b l s consists of approx
imately 280,000 units in private industry. Survey data are
solicited from employers having 11 employees or more in
agricultural production and from all employers in
agricultural services, forestry, and fishing; oil and gas
extraction; construction; manufacturing; transportation
and public utilities; wholesale trade; retail trade; finance,
insurance, and real estate; and services industries (except
private households). Data for employees covered by other
Federal safety and health legislation are provided by the
Mine Safety and Health Administration of the U.S. De
partment of Labor and the Federal Railroad Administra
tion of the U.S. Department of Transportation. The
Occupational Safety and Health Administration collects
and compiles comparable data for Federal agencies.
Although State and local government agencies are not
surveyed for national estimates, several States have legisla
tion which enables them to collect these data. Selfemployed persons are not considered to be employees
under the act.
State Participation
Federal grants covering about 50 percent of the
operating cost permit States to develop estimates of
occupational injuries and illnesses and to provide the data
from which BLS produces national results. National data
for selected States which do not have operational grants
are collected directly by BLS and by the State agencies
under contract. The participating State agencies collect
and process the data and prepare estimates using stand
ardized procedures established by b l s to insure uniform
ity and consistency among the States. To further insure
comparability and reliability, b l s designs and identifies
the survey sample for each State and, through its regional
offices, validates the survey results and provides technical
assistance to the State agencies on a continuing basis.
Data Collection
State agencies mail report forms ( o s h a N o . 200-S) to
selected employers in February to cover the previous calen
dar year’s experience. For those States not participating
99
in the program, reporting forms are mailed by b l s . Each
employer completes a single report form which is used
for both national and State estimates of occupational
injuries and illnesses. This procedure eliminates duplicate
reporting by respondents and, together with the use of
identical survey techniques at the national and State levels,
insures maximum comparability of estimates. (A copy of
o s h a No. 200-S is included at the end of the chapter.)
Information for the injury and illness portion of the
report form is copied directly from the Log and Summary
of Occupational Injuries and Illnesses. The form also con
tains questions about the number of employee hours
worked (needed in the calculation of incidence rates), the
reporting unit’s principal products or activity, and average
employment to insure that the establishment is classified
in the correct industry and employment-size class. State
agency personnel edit the completed report forms and
verify apparent inconsistencies through phone calls, cor
respondence, or visits. The data are keypunched and
mechanically edited. Reports which do not meet the com
puter screening criteria are verified with the employer.
By midsummer, the active collection phase of the survey
is completed and the preparation of data for both national
and State estimates of occupational injuries and illnesses
begins.
Sample
Because the survey is a Federal-State cooperative pro
gram and the data must meet the needs of participating
State agencies, an independent sample is selected for each
State. The sample is selected to represent all private
industries in the States and territories. The sample size
for the survey is dependent upon (1) the characteristics
for which estimates are needed, (2) the industries for
which estimates are desired, (3) the characteristics of the
population being sampled, (4) the target reliability of the
estimates, and (5) the survey design employed.
While there are many characteristics upon which the
sample design could be based, the Bureau elected to use
the total recorded case incidence rate. This is considered
to be one of the most important characteristics and, im
portantly, the least variable; therefore, it requires the
smallest sample size.
The salient features of the sample design employed are
its use of stratified random sampling with a Neyman
allocation and a ratio estimator. The characteristics used
to stratify the establishments are the Standard Industrial
Classification (sic) code and employment. Since these
characteristics are highly correlated with an establish
ment’s number and rate of recorded injuries and illnesses,
stratified sampling provides greater precision and, thus,
results in a smaller sample size. The Neyman allocation
produces the minimum sample size which will provide an
estimate with a given sampling variance. For the larger
employment-size classes, the allocation procedure places
all of the establishments of the frame in the sample; as
employment decreases, smaller and smaller proportions
of establishments are included in the sample. The cer
tainty strata are usually the size groups with more than
100 employees. The precision of the sample is further
improved, hence permitting a reduction in sample size,
by using the ratio estimator which utilizes available aux
iliary information (employment) that is correlated with
the characteristics which are to be measured.
The sample is designed to produce data at the 2-digit
sic industry level in agriculture, forestry, and fishing; the
3-digit level in oil and gas extraction, construction, and
transportation and public utilities; the 4-digit level in
manufacturing; and the 2-digit level in SIC’s 50-89,
except for some 3-digit estimates in this range of s i c ’s.
Estimating Procedures
Weighting
By means of a weighting procedure, sample units are
made to represent all units in their size class for a par
ticular industry. The weight is determined by the inverse
of the sampling ratio for the industry/employment-size
class from which the unit was selected. Because a small
proportion of survey forms are not returned, weights of
responding employers in a sampling cell are adjusted to
account for the nonrespondents. The respondents are
then shifted into the estimating cell determined by the
employment and business activity reported. Data for each
unit are multiplied by the appropriate weight and
nonresponse adjustment factor. The products are then
aggregated to obtain a total for the estimating cell.
Data for an individual estimating cell are weighted
according to the following formula:
n
where:
X; = weighted estimate of characteristics, e.g., number
of cases reported, in size class i
Wy = weight of sample unit (establishment) j in size
class i, adjusted for nonresponse
Xjj = characteristics reported by sample unit j in size
class i
Benchmarking
Since the universe file which provides the sample frame
is not current to the reference year of the survey, it is
necessary to adjust the data to reflect current employ
ment levels. This procedure is known as benchmarking.
In the annual survey, all estimates of totals are adjusted
by the benchmark factor at the estimating cell level. The
100
benchmarking procedure requires a source of accurate
employment data which can be converted into annual
average employment figures for the cell level in which
separate estimates are desired. Because industry/employment-size data are required for national estimates, bench
mark factors are calculated using both industry level
employment data and size class level employment data.
The benchmark factors are applied to the size class “ blow
up” estimates.
Incidence rate calculation
Incidence rates are calculated using the total obtained
through the weighting and benchmarking procedures. The
adjusted estimates for a particular characteristic are
aggregated to the appropriate level of industry detail. The
total is multiplied by 200,000 (the base of hours worked
by 100 full-time employees for 1 year). The product is
then divided by the weighted and benchmarked estimate
of hours worked as reported in the survey for the industry
segment.
The formula for calculating the incidence rate at the
lowest level of industry detail is:
(Sum of characteristic reported) X 200,000
Incidence rate = -------------------------------------------------------Sum of number of hours worked
Incidence rates for higher levels of industry detail are
produced using aggregated weighted and benchmarked
totals. Rates may be computed by industry, employment
size, geographic area, extent or outcome of case, number
of lost workdays, etc.
Reliability of Estimates
All estimates derived from a sample survey are sub
ject to sampling and nonsampling errors. Sampling errors
occur because observations are made on a sample, not
on the entire population. Estimates based on the different
possible samples of the same size and sample design could
differ. The relative standard errors, which are a measure
of the sampling error in the estimates, are calculated as
part of the survey’s estimation process. For the all-indus
try estimate of the total occupational injuries and illnesses
rate, the sample size is set to insure that a year-to-year
difference of 0.10 or more will be statistically significant
at the 95-percent confidence level. Target relative sam
pling errors for year-to-year changes in the total injury
and illness rate are also set for each industry. These
targets vary from 7 percent to 38 percent at the 95-percent
confidence level,with the average being 11 percent. Both
the estimates and the relative standard errors of the
estimates are published in the b l s annual bulletin
Occupational Injuries and Illnesses in the United States
by Industry.
Nonsampling errors in the estimates can be attributed
to many sources; e.g., inability to obtain information
about all cases in the sample, mistakes in recording or
coding the data, definitional difficulties, etc. To minimize
the nonsampling errors in the estimates, the completed
forms are edited and apparent inconsistencies are checked
with the employer. Even with careful editing, errors caused
by misinterpretation of definitions may not be uncovered.
For this reason, a quality assurance program is conducted
periodically to evaluate the extent of nonsampling errors
in the estimates. A sample of the participating establish
ments is visited by survey personnel. The entries on the
log and summary are compared with supplementary
records ( o s h a N o . 101) and other available information
to evaluate the reliability of the log entries which pro
vide the basic data for the annual survey reports.
Presentation
Each year, b l s publishes a bulletin covering national
results. Selected national data also are published in a news
release and periodically in Monthly Labor Review articles.
The data are also available on b l s data diskettes. The
data are published in safety and trade journals and in the
President’s Annual Report on Occupational Safety and
Health to the U.S. Congress.
In addition, State data on microfiche are available from
the National Technical Information Service, U.S. Depart
ment of Commerce, Springfield, v a 22161.
Uses and Limitations
National and State policymakers use the survey as an
indicator of the magnitude of occupational safety and
health problems, o s h a uses the statistics to help deter
mine which industries have the greatest need to improve
safety programs and to measure the effectiveness of the
act in reducing work-related injuries and illnesses.
Both labor and management use the estimates in
evaluating safety programs. Other users include insurance
carriers involved in workers’ compensation, industrial
hygienists, manufacturers of safety equipment, research
ers, and others concerned with job safety and health.
In terms of the recording and reporting of occupational
illnesses, the statistics generated through the annual sur
vey are a reliable measure of disease cases that are
unequivocally visible. However, in terms of statistical
validity, the data may be wanting because chronic and
long latent diseases, although not totally excluded, are
largely beyond the scope of the survey system. To this
extent, an undercount exists in the illness estimates. There
is, as yet, no reliable measure of that undercount. The
only other comprehensive source of occupational disease
statistics lies in State workers’ compensation records.
However, the same difficulties in establishing an occupa
tional link apply to workers’ compensation cases.
101
Part II.
Supplementary Data System
The Bureau of Labor Statistics’ Supplementary Data
System ( s d s ) is a comprehensive effort to standardize
occupational injury and illness data from State workers’
compensation information to achieve some degree of com
parability. The SDS data are unique in the detail available,
providing analysts with opportunities for more extensive
research than heretofore possible.
Background
While the annual survey program provided the infor
mation required by the Occupational Safety and Health
(OSH) Act of 1970, there was an increasing demand for
information about characteristics of the occupational
injuries and illnesses and the workers to whom they were
occurring. In 1973, in response to this demand, the Bureau
began testing the feasibility of collecting such informa
tion through contracts with States.
Records routinely generated by State workers’ compen
sation programs—employee and employer reports, medi
cal reports, compensation award records, etc.—were long
recognized as potentially valuable sources of information
about occupational injuries and illnesses. However, most
workers’ compensation agencies were primarily concerned
with administering claims systems, and were not particu
larly concerned with availability and accuracy of industry,
occupation, or injury and illness data. Additionally, States
processing such data had different coding systems, some
times with identical terms being defined differently.
States were urged to supply the desired information in
machine-readable form. However, the different classifica
tion systems and record formats resulted in noncom
parabilities and processing difficulties. The Bureau revised
the program to require participating States to use com
parable record formats and classifications.
In 1976, the current structure of the Supplementary
Data System was established, in cooperation with 27
States. The name was chosen from the role SDS plays of
providing supplementary information to the annual survey
of injuries and illnesses. Although the sds does not affect
the variations in coverage and reporting requirements
among States, it requires that participating States provide
prescribed data elements, and use specific classification
systems, standard record formats, and uniform procedures.
Description of SDS
The primary source of information for the SDS is a first
report of injury or illness, which employers and insurance
carriers submit to State workers’ compensation agencies.
All jurisdictions require such reports. There are four basic
types of information on the report. The first identifies
the employer and permits classification of the case by
industry and geographic location. The second lists
characteristics of the employee such as age, sex, salary,
and occupation. The third describes how the accident or
exposure occurred, any objects or substances involved, the
nature of the injury or illness, and the part of body
affected. The fourth provides information on the workers’
compensation carrier, possible disability, and other items
needed to process the claim. Under 50-50 grant funding
agreements, State agencies classify, code, and process the
information from the various workers’ compensation
reports. Since these are administrative reports which
employers, employees, or physicians must file under State
regulations, the information does not constitute an addi
tional burden on employers.
The prescribed data elements which must be uniformly
defined and submitted by all participating States are:
State code
Reference year
Case number
Year and month of occurrence
Occupation
Industry
Ownership (public or private industry)
Nature of injury or illness
Part of body affected
Source of injury or illness
Type of accident or exposure
Sex of employee
At their option, States may also submit other data ele
ments, such as duration of employment, extent of
disability, indemnity compensation, and medical costs,
some of which may be defined differently from State
to State. For example, “duration of employment” may
102
refer to time with an employer, in a particular occupa
tion, or in a particular job. The following optional items
as of 1986 may be submitted by participating States. (The
number in parentheses indicates the number of States pro
viding that information.)
Day of occurrence (32)
Hour of shift (5)
Associated object or substance (16)
Age of employee (31)
Duration of employment (20)
Weekly wages (22)
Extent of disability (13)
Kind of insurance (14)
Indemnity compensation (8)
Medical payments (6)
The indemnity and medical costs of a case are par
ticularly important optional items. Workers’ compensa
tion programs indemnify injured or ill workers with
income-replacing cash benefits. These payments are
awarded for fatalities, disfigurements, permanent disa
bilities, and for temporary disabilities which exceed some
specified number of days (that is, a waiting period).
Medical expenses associated with an injury or illness are
usually paid in full without a waiting period. Although
indemnity compensation and medical payments data are
useful economic and social indicators, and some measure
of severity, only a small number of the participating
States are able to provide these data.
Classification systems used by all States in the sd s
include: (1) the 1972 Standard Industrial Classification
Manual to code industry; (2) the 1980 Bureau o f the
Census Alphabetical Index o f Industries and Occupations
to code the occupation of the injured or ill employee; (3)
the American National Standards Institute Z16.2—1962
Method o f Recording Basic Facts Relating to the Nature
and Occurrence o f Work Injuries (with codes expanded
and modified by the Bureau) to classify the nature of the
injury or illness, the part of body affected, the source
of the injury or illness, and the type of accident or
exposure; and (4) a newly developed classification, the
associated object or substance, which provides additional
information about the factors associated with the injury
or illness.
The sd s requires close cooperation between the State
agencies and the Bureau. In order to achieve uniform
data, the Bureau establishes conceptual and operational
standards which are developed in consultation with the
State agencies. Federal/State cooperation is achieved
through specific actions and groups tailored toward
improving the s d s . For example, State coding is periodi
cally reviewed by regional and national office personnel
for uniformity among all States. Uniformity is also
achieved through State participation on the sd s Interpre
tations Committee, which resolves differences in coding
difficult cases, and the State task force on coding revi
sions, which is composed of nine State members and
reviews classification systems and coding practices with
the view to changing current procedures when necessary.
Presentation
sd s data are available from the National Technical
Information Service (n t is ). Beginning in 1979, these data
have included individual case records for 30 States
organized into multi-State files which make a large body
of data available at moderate cost on machine-readable
magnetic tapes. Information on the tabulations available
from each State can be obtained from the Office of
Occupational Safety and Health Statistics, Bureau of
Labor Statistics, Department of Labor.
Uses and Limitations
The Supplementary Data System provides valuable
information in three general areas: (1) defining workrelated safety and health problems for policymakers; (2)
guiding professional investigations and research; and (3)
making available information for the administration of
workers’ compensation programs. For example, The Re
port o f the National Commission on State Workmen’s
Compensation Laws suggested that systematic collection
and exchange of data would be a valuable source of infor
mation for both compensation and safety agencies.
The sd s is a step in this direction. Because sd s is a
machine-readable categorization of workers* compensa
tion information, a final product will be a State capability
to analyze its cases in considerable detail, including the
types of cases handled and the predominant types of
affected workers and work situations. The data direct
attention to problem areas which can be most effectively
handled by safety and health standards, training, or com
pliance programs.
Although the Supplementary Data System standardizes
the classification, processing, and tabulations of data, it
is not a complete census of occupational injuries and
illnesses; as of 1986, 33 States were participating. In addi
tion, coverage and reporting requirements variations
reflect differences in State workers’ compensation laws.
Differences also exist because of statutory and ad
ministrative variations in workers covered and reports
processed, and in the kinds of cases required to be
reported to workers’ compensation agencies.
Finally, occupational illness data from the SDS suffer
from the same low degree of identification as that
experienced in the annual survey of occupational injuries
and illnesses. Recognition of occupational illness depends
on the “ state of the art.” As medical knowledge in
creases, illness identification will improve in both data
collection systems.
103
Part III.
Work Injury Report Program
The Bureau of Labor Statistics’ Work Injury Report
(w ir ) survey program examines specific types of injuries
(and illnesses) in the work environment or focuses on
selected high-hazard jobs or industries to obtain infor
mation not available from the Annual Survey of Occupa
tional Injuries and Illnesses and the Supplementary Data
System. By surveying the injured worker directly, the w ir
introduces a unique perspective to the pool of injury data,
which has traditionally been provided by employers.
Background
The annual survey produces measures of the incidence
and severity of work-related injuries and illnesses, while
the SDS complements the annual survey by providing
information on selected general characteristics of the
injured worker as well as the injury. Neither program,
however, provides extensive inform ation on the
numerous, more detailed factors associated with certain
types of injuries. When the Occupational Safety and
Health Administration (o s h a ) indicated a need for this
type of data in 1978, BLS established the w ir program.
The o s h a safety and health program must address a
wide variety of topics, such as hazardous worksite con
ditions, the safe operation of tools or equipment, the use
of protective equipment, and special industry standards.
The w ir survey program provides o s h a with a broad
spectrum of support information by allowing flexibility
in both the survey subject matter and the data elements
collected. A w ir survey is able to identify patterns of
accident causes, as well as provide detail on such topics
as the activity of the worker at the time of accident, the
equipment used, the protective equipment worn, and the
training given for the work being done.
Because it would be difficult, if not impossible, for
employers to provide some of the needed information,
the decision was made to survey the injured worker
directly. For example, questions regarding worker activity
are so specific that, in the absence of witnesses, employers
would be required to question the injured workers.
Similarly, only the workers could explain why they did
or did not use protective equipment. Finally, by survey
ing the worker directly, it is possible to expand the scope
of questions on training and work experience to include
previous jobs held by the worker.
The Survey Process
Selection of the subject of a study is the first phase of
the survey process and is based on o s h a ’s assessment of
data requirements. The w ir program permits substantial
flexibility in subject matter because it uses the sd s source
document, the Employer’s First Report of Injury, from
which the following injury characteristics are available:
Nature of injury, part of body affected, source of injury,
and type of accident. Also identified are the industry and
the worker’s occupation. Any of the sd s data elements
can be selected for study, w ir surveys have focused on
specific industries (oil and gas drilling and well servic
ing, and logging), injuries to a particular part of the body
(back, hand, eye, face, and head injuries), selected types
of accidents (falls on stairs, falls from elevations, falls
from scaffolds, and falls from ladders), specific nature
of injuries (upper extremity amputations), and occupa
tions (construction laborers, warehouse workers). The
scope of a w ir survey, however, can be expanded beyond
the sd s classification categories. Three surveys have used
worker activity, which is not coded in s d s , as a selection
criterion (workers injured while welding, using power
saws, and servicing equipment). The range of the pro
gram is limited only by the amount of information pro
vided on employers’ first reports of injury and the
geographic and industrial representation of the par
ticipating States.
The heart of the program is a questionnaire tailored
to obtain information relevant to each area selected for
study by o s h a . Because development of the survey ques
tionnaire requires close interaction with OSHA, a w ir task
force was formed of representatives from the BLS Office
of Occupational Safety and Health Statistics and the
OSHA Offices of Compliance, Standards Development,
Training, and Regulatory Analysis. Also included on the
task force are representatives of the National Institute for
Occupational Safety and Health ( n io s h ), who provide
expertise in the area of occupational injury and illness
epidemiology. Because of the wide variety of survey
104
topics, each study has a unique questionnaire. All pro
posed survey questions are discussed by the task force
and, where possible, formatted with multiple choice
responses. Each questionnaire developed is also designed
to be brief in order to minimize the burden on respond
ents and encourage participation. In general, informa
tion is sought on how the injury occurred, the worker’s
activity and location, what hazardous conditions pre
vailed, the nature of the equipment involved, the safe
guards used, and the extent of related safety training.
Additionally, each questionnaire is tested on a small panel
of workers before it becomes final.
Current surveys employ systematic random sampling,
unless the expected population size is small enough to
warrant a complete census. Survey estimates are derived
using a Horwitz-Thompson estimator with a nonresponse
adjustment procedure. Based on a predicted population
size, the sample size is targeted so that any proportion
estimate based on the entire universe will have a sam
pling error no greater than plus or minus 5 percent at
the 2 sigma or 95-percent confidence level. Early w ir
surveys were based on a purposive sample of about
750 respondents.
The universe of potential respondents generally
includes all workers in the participating States who were
classified through the SDS as being injured or made ill
under the criteria established for the survey during the
selected time period. The time frame is usually limited
to 1 month, but may be longer if the expected popula
tion is very small or if seasonality is a concern.
Participating State agencies screen incoming first
reports of injury, using bls criteria for defining relevant
cases, to identify the target population of injured workers
to be surveyed. Excluded from the screening criteria for
each survey are cases where the industry classification is
coal and metallic and nonmetallic mining, or government,
because these industries are not regulated by o s h a . Also
excluded are cases involving fatalities or assaults.
A table o f sample selection numbers, generated by b l s
for each survey, determines which cases are to be sam
pled. If a case is selected for inclusion in the sample, a
questionnaire is mailed directly to the injured worker’s
home address.
Returned questionnaires are matched with the appro
priate first report o f injury forms, and the primary sd s
data elements, such as nature of injury, industry, occupa
tion, age, and sex, are coded for each case. All informa
tion which could identify a worker or an employer is
deleted from the questionnaire and first report by State
agency personnel. At the completion o f the data collec
tion phase, all returned questionnaires are transmitted to
BLS-Washington, along with refusals, nonmailables, and
Post Office returns.
Questionnaires are screened by BLS-Washington for
completion and consistency. Results of the survey are then
keypunched and mechanically edited. Finally, estimates
are generated and published in a bulletin, with an accom
panying text which highlights the findings.
Weighting and Estimation Procedures
Unless a census has been taken, the sample of injured
workers is weighted to account for all injured workers
within the scope of the survey in the participating
States. The weight assigned is the inverse of the probabil
ity of selection, and is applied to each sample member’s
response.
In each survey, a number of the selected injured workers
do not return the questionnaires. These workers are re
ferred to as unit nonrespondents. A weighting-class non
response adjustment procedure is used to reduce the
potential bias due to nonresponse in the estimates. In this
procedure, the sample is partitioned into cells, and a unit
nonresponse adjustment factor is calculated within each
cell. This procedure is based on the assumption that, with
in each cell, the distribution of the unit nonrespondents
would be the same as the distribution of the respondents.
To determine the set of cells for unit nonresponse
adjustment, a comparison of the following characteristics
is made between respondents and nonrespondents: Age,
sex, nature of injury, part of body affected, source of
injury, type of accident, industry, and occupation. If it
is determined that there are differences in the distribu
tion of a particular characteristic between respondents
and nonrespondents, a partition based on this charac
teristic is used to adjust for unit nonresponse.
In addition to workers not returning the questionnaire,
a small number of workers respond to the survey but do
not answer all of the questions. These are referred to as
item nonrespondents. To account for this type of non
response, it is assumed that the response distribution of
the item nonrespondents would be the same as the
response distribution of the item respondents.
For each question, a final weight for each respondent
is calculated as the product of the original weight, the
unit nonresponse factor, and the item nonresponse fac
tor. The estimate of the total number of in-scope injured
workers for each question is equal to the sum of the final
weights of the respondents. The estimate of the percent
of workers giving a particular answer to a question is the
sum of the final weights of the respondents giving a par
ticular answer divided by the estimate of the total number
of in-scope injured workers.
Reliability of Estimates
All estimates derived from a sample survey are sub
ject to sampling and nonsampling errors. Sampling errors
occur because observations are made on a sample, not
on the entire universe. Estimates based on the different
possible samples of the same size and sample design could
105
differ. The standard errors, which are a measure of the
sampling error in the estimates, are calculated as part of
the survey’s estimation process and are available upon
request.
Nonsampling errors in the estimates can be attributed
to many sources, e.g., inability to obtain information
about all cases in the sample, mistakes in recording or
coding the data, definitional difficulties, etc. To minimize
the nonsampling errors in the estimates, the completed
questionnaires are edited, and apparent inconsistencies
are checked.
Presentation
At the completion of each survey, the results are
tabulated and published along with an analysis of the
survey findings. Since 1979, 18 survey reports have
been published. The following reports are available from
the U.S. Department of Commerce, National Technical
Information Service ( n t is ), 5285 Port Royal Road,
Springfield, VA 22161:
Survey of Ladder Accidents Resulting in Injuries
Survey of Welding and Cutting Accidents Resulting
in Injuries
Survey of Scaffold Accidents Resulting in Injuries
Survey of Power Saw Accidents Resulting in Injuries
Accidents Involving Eye Injuries
Accidents Involving Face Injuries
Accidents Involving Head Injuries
Accidents Involving Foot Injuries
Injuries Related to Servicing Equipment
Back Injuries Associated with Lifting
Work-Related Hand Injuries and Upper Extremity
Amputations
Reports available to date from the U.S. Department of
Labor, Bureau of Labor Statistics, Office of Occupa
tional Safety and Health Statistics, Room 4014, 601 D
Street NW., Washington, DC 20212 include:
Injuries
Injuries
Injuries
Injuries
Injuries
Injuries
in Oil and Gas Drilling and Services
Resulting From Falls From Elevations
in the Logging Industry
Resulting From Falls on Stairs
to Construction Laborers
to Warehouse Workers
Uses and Limitations
Because the design of the w ir program allows for flex
ibility in both the types of surveys done and the kinds
of information collected, these surveys produce a broad
range of data on work-related accidents. The ways in
w hich the data can be used, as w ell as the p eop le w ho
use it, are as varied as the in fo rm a tio n itself, o s h a , o f
course, is the prim ary user o f w ir data.
w ir surveys can be used by o s h a in the development
or revision of safety standards and in the planning of
compliance strategy and training programs. In standards
setting, w ir data can be used to test the need for a par
ticular standard and to support individual requirements
of the standard. They can also be used to direct o s h a ’s
attention to an area where a problem may exist and to
assist in determining the corrective action that may need
to be taken.
Once standards are approved, OSHA is responsible for
their enforcement. Enforcement is carried out by com
pliance officers who inspect workplaces for adherence to
standards. Because w ir survey data can provide specific
information on how and why a particular type of acci
dent occurs, these data have proven a valuable tool in
training compliance officers to be aware of situations in
which there have been a large number of injuries.
In training and education, o sh a needs information for
targeting the workers who might benefit the most by
knowing the injury potential inherent in certain work
situations. Once these areas are identified, o s h a can
tailor educational programs to increase awareness of these
problems. Because of the detailed information provided
by w ir surveys, o s h a has been able to incorporate data
from several surveys into their educational programs.
Other data users include employers and safety officials
throughout private industry, government, and labor
organizations, as well as special interest groups such as
lawyers, consumer organizations, and manufacturers of
safety equipment. For example, n io s h uses w ir data in
conjunction with their testing procedures and special
studies, while State government agencies use the data to
evaluate standards specific to safety issues particular to
their States.
In spite of the unprecedented amount of information
provided through the w ir program, there are limitations
associated with the survey data. Data collection pro
cedures limit coverage, for the most part, to those States
which participate in the SDS program. For this reason,
the w ir program does not produce national estimates.
The program is also subject to differences in State
workers’ compensation reporting requirements. At the
present time, the program does not permit detailed com
parisons between injured workers and the rest of the
working population. For example, there are no data on
the use of protective equipment for workers who were
not injured. Lack of data on workers who were exposed
to the same hazards but not injured precludes the develop
ment of incidence rates as a measure of the relative risk
by activity, occupation, etc. In addition, the data reflect
the injury experience for a particular reference period
ranging from 1 to 6 months.
106
Technical References
Hilaski, Harvey J . ‘‘Understanding Statistics on Occupational
Illnesses,” Monthly Labor Review, March 1981.
Hilaski, Harvey J., and Wang, Chao Ling. “ How Valid Are
Estimates of Occupational Illnesses?” Monthly Labor
Review, August 1982.
McCaffrey, David. “ Work-Related Amputations by Type and
Prevalence,” M onthly Labor Review, March 1981.
Root, Norman, and McCaffrey, David. “ Providing More
Information on Work Injury and Illness,” Monthly
Labor Review, April 1978.
Root, Norman, and Hoefer, Michael. “ The First Work-Injury
Data Available From New bls Study,” Monthly Labor
Review, January 1979.
Root, Norman, and McCaffrey, David. “ Targeting Worker
Safety Programs: Weighing Incidence Against Expense,”
Monthly Labor Review, January 1980.
Root, Norman, and Daley, Judy. “ Are Women Safer
Workers? A New Look at the D ata,” Monthly Labor
Review, September 1980.
Root, Norman. “ Injuries at Work Are Fewer Among Older
Employees,” Monthly Labor Review, March 1981.
Root, Norman, and Sebastian, Deborah, “ b ls Develops
Measure of Job Risk by Occupation,” Monthly Labor
Review, October 1981.
Schauer, Lyle, and Ryder, Thomas. “ New Approach to
Occupational Safety and Health Statistics,” Monthly
Labor Review, March 1972.
U.S. Department of Labor, Bureau of Labor Statistics. Occu
pational Injuries and Illnesses in the United States by
Industry. Bulletin issued yearly.
Bureau of Labor Statistics. Recordkeeping Guidelines fo r
Occupational Injuries and Illnesses, September 1986
(Effective April 1986).
The official Department of Labor interpretation of
employer recordkeeping requirements under the Occu
pational Safety and Health Act of 1970 and 29 cfr
Part 1904. Provides supplemental instructions to the
recordkeeping forms, o s h a N os . 200 and 101.
Bureau of Labor Statistics. A Brief Guide to Recordkeeping
Requirements fo r Injuries and Illnesses, June 1986 (Effec
tive April 1986).
An abbreviated version of the guidelines to be used
as a ready reference, and as an aid to plant managers
and employers of small-size establishments or firms.
107
Chart 1. Guide to recordability of cases under the
Occupational Safety and Health Act
108
U.S. Department of Labor
Bureau of Labor Statistics
Log and Summary of Occupational
Injuries and Illnesses
NOTE:
This form is required by Public Law 91-586 and must ba leapt
in tha establishment for 5 years. FaHura to maintain and post
can rasult in tha issuance of citations and asaaasmant of penalties.
(Sea potting requirements on the other tide of form.)
Casa or
Date of
Number
Onset of
Illness
Enters
Enter
nondupli- Mo./day.
eating
number
Employes's Nama
Enter first name or initial,
middle initial, last name.
will
facilitate
Occupation
Enter regular job title, not
activity employee was performing when injured or at
onset of illness. In the aosence
of a formal title, enter a brief
description of the employee's
duties.
Enter department in which
the employee is regularly
employed or a description
■upptaTypical entries for this column might be:
Strain of lower back Contact dermatitis
(A)
(B)
(C)
(D)
(E)
Establishment Address'
[Extent of and Outcome of INJURY
<F)
Type, Extent of, and Outcome of ILLNESS
[Fatalities
j
Nonfatal Injuries
[Injury
■Related
Injuries With Lost Workdays
I
|
i
I
(Enter DATE
lof death.
ment at the time of injury
or illness.
periaons
Establishment Nome
Enter a brief description of the injury or illness
and indicate the part or parts of body affected-
which employee is assigned,
even though temporarily
Form Approvod
O.M.Q. No. 1220-0029
Company Name
RECORDABLE CASES: You are required to record information about every occupa
tional death; every nonfatal occupational Ulnass; and those nonfatal occupational in
juries which involve one or more of the following; loss of consciousness, restriction
of work or motion, transfer to another job, or medical treatment (other than first aid).
(See definitions on the other tide of form.)
Department
Description of Injury or Illness
Enter a
CHECK if
injury In-
Injuries
Without Lost
Workdays
Enter num Enter num
ber of
ber of
DAYS ewey DAYS of
daysawey away from
from
work ectiv-
days of
restricted
activity,
or both.
(2)
I
r
Type of Illness
Fatalities
CHECK Only One Column for Each Illness
(See other tide of form for terminations
or permanent transfora.
Illness
Related
Enter a CHECK
if no entry was
made in colbut the injury
as defined
above.
|
5
I I
(3)
(4)
(5)
1
51 8
\
1 |» f i t
If i
i! il
|
(a)
| l
(7)
(d)
is)
£
(6)
(b)
(C)
* ,
8
Iit
is
81
l 1
Illnesses With Lost Workdays
Enter D ATE Enter a Enters
CHECK CHECK if
of death.
If Illness illness indays away away from
from
Illnesses
Without Lost
Entor num Entor num Enter a CHECK
if no entry was
ber of
ber of
made In colDAYS ewey DAYS of
restricted
v/ork cctfv-
days of
restricted
II
(8)
(f)
Nonfatal lllnocses
activity,
or both.
(9)
(10)
(ID
(12)
(13)
(0)
I
----- 1----------------I
I
I
*
I
I
_________________________________ !__________
'
5^
i
i
........I
i
i
--------------------------------------------------------------------1----------------I
i
--------------------------------------------------------------------1-----------------
i
i
--------------------------------------------------------------------1
I
i
I
-------------------------------------------------------------------- 1—
i
i
-------------------------------------------------------------------- 1—
I
___________ I
I
Certification of Annual Summary Totals By
OSHA No. 200
9
S
OSHA No. 200
POST ONLY THIS PORTION OF THE LAST PAGE NO LATER THAN FEBRUARY 1.
Date
I
I
Instructions for OS H A No. 200
I.
Log and Summary of Occupational Injuries and Illnesses
Column B -
Each employer who is subject to the recordkeeping requirements of the
Occupational Safety and Health Act of 1970 must maintain for each estab
lishment a log of all recordable occupational injuries and illnesses. This
form (O SH A No. 200) may be used for that purpose. A substitute for the
O S H A No. 200 is acceptable if it is as detailed, easily readable, and under
standable as the O S H A No. 200.
Enter each recordable case on the log within six (6) workdays after learn
ing of its occurrence. Although other records must be maintained at the
establishment to which they refer, it is possible to prepare and maintain
the log at another location, using data processing equipment if desired. If
the log is prepared elsewhere, a copy updated to within 45 calendar days
must be present at all times in the establishment.
Logs must be maintained and retained for five (5) years following the end
of the calendar year to which they relate. Logs must be available (normally
at the establishment) for inspection and copying by representatives of the
Department of Labor, or the Department of Health and Human Services,
or States accorded jurisdiction under the Act. Access to the log is also
provided to employees, former employees and their representatives.
II.
Changes in Extent of or Outcome of Injury or Illness
If, during the 5-year period the log must be retained, there is a change in
an extent and outcome of an injury or illness which affects entries in
columns 1. 2, 6, 8, 9, or 13, the first entry should be lined out and a new
entry made. For example, if an injured employee at first required only
medical treatment but later lost workdays away from work, the check in
column 6 should be lined out. and checks entered in columns 2 and 3 and
the number of lost workdays entered in column 4.
Columns
1 and 8
-
IN JU R Y OR IL L N E S S -R E L A T E D D E A T H S .
Self-explanatory.
Columns
2 and 9
-
IN JU R IE S OR ILLN ES S ES W IT H L O S T W O R K D A Y S .
Self-explanatory.
Posting Requirements
Even though there were no injuries or illnesses during the year, zeros must
be entered on the totals line, and the form posted.
The person responsible for the annual summary totals shall certify that the
totals are true and complete by signing at the bottom of the form.
IV.
Instructions for Completing Log and Summary of Occupational Injuries
and Illnesses
Column A
-
CA SE O R F IL E N U M B ER . Sal E xp la n a tory.
Columns 7a
through 7g -
V.
Columns
3 and 10
-
IN JU R IES O R ILLN ES S ES IN V O L V IN G D A Y S A W A Y
FROM W ORK . Self-explanatory.
Columns
4 and 11
-
L O S T W O R K D A Y S ------D A Y S A W ^ Y F R O M W O R K .
Enter the number of workdays (consecutive or not) on
which the employee would have worked but could not be
cause of occupational injury or illness. The number of lost
workdays should not include the day of injury or onset of
illness or any days on which the employee would not have
worked even though able to work.
N O T E : For employees not having a regularly scheduled
shift, such as certain truck drivers, construction workers,
farm labor, casuai labor, part-time employees, etc., it may
be necessary to estimate the number of lost workdays. Esti
mates of lost workdays shall be based on prior work history
of the employee A N D days worked by employees, not ill or
injured, working in the department and/or occupation of
the ill or injured employee.
Columns
5 and 12
-
L O S T W O R K D A Y S — D A Y S O F R E S T R IC T E D W O R K
A C T I V IT Y .
Enter the number of workdays (consecutive or not) on
which because of injury or illness:
(1) the employee was assigned to another job on a tem
porary basis, or
(2)
the employee worked at a permanent job less than
full time, or
(3)
the employee worked at a permanently assigned job
but could not perform all duties normally connected
with it.
The number of lost workdays should not include the day of
injury or onset of illness or any days on which the employ
ee would not have worked even though able to work.
Poisoning (Systemic Effect of Toxic Materials)
Examples: Poisoning by lead, mercury, cadmium, arsenic, or
other metals; poisoning by carbon monoxide, hydrogen sulfide,
or other gases; poisoning by benzol, carbon tetrachloride, or
other organic solvents; poisoning by insecticide sprays such as
parathion, lead arsenate; poisoning by other chemicals such as
formaldehyde, plastics, and resins; etc.
7e.
Disorders Due to Physical Agents (O ther than Toxic Materials)
Examples: Heatstroke, sunstroke, heat exhaustion, and other
effects of environmental heat; freezing, frostbite, and effects of
exposure to low temperatures; caisson disease; effects of ionizing
radiation (isotopes. X-rays, radium); effects of nonionizing radia
tion (welding flash, ultraviolet rays, microwaves, sunburn); etc.
7f.
Disorders Associated With Repeated Trauma
Examples: Noise-induced hearing loss; synovitis, tenosynovitis,
and bursitis. Raynaud's phenomena; and other conditions due to
repeated motion, vibration, or pressure.
7g
All Other Occupational Illnesses
Examples: Anthrax, brucellosis, infectious hepatitis, malignant
and benign tumors, food poisoning, histoplasmosis, coccidioido
mycosis. etc.
T Y P E O F ILLN ES S .
Enter a check in only one column for each illness.
Totals
Ad d number of entries in columns 1 and 8.
Add number of checks in columns 2, 3, 6, 7, 9, 10, and 13.
Add number of days in columns 4, 5. 11, and 12.
Yearly totals for each coiumn (1-13) are required for posting. Running or
page totals may be generated at the discretion of the employer.
If an employee's loss of workdays is continuing at the time the totals are
summarized, estimate the number of future workdays the employee will
lose and add that estimate to the workdays already lost and include this
figure in the annual totals. No further entries are to be made with respect
to such cases in the next year's log.
V I.
7d.
IN JU R IE S O R ILLN E S S E S W I T H O U T L O S T
W O R K D A Y S . Self-explanatory.
T E R M IN A T IO N O R P E R M A N E N T T R A N S F E R -P la c e an asterisk to
the right of the entry in columns 7a through 7g (type of illness) which
represented a termination of employment or permanent transfer.
An y injury which involves days away from work, or days of
restricted work activity, or both must be recorded since it
always involves one or more of the criteria for recordability.
The entire entry for an injury or illness should be lined out if later found
to be nonrecordable.
For example: an injury which is later deter
mined not to be work related, or which was initially thought to involve
medical treatment but later was determined to have involved only first aid.
A copy of the totals and information following the fold line of the last
page for the year must be posted at each establishment in the place or
places where notices to employees are customarily posted. This copy must
be posted no later than February 1 and must remain in place until March 1.
Columns
6
and 13 -
Columns
C through F - Self-explanatory.
In another example, if an employee with an occupational illness lost work
days, returned to w ork, and then died of the illness, any entries in columns
9 through 12 should be lined out and the date of death entered in column 8.
III.
D A T E O F IN JU R Y O R O N S E T O F ILLN ES S .
For occupational injuries, enter the date of the work acci
dent which resulted in injury. For occupational illnesses,
enter the date of initial diagnosis of illness, or, if absence
from work occurred before diagnosis, enter the first day of
the absence attributable to the illness which was later diag
nosed or recognized.
Definitions
O C C U P A T IO N A L IN JU R Y is any injury such as a cut, fracture, sprain,
amputation, etc., which results from a work accident or from an expo
sure involving a single incident in the work environment.
N O T E : Conditions resulting from animal bites, such as insect or snake
bites or from one-time exposure to chemicals, are considered to be injuries.
O C C U P A T IO N A L IL L N E S S of an employee is any abnormal condition or
disorder, other than one resulting from an occupational injury, caused by
exposure to environmental factors associated with employment. It in
cludes acute and chronic illnesses or diseases which may be caused by in
halation, absorption, ingest'on, or direct contact.
The following listing gives the categories of occupational illnesses and dis
orders that will be utilized for the purpose of classifying recordable ill
nesses. For purposes of information, examples of each category are given.
These are typical examples, however, and are not to be considered the
complete listing of the types of illnesses and disorders that are to be count
ed under each category.
M E D IC A L T R E A T M E N T includes treatment (other than first aid) admin
istered by a physician or by registered professional personnel under the
standing orders of a physician. Medical treatment does N O T indude firstaid treatment (one-time treatment and subsequent observation of minor
scratches, cuts, burns, splinters, and so forth, which do not ordinarily re
quire medical care) even though provided by a physician or registered
professional personnel.
E S T A B L IS H M E N T : A single physical location where business is conduct
ed or where services or industrial operations are performed (for example,
a factory, mill, store, hotel, restaurant, movie theater, farm, ranch, bank,
sales office, warehouse, or central administrative office). Where distinctly
separate activities are performed at a single physical location,such as con
struction activities operated from the same physical location as a lumber
yard, each activity shall be treated as a separate establishment.
For firms engaged in activities which may be physically dispersed, such as
agriculture; construction; transportation; communications; and electric,
gas, and sanitary services, records may be maintained at a place to which
employees report each day.
7a.
Occupational Skin Diseases or Disorders
Examples: Contact dermatitis, eczema, or rash caused by pri
mary irritants and sensitizers or poisonous plants; oil acne;
chrome ulcers; chemical bums or inflammations; etc.
Records for personnel who do not primarily report or work at a single
establishment, such as traveling salesmen, technicians, engineers, etc., shall
be maintained at the location from which they are paid or the base from
which personnel operate to carry out their activities
7b.
Dust Diseases of the Lungs (Pneumoconioses)
Examples: Silicosis, asbestosis and other asbestos-related dis
eases, coal worker's pneumoconiosis, byssinosis, siderosis, and
other pneumoconioses.
materials processed or used, and the kinds of operations performed in the
course of an employee’s work, whether on or off the employer's premises.
7c.
Respiratory Conditions Due to Toxic Agents
Examples: Pneumonitis, pharyngitis, rhinitis or acute conges
tion due to chemicals, dusts, gases, or fumes; farmer's lung; etc.
W O R K E N V IR O N M E N T is comprised o f the physical location, equipment
SUPPLEMENTARY RECORD OF OCCUPATIONAL
INJURIES AND ILLNESSES
U .S . D e p a r t m e n t o f L a b o r
Cat* o r F ila N o .
F o rm Ap p ro ve d
Failure to maintain can raault in tha issuance of citations and assassnMmt of penalties.
O .M .B N o. 1220 0029
Emetayar
suitable report is made fo r other purposes, th is fo rm (O SHA No. 101) m ay be used o r the necessary facts can
listed on a separate plain sheet o f paper. These records m ust also be available in th e establishm ent w ith o u t delay and
at reasonable tim es fo r exam ination b y representatives o f the D epartm ent o f Labor and the D epartm ent o f Health
2. M ail address (N o . a n d street. c ity o r tow n . State, a n d zip coda)
3.
T o supplem ent the Log and Sum m ary o f O ccupational Injuries and Illnesses (O S H A No. 20 0 ), each establishm ent
m ust m aintain a record o f each recordable occupational in ju ry o r illness. W orker's com pensation, insurance,
o ther reports are acceptable as records if th e y contain all facts listed below o r are supplem ented t o d o so. I f
1. Nam e
? § s
B u r e a u o f L a b o r S t a t is t ic s
S u p p le m e n t a r y R e c o r d o f
O c c u p a t io n a l I n ju r ie s a n d Illn e sse s
L o ca tion , if different from mail address
and Human Services, and States accorded ju ris d ic tio n under th e A c t. The records m ust be m aintained fo r a period
Injured or III Employee
o f n o t less than five years fo llo w in g the end o f the calendar year to w h ich th e y relate.
4.
Nam e (F irs t, m id d le , a n d last)
5
H o m e address (N o . a n d s tre e t c it y o r to w n . State, a n d z ip code )
6.
Age
Social S ecurity N o.
_ L-L...1 M
7. Sex
□
(Check one)
i l l
□
8. O ccu p a tio n (E n t e r regular jo b title, not the specific a c tivity he was pe rfo rm in g a t tim e o f in ju ry .)
9.
Such records must contain at least the fo llo w in g facts:
1 ) A b o u t the em plo yer—name, m ail address, and lo cation if d iffe re n t fro m m ail address.
2) A b o u t the in ju re d o r i l l em ployee—name, social security num ber, hom e address, age, sex, occup ation, and depart
Departm ent (E n t e r nam e o f dep artm e n t o r division in w hich the inju re d person is regularly em p loye d, even thou gh he m a y have been te m p ora rily
ment.
w o rk in g in a no th er d e p artm e n t a t the tim e o f in ju ry .)
3) A b o u t the accident o r exposure to occup ational illness—place o f accident o r exposure, w h eth er it was on em
ployer's premises, w h a t the employee was doing w hen in jured, and how th e accident occurred.
The Accident or Expoeure to Occupational Illness
If accident or e xposure occurred on e m p loye r's premises, give address of plant or establishment in w h ic h it o ccurred. D o not indicate depa rtm e n t or
division w ith in the pla n t or establishment. If accident occurred outside e m p loye r’s premises at an identifiable address, give that address If it occurred on
a p u b lic highw ay or at any other place w h ic h cannot be identified b y num ber and street, please provide place references locating the place of in jury as
accurately as possible.
4) A b o u t the occupational in ju r y o r illness—description o f th e in ju ry o r illness, including pa rt o f b o d y affected;
name o f the object or substance w h ich d ire c tly injured the em ployee; and date o f in ju ry o r diagnosis o f illness.
10. Place of accident or exposure (N o . a n d street, c it y o r tow n. State, a n d zip code)
5) O th e r-n a m e and address o f physician; if hospitalized, name and address o f hospital; date o f rep o rt; and .name
11 Was place o f accident or exposure on e m p loye r's premises?
Yes I
l
No □
______________________________________________________________
12. W hat was the e m ployee d o in g when injured? (B e specific. I f he was using tools o r equipm ent o r ha ndlin g material, name them a n d te ll w h a t he was
d oing w ith th e m .)
13. H o w did the accident occur? (D escribe f u lly the events w hich resulted in the in ju ry o r o ccupational illness. T e ll w h a t happ e n e d a n d h o w i t happened.
N a m e a n y objects o r substances in v o lv e d a n d tell h o w they were involved. G ive fu ll details on a ll factors w h ich le d o r c o n trib u te d to the a cc ide nt
Use separate sheet fo r rd d itio n a i space.)
Occupational Injury or Occupational Illness
14. Describe the in ju ry o r illness in detail and indicate the part of b o dy affected. (E .g ., am putation o f rig h t inde x finger a t second j o i n t ; fracture o f ribs;
le ad p o is o n in g ; d erm atitis o f le ft hand, etc.)
15. N am e the object or substance w h ic h d ire c tly injured the employee. (F o r example, the machine o r thin g he struck against o r w h ich struck h im ; the
vapor o r p o is o n h e inh a le d o r sw a llow e d ; the c he m ica l o r radiation w h ich irriatated his skin; o r in cases o f strains, hernias, etc., the th in g he was
liftin g, p u lling , e tc .)
16. Date o f in jury or in'tiai diagnosis o f o ccupational illness
17. Did e m p loye e die? ( Check o ne )
18. Nam e and address o f physician
19. If hospitalized, nam e and address of hospital
Date of report
Prepared b y
O S H A N o . 101 (F e b . 1 981)
Official position
□
and position o f person preparing the report.
SEE D E F IN IT IO N S ON TH E B A C K O F O S H A FO RM 200.
1986 OSH A No. 200-S
Annual Occupational Injuries and Illnesses Survey Covering Calendar Year 1986
B ureau of L a b o r Statistics for the O ccu p a tio n a l Safe ty and Health A d m in is tra tio n
T H IS R E P O R T IS M A N D A T O R Y U N D E R P U B L IC L A W 91 59 6 . F A IL U R E T O R E P O R T
O .M .B . N o . 1 2 2 0 -0 0 4 5
A p p ro v al E x p . 6/30/87
C A N R E S U L T IN T H E IS S U A N C E O F C IT A T IO N S A N D A S S E S S M E N T O F P E N A L T IE S .
T h e i n f o r m a t i o n c o lle c t e d o n t h i j f o r m w i l l be u se d f o r s ta tis tic a l p u r
p o s e s o n ly b y th e B L S . O S H A , a n d t h e c o o p e r a t i n g S ta t e A g e n c ie s .
St.
Sch N o.
c lT
Suf
~
C o m p le te this re p o rt w h e th e r o r n o t there were
recordable o c cu p atio n a l in juries o r illnesses.
SIC
II. T O T A L HOURS
W ORKED IN 1986
E n te r the average n u m be r
E n te r the total n u m be r
of em ployees w h o w o rk e d
d u rin g calendar year 1986
in the establishm entls)
co vered b y this report. In
c lu d e all classes of e m p lo y
ees: full tim e, part tim e,
seasonal, te m p ora ry, etc.
See the instructions for an
exam ple of an annual aver
of hours a ctually w o rk e d
d u rin g 19 86 b y all e m
ployees covered b y this
re p ort D O N O T include
a n y n o n -w o rk tim e even
th ou gh paid such as va ca
tions, sick leave, etc. If
em ployees w o rk e d low
ho urs in 19 86 d ue to la y
age e m p lo ym e n t calcula
tion (Round to the
offs. strikes, fires, etc ,
e xpla in u n der C o m m e n ts
nearest whole number.)
(section V I I ) . (Round to
the nearest whole number.)
EMPLOYER S COPY
DO NOT RETURN
P L E A S E R E A D T H E E N C L O S E D IN S T R U C T IO N S
EDIT
A N N U A L AVER AG E
EM PLOYM ENT IN 1986
U .S . D e p a r t m e n t o f L a b o r
III. N A TU R E OF BUSINESS IN 1986
A . C heck the b o x w h ich
best describes the general
ty p e of a ctivity perfo rm e d
b y the establishm entls)
included in this report.
□
□
□
A g ric u ltu re
F o re stry
Fishing
□
Mining
□
O
□
□
□
□
□
□
□
□
□
□
C o n s tru ctio n
M an u fa cturin g
Tra n s p o rta tio n
C o m m u n ic a tio n
P u blic U tilities
Wholesale Tra d e
Retail Tra d e
Fin a nce
Insurance
Real Estate
Services
P u blic A d m in is tra tio n
REPORT LO C A TIO N AN D ID E N TIFIC A TIO N
B. E n te r in order of im
porta n ce the principal
p ro d u cts, lines of trade,
services or other a ctivi
ties. F o r each e n try also
include the ap proxim ate
percent o l total 19 86
annual value of p ro du c
tio n . sales or receipts
IV. M O N TH OF OSHA
INSPECTION
C If this re p ort in
cludes a n y establish
m e n tis ) w h ic h per
fo rm services for
other u nits of y o u r
c o m p a n y , indicate
the p rim a ry ty p e of
service or s up p ort
p ro vid e d (Check
as many as apply.)
1. □ C entral
a d m in istra tio n
2. D Research, d evelop
m e n t a nd testing
3. □ Storage
(w areho use)
4 □ O th e r (sp e cify)
V. R ECO R D AB LE
IN JUR IES AN D
ILLNESSES
If the e stablishm entls)
covered b y this report
had e ither a Federal o r
S tate O S H A co m plian ce
inspection d u rin g ca l
endar year 1 986,
please enter the nam e
of the m o n th in w h ich
the first inspection
occurre d .
IL u h this
box blank.)
D id the estab
lishm e n tls) have
a ny recordable
injuries or ill
nesses d u rin g
calendar year
1986?
1. □
N o (Please
co m ple te
section V II.)
2.
D Ye s (Please
co m ple te
sections V I
and V I I )
SEE R E V E R S E D
Please indicate a ny address changes b e lo w .
C om p le te this re p ort for the e stablishm entls) covered b y the d escriptio n b e lo w
For Information Call:
O S H A N o . 2 0 0-S (R evised D ecem ber 1 9 86 )
V I . O C C U P A T I O N A L I N J U R Y A N D I L L N E S S S U M M A R Y ( C o v e rin g C a le n d a r Y e a r 1 9 8 6 )
• Complete this section by copying totals from the annual summary o f your 1906 OSHA No. 200.
#
•
•
•
R e m e m b er to reverse the ca rb o n insert before c o m p le tin g this side.
Leave section V I blank if there w ere n o O S H A recordable injuries o r illnesses d u rin g 1 986.
OCCUPATIONAL ILLNESS CASES
OCCUPATIONAL INJURY CASES
IN JU R Y
IN JUR IES W ITH LO S T W O R KD AYS
R ELATED
FA TA L
ITIE S ”
(D EA TH S )
Injury caaea In ju ry
T o ta l
with day■
caaea
days
away from w ith daya
away
work and/or aw ay
fro m
w ork
reetrlcted
fro m
work daya
w o rk
INJURIES
W ITH O U T
LOST
W ORK
DAYS*
ILLNESS ILLNESSES
R E LA TE D
FATAL
ITIES**
(D EATH S )
lllneaa caaea
with daya
away from
work and/or
reetrlcted
workdeya
TYP E O F ILLNESS
Entar tha number of chacks
from tha appropriate columns
of tha log (OSHA No. 200).
T o ta l
daya o f
rettrlcted
activity
N u m b e r of
D EA TH S
in co l. 1
o f the log
(O S H A
No. 200)
N u m b e r of
CHECKS
in col. 2
of the log
(O S H A
N o . 2 0 0)
N u m b e r of
CHECKS
in co l. 3
of the log
(O S H A
No. 200)
S u m of
the D A Y S
in col. 4
of the log
(O S H A
N o. 200)
S u m of
the D A Y S
in co l. 5
of the log
(O S H A
No. 200)
N u m b e r of
CHECKS
in col. 6
of the log
(O S H A
N o. 200)
(1 )
(2 )
(3 )
(4)
(5 )
(6 )
If
15
(a)
(b )
(c)
DEATHS
Id)
(e)
(f)
(g)
: DAYS LOS' FBSULTING \
: t i c \ . t p a '.S c ER-tc a \ c ~ -
work
Total
daya
away
from
work
Total
days of
rettrlcted
activity
N u m b e r of
CHECKS
in co l. 9
o f the log
(O S H A
N o . 200)
N u m b e r of
CHECKS
in co l. 10
of the log
(O S H A
No. 200)
S u m of
the D A Y S
in co l 11
o f the log
(O S H A
No. 200)
S u m of
the D A Y S
in col. 12
of the log
(O S H A
N o 200)
N u m b e r of
CHECKS
in col. 13
of the log
(O S H A
No. 200)
(B)
(9 )
(10)
(11)
(12)
(13)
■■
■i■i
• I F Y O U L I S T E D F A T A L I T I E S IN C O L U M N S (1) A N D / O R (8), P L E A S E G I V E A B R I E F
D E S C R I P T I O N O F T H E O B J E C T O R E V E N T W H I C H C A U S E D E A C H F A T A L I T Y IN
T H E " C O M M E N T S " SE CTIO N BELOW .
N A M E ___________________________________________
T I T L E ___________________________________________
SIGN ATUR E .
PHONE
D A TE
lllneaa
caaea
with daya
ewey
ILLNESSES
W ITH O U T
LO ST
WORK
DAYS*
: V 'V O CIFIST AID
R E P O R T P R E P A R E D B Y (P le a se t y p e o r p r in t)
AR EA CODE
W ITH LOST W ORKDAYS
N om b e r of
D EA TH S
in col. 8
of the log
(O S H A
N o. 200)
DEATHS
•’- S — C A S E j '. . 1
V II.
#
Note: First aid even when administered by a doctor or nurse is n ot recordable.
Please check y o u r figures to be certain th at the sum o f entries in co lu m n s (7 a ) + (7 b )
+ (7 c ) + (7 d ) + (7 e ) + (7 f ) + (7 g ) = the sum o f entries in co lu m n s (8 ) + (9 ) + (1 3 ).
If y o u listed fatalities in co lu m n s (1 ) and/or (8 ) . please give a b rief d e scrip tio n o f
the object o r event w h ic h caused each fa tality in the " C o m m e n ts " section.
112
SURVEY REPORTING REGULATIONS
SECTION I.
T it le 2 9 , Part 1 9 0 4 .2 0 -2 2 o f the C o d e o f Federal Regulations requires that:
each e m p lo ye r shall re turn the co m ple te d survey fo rm , O S H A N o . 2 0 0 -S , w ith in
3 weeks o f receipt in accordance w it h the instru ctio ns show n b elo w .
than one establishm ent is included in this report, add together the annual average e m p lo y
m e n t for each establishm ent and enter the sum In clu d e all classes of em p lo yee s— seasonal,
INSTRUCTIONS FOR COMPLETING THE OSHA NO. 200-S FORM
1986 OCCUPATIONAL INJURIES A ND ILLNESSES SURVEY
(Covering Calendar Year 1986)
te m p o ra ry, adm inistrative, supervisory, clerical, professional, technical, sales, d e live ry, in
stallation, co n s tru ctio n and service personnel, as w ell as operators and related w orkers.
Chane* of Ownerdlip— W hen there has been a change of o w n e rship d u rin g the report period,
o n ly the records o f the cu rre n t o w n e r are to be entered in the report. E x p la in fu lly under
C o m m e n ts (S e ctio n V I I ) , and include the date o f the ow n e rship change a nd the tim e period
this re p ort covers.
Partial-Year Reporting— F o r any establishm ent(s) w h ic h was n o t in existence for the entire
re p ort year, the re port s hould cover the p o rtio n o f the pe riod d u rin g w h ich the establish
m e n t(s ) was in existence. E x p la in fu lly u n de r C o m m e n ts (S e ctio n V I I ) , inclu d ing the tim e
period this re port covers.
ESTABLISHMENTS INCLUDED IN THE REPORT
T h is report should include o n ly those establishm ents located in, o r identified b y , the Re p ort
Lo ca tio n and Id e ntification designation w h ich appears n e xt to y o u r m ailing address Th is
designation m a y be a geographical area, usually a c o u n ty o r c ity , o r it co u ld be a brief de
scrip tion o f y o u r op e ratio n w ith in a geographical area. If y o u have any questions conce rnin g
the coverage o f this re p o rt, please co nta ct the agency identified o n the O S H A N o . 20 0-S
re port form .
s to re , h o te l, r e s ta u r a n t, m o v ie th e a tr e ,
fa rm ,
( F o r e x a m p le :
rs n c h , b a n k ,
A n n u a l Average e m p lo ym e n t should be c o m p u te d b y sum m in g the e m p lo ym e n t fro m all
pay periods d u rin g 1 9 8 6 and then d ivid in g that sum b y the total n u m be r o f such pay periods
th ro u g h o u t the e n tire ye a r, in clu d ing periods w ith no e m p lo ym e n t. F o r e xam ple, if y o u had
the fo llow in g m o n th ly e m p lo y m e n t— Ja n .-1 0 ; F e b .-1 0 , M a r.-1 0 ; A p r -5 ; M ay 5; Ju n e 5,
J u ly -5 , A u g .-O ; Sept.-O . O c t.-O ; N o v .-5; D e c -5 — yo u w o u ld sum the n u m b e r o f em ployees
for each m o n th ly pay period (in this case, 6 0 ) and then d ivid e that total b y 12 (the n u m be r
of pay periods d u rin g the ye a r) to de rive an annual average e m p lo ym e n t o f 5.
SECTION II.
TO TA L HOURS WORKED IN 1986
E n te r in Section II the total n u m b e r of ho urs actually worked b y all classes o f em ployees
d u rin g 1986. Be sure to include O N L Y tim e on d u ty OO N O T include any non-work time
even though paid, such as vacations, sick leave, ho lid a ys, etc. T h e hours w orke d figure should
be obtained from p a yroll o r other tim e records w herever possible, if hours worked are not
m aintained separately from hour* paid, please enter y o u r best estimate. If actual hours
w orke d are not available for em ployees paid on co m m issio n , salary, b y the m ile , e tc., hours
w orke d m a y be estim ated o n the basis o f scheduled hours o r 8 hours per w o rk d a y.
F o r e xam ple, if a g rou p o f 10 salaried em ployees w o rke d an average o f 8 hours per d a y , 5
days a w eek, for 5 0 weeks o f the report p e riod , the total hours w orke d for this g ro u p w o u ld
b e 1 0 x 8 x 5 x 5 0 = 2 0 ,0 0 0 hours for the re port period.
D EFIN ITIO N OF ESTABLISH M EN T
A n E S T A B L IS H M E N T is d e fin e d as a s in g le p h y s ic a l l o c a tio n w h e re b u s in e s s is c o n d u c te d
o r w h e re s e rv ic e s o r in d u s tr ia l o p e r a tio n s a re p e r fo r m e d .
ANNUAL AVERAGE EMPLOYMENT IN 1986
E n te r in Section I the average (n o t the to ta l) n u m be r of full and p a rt-tim e em ployees w h o
w o rke d d u rin g calendar year 1 9 86 in the establishm ent(s) included in this report If m ore
a fa c t o r y , m ill,
SECTION III.
sales o f fic e , w a re h o u s e , o r
c e n tr a l a d m in is tr a tiv e o f f ic e . )
NATURE OF BUSINESS IN 1986
In o rd er to verify the nature o f business code, w e m ust have in form ation ab ou t the specific
e con o m ic a ctivity carried on b y the establishm ent(s) included in y o u r report d u rin g calendar
year 1986.
F o r fir m s e n ga ge d in a c t iv itie s s u c h as c o n s tr u c t io n , tr a n s p o r t a tio n , c o m m u n ic a tio n , o r
e le c tr ic , gas a n d s a n ita r y s e rv ic e s , w h ic h m a y b e p h y s ic a lly d is p e rs e d , r e p o r ts s h o u ld co v e r
th e p la c e t o w h ic h e m p lo y e e s n o r m a lly r e p o r t e a c h d a y
C om p le te Parts A , B a n d C as indicated in Section III on the O S H A N o 2 0 0-S fo rm . C om p le te
Part C only if sup p ortin g services are p ro vid e d to other establishments o f your co m p a n y.
Leave Part C blank if a) sup p ortin g services are n ot the p rim a ry fu n ctio n of a ny establish
m e n t s ) included in this report or b ) sup p ortin g services are pro vid e d b u t o n ly on a contract or
fee baiis fo r the general p ub lic o r fo r other business firms, (instructions continued on page 2.)
R e p o r ts f o r p e r s o n n e l w h o d o n o t p r i m a r il y r e p o r t o r w o r k a t a s in g le e s ta b lis h m e n t, su ch
as tr a v e lin g s a le s p e rs o n s , te c h n ic ia n s , e n g in e e rs , e tc ., s h o u ld c o v e r th e lo c a tio n f r o m w h ic h
th e y a re p a id o r
th e b a se f r o m
w h ic h p e r s o n n e l o p e r a te t o c a r r y o u t th e ir a c tiv itie s .
N O TE :
If m ore than on e establishm ent Is in clu d e d, in fo rm a tio n in S ection III should reflect
the co m b in e d activities o f all such establishm ents. O n e code w ill be assigned w h ich best
indicates the nature o f business o f the gro u p o f establishm ents as a w hole.
SECTION IV. M ON TH O F OSHA INSPECTION
E n te r the nam e o f the first m o n th in 1 9 86 d u rin g w h ich y o u r establishm ent(s) had an
O S H A co m plian ce inspection. In clu d e inspections u n de r the Federal o r State equivalents of
the O ccu p a tion a l Safety and Health A c t b y Federal o r State inspectors and othe r inspections
w h ich m a y result in penalties fo r vio la tio n s o f safety and health standards D o n o t include
inspections lim ite d to elevators, boilers, fire safety o r those w h ic h are consultative in nature.
S ECTION V.
R ECORDABLE IN JUR IES OR ILLNESSES
C heck the approp ria te b o x . If yo u checked " Y e s ," co m ple te Sections V I and V I I on the
back o f the fo rm . If y o u checked " N o , " co m ple te o n ly S ection V I I .
SECTION V I. O CCU P A TIO N A L IN JUR Y AN D ILLNESS SUMMARY
T h is section can be co m ple te d easily b y co p y in g the totals fro m the annual s um m a ry of
y o u r T986 O S H A N o . 2 0 0 fo rm (L o g and S u m m a ry o f O ccu p a tion a l Injuries and Illnesses).
Please n ote that if this re port covers more than one establishm ent, the final totals on the
" L o g ” fo r each m ust be added and the sums entered in S ection V I.
Leave S ection V I blank if the em ployees covered in this re p ort experienced no recordable
injuries o r illnesses d u rin g 1986
If there w ere recordable injuries o r illnesses d u rin g the ye a r, please review y o u r O S H A
N o. 2 0 0 fo rm fo r each establishm ent to be included in this re port to m ake sure that all
entries are co rre ct and co m ple te before c o m p le tin g S ection V I . Ea ch recordable case should
be included on the " L o g ” in o n ly one o f the six m ain categories of injuries o r illnesses:
A ls o review each case to ensure that the approp ria te entries have been m ade fo r the o th e r
co lu m n s if applicable. F o r e xam ple, if the case is an In ju ry w ith Lo st W o rk d a ys , be sure that
the check fo r an in ju ry in vo lving days away from work (L o g c o lu m n 3 ) is entered if necessary.
A ls o v erify that the co rre ct n u m b e r o f days a w ay fro m w o rk (L o g co lu m n 4 ) an d /or days of
restricted w o rk a ctiv ity (L o g c o lu m n 5 ) are recorded. A s im ilar review should be m ade fo r a
case w h ich is an Illness w ith Lo st W o rk d a ys (in clu d in g Lo g co lu m n s 10, 11 and 1 2 ). Please
rem em ber that if y o u r e m plo yees' loss o f w o rk d a ys is still co n tin u in g at the tim e the annual
sum m a ry for the year is co m p le te d , y o u should estim ate the n u m b e r o f fu tu re w o rk d a ys
they w ill lose and add this estimate to the actual w o rk d a y s already lost. Each partial d a y
a w ay from w o r k , o th e r than the d a y o f the occurrence o f the in ju ry o r onset o f illness,
should be entered as one full restricted w o rk d a y.
Als o , for each case w h ich is an Illness, m ake sure that the approp ria te co lu m n in dicating
T y p e of Illness (L o g co lu m n s 7a-7g) is checked.
A f te r co m ple tin g y o u r review of the in dividu a l case entries on the " L o g , " please m ake sure
that the " T o t a ls " line has been co m ple te d b y sum m a rizin g C o lu m n s 1 th ro u gh 13 a ccord in g
to the instructions on the back o f the " L o g " form . T h e n , c o p y these " T o t a l s " o n to S ection
V I o f the O S H A N o . 2 0 0 -S fo rm . If yo u entered fatalities in co lu m n s ( I I and/or (81, please
include in the 'C o m m e n ts " section a brief d e scription of the ob je ct or e vent w h ic h caused
each fatality.
FIR ST AID
only Firit Aid
Treatment, even w hen adm inistered b y a d o c to r o r nurse, shou ld n o t be included in this re
F in a lly , please rem em ber that all injuries w h ic h , in y o u r ju d g em e n t, required
p ort. F irs t A id Tre a tm e n t is defined as on e -tim e tre atm e nt and subsequent observation o f
1.
I N J U R Y -r e la t e d deaths (L o g c o lu m n 1)
m in o r scratches.cuts, b u rn s, splinters, e tc., w h ic h d o n o t o rd in a rily require m edical care.
2. I N J U R I E S w ith days aw ay fro m w o rk and/or restricted days (L o g c o lu m n 2 )
3. I N J U R I E S w ith o u t lost w o rk d a ys (L o g c o lu m n 6 )
4. IL L N E S S -r e l a t e d deaths (L o g co lu m n 8 )
5.
6.
SECTION V II.
I L L N E S S E S w ith d ays a w ay fro m w o rk and/or restricted d ays (L o g c o lu m n 9)
I L L N E S S E S w ith o u t lost w o rk d a ys (L o g c o lu m n 13)
113
COM MENTS AN D ID E N TIF IC A TIO N
Please co m ple te all parts in clu d ing y o u r area co de and telephone n u m be r. T h e n re turn the
O S H A N o . 2 0 0-S fo rm in the pre-addressed envelope.
KEEP y o u r file c o p y.
Dear Employer:
The Occupational Safety and Health Act of 1970 requires the Secretary of Labor to collect, compile, and analyze statistics on occupational injuries and illnesses. This is accomplished through a joint Faderal/State survey program with
States that have received Federal grants for collecting and compiling statistics. Establishments are selected for this sur
vey on a sample basis with varying probabilities depending upon size. Certain establishments may be included in each
year's sampla because of their importance to the statistics for their industry.
You have been selected to participate in the nationwide Occupational Injuries and Illnesses Survey for 1986. Under the
Occupational Safety and Health Act, your report is mandatory.
The following items are enclosed for your use: (1) Instructions for completing the form; (2) The O SHA No. 200-S form
and a copy for your files; and (3) An addressed return envelope. Please complete the OSHA No. 200-S form and return
it within three weeks in the envelope provided.
If you have any questions about this survey, contact the survey collection agency indicated on the OSHA No. 200-S form.
Thank you for your cooperation w ith this important survey.
JOHN A. PENDERGRASS
Assistant Secretary for
Occupational Safety and Health
114
Chapter 15. Economic Growth
and Employment Projections
Part I.
bls develops and publishes 10- to 15-year projections
of the U.S. economy on a 2-year cycle, with the initial
results published in the fall of odd-numbered years. The
projections cover the labor force, aggregate economy,
final demand, output by industry, employment by
industry, and employment by occupation. In recent years,
alternative projections have been made based on differ
ing assumptions with regard to basic economic activity—
such as the level of business investment or path of govern
ment spending.
The basic principles underlying the procedures used to
develop the projections have remained constant over the
years, but many changes in procedures have been made
as new series of data become available and as statistical
tools improve. The current methodology has been
relatively unchanged since the late 1970’s.
Projections are made from a base year to a target year.
The base year is the latest year for which data are
available at the time the projections are being developed,
usually the year immediately preceding publication. The
target year is the last year of the period being projected,
usually 10 to 15 years in the future. Generally, each set
of projections has a single target year, but some projec
tions are prepared for both the target year and an interim
year. The projections do not attempt to measure yearto-year change; they only project levels of economic
activity for the target year.
The bls projections are developed in a series of six
steps or stages, each of which is based on a separate
model: (1) labor force, (2) aggregate economy, (3) in
dustry final demand, (4) input-output, (5) industry
employment, and (6) occupational employment. While
each of these steps is taken separately, they are closely
interrelated, the results of one usually being needed for
the next; the third and fourth steps, in particular, are very
tightly interwoven. Details on the data used in each model
and the methods used for each projection appear in parts
II through VII of this chapter; however, some general
comments follow with regard to assumptions, procedures,
presentation, and limitations of the projections.
Assumptions and Procedures
The development of projections requires analyzing
large amounts of historical data, discovering trends, and
determining a likely course for those trends in the future.
Overview
The determination of the likely course of a trend is very
problematic, and various factors contribute to each deci
sion. Users of BLS projections, like the users of any pro
jections, need to be aware of the underlying assumptions
and should consider the projections as likely outcomes
in the light of those assumptions and current and expected
trends.
First, some assumptions are made concerning general
economic or social conditions. These include the follow
ing: (1) work patterns will not change significantly over
the projection period; for example, the average workweek
will not change markedly; (2) broad social and educa
tional trends will continue; (3) there will be no major war;
(4) there will be no significant change in the size of the
Armed Forces; and (5) fluctuations in economic activity
due to the business cycle will occur. These assumptions
have both an overall and a particular impact. For exam
ple, the assumption that social trends will continue implies
that our society will continue to provide for the educa
tion of the young, which influences the projected level
of local government demand and the demand for
teachers.
Second, particular ranges in the target year are set for
certain factors, such as unemployment—a factor that
obviously affects the figure for total employment and thus
the employment in each industry and occupation. The
projections are particularly sensitive to such target ranges
as real gross national product (GNP), g n p deflator, real
disposable income, civilian labor force, civilian employ
ment, real output per person, and the unemployment rate.
For these, BLS specifies up to three alternatives rather
than one, thus generating three scenarios, or projected
levels of economic activity, for the target year.
Third, the analysis of the historical data calls for
judgments as to influences on the past rate of change that
will become more or less important in the future. Dur
ing the 1970’s and early 1980’s, for example, employment
of cashiers in retail stores grew at the expense of other
sales occupations as these stores centralized their cashier
services. However, this factor will no longer cause
changes in the kinds of workers retail stores hire, in the
judgment of bls analysts, because most such stores now
have centralized cashier operations.
To insure the consistency of the six models and many
assumptions, goals, and variables, the bls projection
procedure encompasses detailed review and analysis of
115
the results at each stage for soundness of economic logic,
reasonableness, and consistency. Preliminary projections
made in each step are reviewed by both the analysts who
work primarily on that projection and analysts who work
primarily on other steps. As a result of detailed analyses,
the models are rerun and reanalyzed several times until,
in the judgment of the bls staff, projections that are
both reasonable and internally consistent are achieved for
the entire system. In short, the final results reflect
innumerable interactions among staff members who focus
on particular variables in the models.
Data are also released in machine-readable form. Com
puter tapes in both IBM standard and unlabeled formats
are prepared containing aggregate projections, inputoutput data, historical macro data, historical industry
data, projected industry data, and occupational projec
tions. A diskette for use with Lotus 1-2-3 contains
industry data on output, total employment, wage and
salary employment, and total hours. Both tapes and
diskettes are sold through the bls Division of Financial
Planning and Management.
Limitations
Presentation
The projections are first published in the Monthly
Labor Review, usually in the fall of odd-numbered years,
and subsequently appear in bls bulletins and the Occupa
tional Outlook Quarterly, which also prints articles on
such topics as new and emerging occupations and chang
ing job market conditions for existing occupations.
Three bulletins were published for the 1986-1995 pro
jections: b l s Economic Growth Model System Usedfor
Projections fo r 1995: Data and Methods; the Occupa
tional Outlook Handbook, 1986 Edition; and Occupa
tional Projections and Training Data, 1986Edition. The
first reprints the Monthly Labor Review articles and con
tains supplementary data, such as selected aggregate
economic variables, the size of each labor force cohort,
final demand by industry, gross output and employment
by industry, and wage and salary employment by detailed
industry. The Occupational Outlook Handbook discusses
more than 200 occupations; besides outlook data, it
includes information on the nature of the work, training
requirements, working conditions, and earnings. The
Handbook is available in the vast majority of career
information centers in the country’s high schools, col
leges, and libraries, where it is used as a primary source
of information for people in the process of choosing a
career. Occupational Projections and Training Data
presents detailed statistics on employment and education
and training completions; most of the data are for the
occupations included in the Handbook.
Part II.
Labor Force
The labor force projections, the first step in the BLS
projections sequence, depend upon an analysis of the cur
rent population and projections of its future size and
composition. Projections are made for the labor force
as a whole and for more than 80 separate age-sex-race
groups and more than 20 age-sex groups for people of
Hispanic origin.
Because of the nature of projections, misunderstand
ings may arise between users, who feel the need for exact
numbers, and producers, who recognize their inability to
predict with such precision. Such conflicts are all the more
likely because the models used to develop the projections
provide numerical answers to specific questions. Users
are inevitably tempted to attribute to those numbers an
exactness which they should not be accorded. The transla
tion of analytical judgments, such as those concerning
the impact of a specific technology on the need for
workers with specific skills, into numerical estimates is
especially subjective. For example, most analysts would
agree that the use of robots will affect employment in
manufacturing industries. Yet numerical estimates of the
reduction in the proportion of assemblers and welders in
an industry affected by robots could vary significantly
among analysts. The Bureau attempts to address this
dilemma by making clear all of the important assump
tions underlying its projections, by developing alternative
versions that reflect some of the uncertainties and dif
fering policy decisions about the future, and by making
projections on a regular 2-year cycle in order to incor
porate new data and assumptions.
The Bureau also seeks to improve the projections
process and to make users aware of their limitations by
reviewing previous projections. Once the target year is
reached, bls evaluates the projections to determine the
errors and to learn what changes in assumptions or
models might have made them more accurate.
Data Sources
Projections of the labor force require projections of
the population and data on labor force participation. The
Bureau of the Census prepares population projections by
age, sex, and race, based on trends in birth rates, death
rates, and migration. Data on participation rates are
116
based on the Current Population Survey (CPS), con
ducted for b l s by the Bureau of the Census. (A descrip
tion of the CPS appears in chapter 1.)
Methods
Various assumptions can be made for either popula
tion growth or labor force participation. Because birth
rates cause much uncertainty in projecting the popula
tion, the Bureau of the Census prepares three projections
of the population based on differing assumptions with
respect to birth rates, b l s selects one of these, generally
the middle one.
In recent years, b l s developed three alternative sets of
labor force projections by specifying different variables
for certain trends of labor force participation. In the
1984-95 projections, for example, a high-growth scenario
assumed rapid growth in the labor force participation of
women in the 1980’s and the convergence of participa
tion rates of black and white men under age 65, rates that
have been diverging since 1955; a middle-growth scenario
assumed only the rapid growth of women’s participation;
a low-growth scenario assumed a moderate increase in
women’s participation and continued divergence in the
participation rates of black and white men.
Using the Census Bureau’s population projections, b l s
projects labor force participation rates—the proportion
of various groups in the population who will be working
or seeking work. Projections for the 1986-2000 period
were made for more than 110 separate demographic
groups because both the level and trends of participation
vary considerably by age, sex, race, and Hispanic origin.
The labor force participation projection for each agesex-race group and for each group of Hispanic origin is
developed by analyzing past rates of growth for that
group and extrapolating it to the target year. Past par
ticipation rates may be generated from data for two dif
Part III.
Uses
Labor force projections are a basic factor in estimating
the amount of economic growth necessary to achieve
specified levels of employment. They provide insight into
the demographic characteristics of future workers and the
implications of these for education and training. In addi
tion, along with other factors, they are used by planners
in business and industry to estimate demand for their
products, develop marketing plans, evaluate expansion
programs, and build macroeconomic models. Interna
tional organizations and Federal, State, and local govern
ment agencies also use the projections.
Aggregate Economy
Projections of the aggregate economy—the second
stage in the b l s projection procedures—are made
through use of a macroeconomic model referred to
throughout this chapter as the macro model. The labor
force projections made in the first stage of the b l s pro
cedure are used along with many other variables to
develop projections of g n p and major categories of
demand and income. Because the variables are so
numerous, sources of data are manifold. They include
b l s itself, the U.S. Departments of Commerce and
Energy, the Federal Reserve Board, and Wharton
Econometric Forecasting Associates, Inc.
ferent time periods, one of about 20 years and one of
less than 10 years. The choice of period is based on a judg
ment of which period will prove to be a better predictor
for the group. Groups whose participation rate is chang
ing rapidly, for example, would be projected on the basis
of the shorter period.
Cross-sectional and cohort analyses are also conducted
for each group. In cases where these analyses show incon
sistencies, the participation rate extrapolated from the
time-series data is modified. The modification is greatest
for black women and smallest for white men.
At each stage, the results for specific groups are
reviewed and modified if not consistent with those for
other demographic groups. The projected participation
rate for each age-sex-race group is then multiplied by the
corresponding population projection to obtain the labor
force projection for that group. Because people of
Hispanic origin may be of any race, the estimates for
those age-sex groups are not added to the projections for
the age-sex-race groups. The latter are summed to obtain
the total labor force. The size of the civilian labor force
is then derived by subtracting the Armed Forces from the
total labor force.
Methods
For the past few cycles of projections, b l s has used
models of the economy—which are basically sets of equa
tions that correlate different aspects of the economy with
each other—created by other organizations. The specific
equations used in a model may differ, but they work in
similar ways to provide a framework for the preparation
of a consistent set of economywide projections for a given
set of assumptions and goals. The 1986-2000 projections
were based on a macro model created by Wharton
Econometric Forecasting Associates, Inc. The macro
117
model is a system of behavioral relationships and iden
tities based on annual data and designed to allow an
analyst to explore the determinants of growth in the U.S.
economy. Made up of approximately 2,400 equations,
the macro model is driven by a set of 900 exogenous
variables—arithmetic values that can be manipulated by
the equations in the macro model, b l s specifies the value
of these variables.
The exogenous variables can be divided into three
groups, according to the degree of certainty to which each
can be determined. Reliable, generally accepted values
are available for some variables, such as the future size
of the population; Census population projections have
proven to be highly accurate. Other variables involve
policy decisions that, while subject to change, have
remained fairly constant for many years; these include
the amount of Federal transfer payments, the response
of the monetary authority to economic growth, and the
size of the Armed Forces. Finally, some exogenous
variables do not follow predictable relationships; these
include the inflation rates in the economies of the major
trading partners of the United States, the exchange value
of the U.S. dollar, and energy prices.
Besides being governed by general assumptions, the
projections are generally approached with certain goals
or targets in mind. Because the goals relate to variables
that are, strictly speaking, results of the aggregate model
rather than inputs to it, they are attained by changes to
the structure of the model itself. Such goals or target
variables include the unemployment rate, the rate of
growth of labor productivity, inflation, and the sectoral
distribution of employment.
Part IV.
represents demand on the part of persons and
certain nonprofit institutions. Rent and the imputed
rental value of owner-occupied dwellings are included
in this category, but the purchase of dwellings is not.
(2) Capital investment represents demand on the part of
business plus residential construction. Capital invest
ment includes both fixed investment—such as the pur
chase of durable equipment and structures—and
change in business inventories of raw materials,
semifinished goods, and finished goods.
Uses
The values produced by the macro model for many
variables are used at later stages in the bls projection
procedures. Among these are the values for GNP and
several different categories of personal consumption
expenditures, gross private domestic investment, exports,
imports, government purchases, and employment.
The projections also form an important part of the
U.S. Government’s report to international organizations
on the long-term economic outlook for the United States.
In addition, other Government agencies use parts of the
economic growth projections to develop projections for
their program needs. State and local governments, area
planning councils, outside research organizations, and
universities use these data when planning programs,
building their own models, or evaluating projections.
Industry Final Demand
Projections of final demand are made in the third stage
of the BLS projection procedures; these projections are
made along with the creation of the input-output model
described in the next section of this chapter. Final demand
is one way to view g n p ; it is g n p distributed among final
users, broadly categorized into four groups: (1) personal
consumption expenditures ( p c e ), (2) capital investment,
(3) foreign trade, and (4) government.
(1 ) p c e
Once the value of each variable has been determined,
the macro model is run, that is, the equations are solved,
producing projected values for numerous kinds of eco
nomic activity, such as g n p , purchases of consumer
goods, and capital investment, b l s analysts review the
aggregate results for soundness of logic and reason
ableness. The review includes checks on internal con
sistency, evaluation of continuity with past trends, and
comparisons with projections made by other organiza
tions. Although the review tends to focus on such items
as g n p , unemployment, and productivity, the macro
model’s framework ensures that other important meas
ures of economic performance are not overlooked.
(3) Foreign trade includes both exports and imports.
These are analyzed separately; imports are subtracted
from exports at the final stage of the projections
procedure.
(4) Government demand is defined as the goods
and services purchased by all government units—
local, State, and Federal. It does not include transfer
payments such as those made in the Social Security
program, interest, or subsidies, all of which are
accounted for under personal consumption expen
ditures or capital investment.
Final demand determines the total output of the
economy, which in turn determines the distribution of
employment. Because the purpose of production is the
satisfaction of demand, variation in the demand for
goods and in the means of producing these goods
changes the level and distribution of employment over
time.
118
Data Sources
In general, projecting final demand entails the com
pilation of historical data in a form that helps determine
the industry distribution of the economy for some future
year, b l s uses “ bridge” tables and bills of goods that
show the industry composition of final demand com
ponents in order to perform this analysis. Projections
made in the macro model and data from the economic
censuses conducted by the Bureau of the Census provide
the information needed. Data for the target year of the
projections come from the macro model, which generates
values for more than 40 different categories of final
demand.
For past years, large amounts of data in various forms
are available. Much of it is provided by the Bureau of
Economic Analysis ( b e a ) of the Department of Com
merce. Although considerable data are collected annually,
the creation of a bill of goods requires a level of detail
that is available only from the economic censuses, and
b e a only produces a bill of goods for those years. The
historical series depends heavily, therefore, on the census
years; more recent economic trends are incorporated into
the projections through supplementary data series.
Besides the economic censuses, b l s analysts use several
other series of data produced by the Department of Com
merce in order to project final demand, p c e are compiled
by BEA in the National Income and Product Accounts
( n i p a ); they are available annually from 1929, disag
gregated into more than 80 components of consumption
expenditures, b e a also produces a bridge table for each
year in which an economic census is held; the bridge table
distributes the more than 80 p c e categories among over
500 producing industries. Two principle sources of data
are available for capital investment: (1) investment by
industry, collected annually by the Office of Business
Analysis, which yields information on investment in
capital goods; and (2) n i p a annual data, which yield
information on both durable goods and structures. A
capital flows matrix, produced by b e a for the years in
which there is an economic census, distributes investment
by industry into the producing industry. For foreign
trade, plentiful data on exports and imports are available
in the detailed merchandise trade statistics published
annually by the Bureau of the Census. Data from the
Department of Commerce and from specialized surveys
allow for the construction of bills of goods for govern
ment demand.
Methods
Analytical judgm ents
In order to project final demand, the same kinds of
judgments and assumptions must be made as those that
enter into the macro model. For some components, the
data available permit reasonable certainty. For example,
the demand for education, a significant factor in State
and local government demand, reflects the projected age
structure of the population. Similarly, demand for
residential construction depends heavily on demographic
and income forecasts. Judgments must also be made,
however, with regard to the effect of technological
developments—such as computers and robots—and other
factors for which data are less reliable.
Projections
The historical data and the initial projections of various
categories of final demand generated by the early runs
of the macro model provide a starting point for the
analysts, who must study all aspects of demand to insure
that the models remain balanced and consistent
throughout the development of a new set of projections.
Although the four categories of final demand—personal
consumption expenditures, capital investment, foreign
trade, and government—are subject to different pro
cedures, each makes use of bridge tables.
The bridge tables distribute final demand to more than
500 producing industries that generally correspond to the
4-digit Standard Industrial Classification (sic) codes.
They provide a percent distribution of the industries sup
plying the various categories into which g n p is appor
tioned. Projected bridge tables are created by b l s for
roughly 220 industries; most are at the 3-digit sic level.
The projections use this level of aggregation because of
the nature of the data. The projected bridge tables reflect
such factors as expected changes in technology, consumer
tastes or buying patterns, the industrial pattern of exports
and imports, and the future composition of each indus
try’s business investment. Thus, the bridge tables allow
the analyst to provide for shifts in the industrial makeup
of a given demand category. Having the data at this level
of allocation allows finer adjustment for technological
and economic change.
Personal consumption expenditures are projected by
aggregating the more than 80 categories in the bridge table
into 15 major product groups. Some categories, such as
gasoline and oil, are identical to a b e a category, but
many are composites of several b e a categories. For
example, the transportation services category in the b l s
macro model includes all of the following BEA categories:
Automobile repair, road tolls, automobile insurance less
claims paid, bus and trolley car transportation, taxicabs,
commuter rail transportation, railway transportation,
intercity buses, airline transportation, and other intercity
transportation. The projection for each of the 15 cate
gories is then modified to reflect technological and
economic assumptions as well as industry trends that are
expected to continue over the projected time span.
119
Capital investment is initially projected by the macro
model, which generates values for investment in durable
goods and new structures for the whole economy and
distributes these figures among aggregate industries. The
investments by purchasing industries in durable goods and
new structures can then be converted into the producing
industries by use of the b e a capital flows tables. The
results must be carefully reviewed by the staff because
the capital flows tables are purely historical and do not
take into account expected changes in technology or other
factors. Changes that result from these reviews may
necessitate complementary adjustments in other parts of
the macro model. Adjustments to the macro model may
also be necessary to reflect changes in inventories.
Foreign trade is initially projected by the macro model,
which generates values for components of exports and
imports, such as foods, feeds, and beverages, consumer
goods, and autos and parts. These values are distributed
across more than 200 industries in the light of past trends,
existing and expected shares of the domestic market,
expected world conditions, and trade agreements. Once
this distribution has been made, it is reviewed by the staff
and adjustments made as necessary. Total imports are
divided into two categories: Those competitive with
domestic products and those that have no domestic
counterparts, such as coffee and diamonds. Competitive
imports are subtracted from final demand in order to
Part V.
Government demand is initially projected by the macro
model for six categories: State and local education; State
and local health, welfare, and sanitation; State and local
safety; State and local other; Federal defense; and Federal
nondefense. Projections of the size of the government
labor force—State and local, Armed Forces, and Federal
civilian—are also made at earlier stages of the b l s pro
jection procedures. Consequently, the analyst is able to
project at this stage the compensation that governments
will provide their employees and subtract this amount
from the value for government demand. The remainder
is distributed across other industries in the economy by
means of a bridge table.
Uses
Projections of final demand are used to refine the
macro model and are major components of the inputoutput model described in the next section.
Input-Output
The creation of an input-output model is the fourth
stage of the b l s projection procedure. Each industry
within the economy relies on other industries to supply
inputs—intermediate products or services—for further
processing. The input-output model insures consistency
between demand estimates and output estimates and per
mits analysis of technological change and product
substitution.
The components and support services that enter into
a product are frequently produced by industries other
than the one that provides the final product. For exam
ple, an automobile is a final product and its value is part
of the final demand model. However, the carmaker must
purchase steel, glass, electricity, secretarial assistance, and
financial services in order to produce the automobile;
these intermediate products do not explicitly appear in
the final demand model. The input-output model does
show these purchases by the carmaker. Analysis of such
interrelationships results in a more precise projection of
each industry’s production than is possible from the pro
jection of final demand alone. This, in turn, allows for
a better projection of employment by industry.
In addition to allowing the examination of techno
logical change, the input-output model makes possible
derive the domestic output. For example, the projected
value of imported automobiles is subtracted from total
demand for autos so that the demand-for-autos compo
nent of the macro model will reflect domestic products
only. Noncompetitive imports are considered purchases
of the industry that uses them; some noncompetitive
imports that do not need any processing, such as bananas,
are only included in the final user category.
the analysis of changes in demand for secondary products
of an industry. In the historical data, establishments are
grouped into industries according to their primary
product—those goods or services that produce the largest
part of its revenue. Many establishments, however, pro
duce more than one product. The input-output model
enables analysts to look at changes in output for all the
products of an industry and in all the industries that sell
a product, allowing for projected changes in demand to
be made proportionately across all relevant industries.
Data Sources
Historical data prepared by b e a , projections of final
demand developed in the third stage of the procedure,
and studies of particular industries and commodities are
all needed to produce the input-output model, b e a
creates historical input-output tables based on data con
tained in the Census o f Manufactures and other economic
censuses. Because the most recent of these is often
somewhat outdated, BLS uses data from more recent
surveys and other sources in order to construct inputoutput tables for more recent years.
120
Methods
The b l s input-output model consists of five matrices,
or tables. The “ use” table shows the sales in dollars of
each commodity to every consuming industry and to final
demand; rows sum to commodity output and columns
to industry output. The “ make” table shows, in dollars,
the production of commodities by each industry; rows
sum to industry output and columns sum to commodity
output. The direct requirements table presents the values
from the use table as coefficients. The market shares
matrix presents the values from the make table as coeffi
cients. The total requirements table combines data from
the direct requirements table and the market shares
matrix; it shows total requirements—direct and indirect—
to produce a dollar of final demand.
After these adjustments are made, b l s converts the
dollar values shown in the tables into coefficients show
ing relative values. These coefficients are used to generate
the projected input-output tables.
Throughout the projection process, the coefficients
Part VI.
used for the projected input-output matrices may be
changed for several reasons, such as technological devel
opments, changes in product mix or relative prices, and
the availability of substitute inputs or more current data.
The coefficients in the two matrices can be analyzed and
adjusted on both an industry-by-industry and a commod
ity-by-commodity basis to insure that they reflect the best
information available and are consistent with other pro
jections. As with the other steps, several reviews and
interim projections are required before the final matrices
are produced.
Uses
The projected output of industries is an important
part of the next stage of projections procedure, the
projection of industry employment. The input-output
model also permits other analytical uses. Specifically, the
model can be used to generate labor requirements for
various economic sectors or types of demand for recent
years.
Industry Employment
The fifth stage of the projection procedure is the
analysis of trends in industry output and employment.
Fast-growing and declining industries are identified, and
employment is projected for each of more than 200
industries. Industry employment is projected through the
use of a labor model, developed by b l s , that correlates
productivity with industry output. An equation is
specified for each industry, relating the demand for labor
in that industry to the output of the industry and to cer
tain other economic variables.
self-employed and unpaid family worker jobs and hours,
and private household workers (chapter 1); and unemploy
ment insurance data for employment in industries not
covered in the establishment survey (chapter 5).
Projected data for this model come from the inputoutput, labor force, final demand, and other b l s models.
Projections of capital, ratio of output to capital, capacity
utilization, average workweek, and total wage and salary
employment in durable goods manufacturing, nondurable
goods manufacturing, and nonmanufacturing come from
the macro model.
Data Sources
Methods
Historical data for the industry employment model are
developed from a wide variety of sources. Time series on
output (in constant dollars) for manufacturing industries
are estimated from the Annual Survey of Manufactures
conducted by the Department of Commerce and from
b l s industry and producer price indexes. For non
manufacturing industries, the sources for the output and
price data are very diverse; they include n i p a , the Serv
ice Annual Survey, ir s business receipts, Agricultural
Statistics, Minerals Yearbook, transportation statistics,
and numerous others. Time series on employment and
hours are derived from three BLS sources for different
groups of workers: The establishment survey for nonagricultural wage and salary employment, production worker
employment, and weekly hours (see chapter 2); The Cur
rent Population Survey for agricultural employment,
Assumptions
Besides the general assumptions that govern the macro
model, specific assumptions may be made for selected
industries in order to project employment. For example,
productivity in a particular industry may be assumed to
equal, exceed, or fall short of the value determined by the
labor model. Other assumptions that affect many indus
tries also have to be made; recently, for example, an as
sumption has been made that new capital spending would
be devoted in large part to high-technology innovations.
Projections
The
121
bls
labor model contains, for each of more than
200 industries, a regression equation that estimates
worker-hours as a function of four factors: Industry out
put, aggregate capacity utilization (as approximated by
the unemployment rate), relative price of labor, and a
technological variable as approximated by the ratio of
output to capital. Worker-hours are then divided by pro
jected average annual hours, yielding the projected
number of wage and salary jobs for each industry.
Estimates are then made, based on analysis of individual
industries, for the self-employed and for unpaid family
workers.
Many adjustments are made while the projections are
being developed. Adjustments are usually required in situ
ations such as the following: An industry does not operate
near the conditions of maximum profit, historical and
projected trends diverge widely, the output and employ
ment time series data are inconsistent, new technology
is expected, or the labor productivity trend implies nega
tive employment. Adjustments to the labor model’s initial
estimate of productivity change are based on detailed
analysis of historical and recent trends, special industry
studies conducted by the staff, and industry technology
studies from the Bureau’s Office of Productivity and
Technology (chapter 12). In order to make all these
Part VII.
Uses
The projections of employment for industries are used
in the industry-occupational matrix and in special studies
and articles reporting on trends in specific industries. In
addition, a labor requirements table is computed, show
ing the number of jobs in each detailed industry related
to final demand for each of the commodities in the inputoutput table described in the previous section. The
historical time series data on output and employment are
also used extensively by researchers in universities, other
government agencies, and businesses for market research,
industry analysis, and economic impact studies.
Occupational Employment
The final stage in the b l s projection procedure is the
development of occupational projections. To generate the
data on wage and salary workers, an industryoccupational matrix, or table, showing the distribution
of occupational employment by industry is constructed
for the base year and projected to the target year. Baseyear data are also developed for self-employed workers
and unpaid family workers. The projection for these
workers, however, is made for the economy as a whole
rather than by industry. The three classes of workers—
wage and salary workers, self-employed, and unpaid
family workers—are summed to derive the projections
for each occupation. The 1986 matrix, which was pro
jected to the year 2000, includes more than 400 occupa
tions in 258 industries.
Data Sources
In order to project occupational employment, data are
needed on staffing patterns of wage and salary workers
by industry, employment of wage and salary workers by
industry, and on self-employment and the employment
of unpaid family workers.
Information on industry staffing patterns of wage and
salary workers is available from several surveys. The Occu
pational Employment Statistics ( o e s ) surveys conducted
periodically by State employment security agencies under
adjustments and to balance total employment from the
aggregate projections with the sum of the industry
employment projections, a number of iterations of the
process are necessary.
Once the projections of employment by industry are
final, they are further disaggregated using a time series
regression model into 258 industries that, with few excep
tions, correspond to the 3-digit SIC codes. The 258
resulting projections are reviewed in light of a broad range
of economic information.
a BLS-State cooperative program provide information for
all but a few industries (chapter 3). Information on agri
culture and private household industries is available from
the c p s and the Decennial Census of the Population.
Economywide data on self-employed and unpaid family
workers are also derived from the CPS and the decennial
census. Information on staffing patterns for the Federal
Government is developed by BLS from data compiled by
the Office of Personnel Management ( o p m ).
Methods
Disaggregation and aggregation
Occupations in the industry-occupational matrix are
classified according to the system used in the OES surveys,
which is compatible with the Standard Occupational
Classification (soc) Manual. It is very similar to the
system used for the census and the c p s . The o e s survey,
however, compiles data for broader groups of occupa
tions in some industries. The development of economy
wide employment estimates for the detailed occupations
in these industries requires disaggregation of the broader
survey data. Data from the decennial census are used for
these adjustments.
Adjustments also have to be made for the industries
that are not included in the o e s : Agriculture, private
122
households, and Federal Government. For example,
the occupational classifications used by OPM are more
detailed than the soc, necessitating the aggregation of
many occupations to make them comparable to the
classification used in the matrix. Similarly, estimates of
employment for self-employed and unpaid family
workers derived from the CPS and decennial census must
be adjusted to make them comparable to the other data
in the matrix.
Once all the data have been reviewed, they are arrayed
in a matrix that shows occupational employment
distributed in percentages by industry. These percentages,
however, are derived from surveys conducted in different
years on a 3-year cycle. The percentages are, therefore,
applied to total industry employment estimates for the
base year in order to develop occupational employment
estimates for the base year.
Projections
When a matrix for the base year has been developed,
a projected matrix for the target year can be made.
Changes in staffing patterns must, of course, be allowed
for in the projections. This is done in several ways.
Historical data are reviewed to identify trends; factors
underlying the trends are identified through analytical
studies of specific industries and occupations, techno
logical change, and a wide variety of economic data; and
judgments are made as to how the pattern will change.
Factors underlying change are numerous, including
technological developments affecting production and
products, innovations in the ways business is conducted,
modifications of organizational patterns, responses to
government policies, and decisions to add new products
and services or stop offering old ones.
Some expected trends may be too recent to be evident
in the historical data. For example, an analysis of the past
would not point toward the future impact of robots on
staffing because this technology has not been used much
in most industries. However, robots are expected to have
a significant impact on some occupations, especially in
the automobile industry. Information of this nature is
identified by studies conducted by the Bureau’s Office
of Productivity and Technology and other organizations.
The industry-occupation matrix can display either the
number employed or coefficients that show the propor
tion of workers in an industry that work in a particular
occupation. The matrix of coefficients is used to project
the staffing pattern of an industry. The change projected
for a specific occupation may be small, moderate, signifi
cant, or very significant; the precise percentage reflects
the analytical judgment of the staff members based on the
analysis described above that relate to that occupation.
Several versions of the projected matrix are developed
during the projection procedure. Each version is reviewed
by members of the staff preparing the Occupational
Outlook Handbook and economists working on other
steps in the projection program; these reviews insure that
all information available to the staff is brought to bear
on the projections and that consistent assumptions are
made for all the Bureau’s projections. Knowledgeable
people outside the Bureau are also asked to comment.
The final matrix represents, therefore, a broad consen
sus on the part of all analysts working on the projections.
Uses
The occupational projections are very widely used by
economists, counselors, students, and others concerned
with the future of the economy.
123
Technical References
Fullerton, Howard N, Jr. “The 1995 Labor Force: b l s ’ s
Latest Projections,” Monthly Labor Review, November
1985.
U.S. Department of Commerce, Office of Federal Statistical
Policy and Standards. Standard Occupational Classif
ication Manual, 1980.
Personick, Valerie A. “A Second Look at Industry Output
and Employment Trends Through 1995,” Monthly Labor
Review, November 1985.
U.S. Department of Labor, Bureau of Labor Statistics.
Economic Growth Model System Used fo r Projec
tions fo r 1995: Data and Methods, Bulletin 2253, 1986.
bls
A similar publication appears biennially.
Silvestri, George T., and Lukasiewicz, John M. “ Occupa
tional Employment Projections: The 1984-95 Outlook,”
Monthly Labor Review, November 1985.
Su, Betty W. “The Economic Outlook to 1995: New As
sumptions and Projections,” Monthly Labor Review,
November 1985.
Tschetter, John. “An Evaluation of b l s ’ s Projections of the
1980 Industry Employment,” Monthly Labor Review,
August 1984.
U.S. Department of Commerce, Bureau of the Census. Pro
jections o f the Population o f the United States: 1983 to
2080, Current Population Reports, series P-25, No. 932,
1984.
U.S. Department of Commerce, National Technical Informa
tion Service. The National OES Survey-Based Indus
try-Occupational Employment Matrix, 1984 and 1995.
Bureau of Labor Statistics. Capital Stock Estimates fo r InputOutput Industries: Methods and Data, Bulletin 2034,
1979.
Bureau of Labor Statistics. Measuring Labor Force Move
ments: A New Approach, Report 81, 1980.
Bureau of Labor Statistics. Occupational Outlook Handbook,
biennial.
Bureau of Labor Statistics. Occupational Outlook Quarterly,
quarterly.
Bureau of Labor Statistics. Occupational Projections and
Training Data, biennial.
U.S. Executive Office of the President, Office of Management
and Budget. Standard Industrial Classification Manual,
1972.
124
Chapter 16.
Producer Prices
The Producer Price Index ( p p i ) measures average
changes in selling prices received by domestic producers
for their output. Most of the information used in
calculating the Producer Price Index is obtained through
the systematic sampling of virtually every industry in the
mining and manufacturing sectors of the economy. The
PPI program (also known as the industrial price program)
includes some data from other sectors as well—
agriculture, fishing, forestry, services, and gas and elec
tricity. Thus the title “ Producer Price Index” refers to
an entire “ family” or system of indexes.
As of January 1987, th e PPI program contained:
• Price indexes for nearly 500 mining and manufactur
ing industries, including approximately 8,000 indexes for
specific products and product categories;
• Over 3,000 commodity price indexes organized by
type of product and end use; and
• Several major aggregate measures of price change
organized by stage of processing.
Together, these elements constitute a system of price
measures designed to meet the need for both aggregate
information and detailed applications, such as following
price trends in specific industries and products.
Measures of price change classified by industry, the
most recent addition to the PPI system, now form the
basis of the program. These indexes reflect the price
trends of a constant set of goods and services which repre
sent the total output of an industry. Industry index codes
are based upon the Standard Industrial Classification
(SIC) system and provide comparability with a wide
assortment of industry-based data for other economic
phenomena, including productivity, production, employ
ment, wages, and earnings.
Background
Known until 1978 as the Wholesale Price Index or w p i ,
the Producer Price Index is one of the oldest continuous
systems of statistical data published by the Bureau of
Labor Statistics, as well as one of the oldest economic
time series compiled by the Federal Government. First
published in 1902, the index covered the years from 1890
through 1901. The origins of the index can be found in
an 1891 U.S. Senate resolution authorizing the Senate
Committee on Finance to investigate the effects of the
tariff laws “ upon the imports and exports, the growth,
development, production, and prices of agricultural and
manufactured articles at home and abroad.” 1
The first index, published on the base period 1890-99,
was an unweighted average of price relatives for about
250 commodities. Since that time, many changes have
been made in the sample of commodities, the base period,
and the method of calculating the index. A system of
weighting was first used in 1914, for example, and major
sample expansions and reclassifications were implemented
in 1952 and 1967.
When it was originally founded, the Wholesale Price
Index program was intended to measure changes in prices
received for goods sold in primary markets of this coun
try. The conceptual framework and economic theory
guiding the program’s evolution, while more implicit than
explicit, concentrated on obtaining the price received by
either a domestic producer or an importer for the first
commercial transaction.
A number of practical gaps, inconsistencies, and other
weaknesses in the industrial price program, combined
with increased development of the theory of price indexes
in preretail markets, spurred several changes in termi
nology and operations during the 1970’s. The 1978
program name change from Wholesale Price Index to
Producer Price Index, for example, was prompted by
widespread misconceptions about the meaning of
“ wholesale.” When the program began at the turn of the
century, “ wholesale prices” was a term commonly
understood to refer to prices for goods sold in large quan
tities. Because that connotation faded over time, how
ever, many in the general public came to assume that the
term “ wholesale prices” referred to prices charged by
wholesalers, jobbers, or other distributors. Adoption of
the term “ producer prices” was intended to reemphasize
that the industrial price program continues to be based
on prices received by producers from whoever makes the
first purchase, rather than on prices paid to wholesalers
by retailers or others further removed in the distribution
chain. This new nomenclature was accompanied in 1978
by a shift in the Bureau’s analytical focus from the all
commodities price index (which was popularly called
“ the” Wholesale Price Index) to the Finished Goods Price
Index and the other stage-of-processing price indexes.
1
Senate Committee on Finance, W holesale Prices, Wages, a n d
T ransportation, Senate Report No. 1394, “ The Aldrich Report,” Part
I, 52nd Congress, 2d sess., March 3, 1893; and U. S. Department of
Labor, Course o f Wholesale Prices, 1890-1901, Bulletin No. 39, March
1902, pp. 205-09.
125
These changes were a prelude to the most comprehen
sive overhaul of industrial price methodology in the pro
gram’s history. Also begun in 1978, this overhaul was
phased in gradually until the transition to the method
ology of what is called the Producer Price Index Revi
sion ( p p i r ) was essentially completed in January 1986.
This chapter describes this new methodology, which now
is used throughout the p p i “ family” of indexes. Elements
of the traditional methodology remaining in effect are
integrated into the following presentation. Differences
between the p p i r and the traditional methodology are
mentioned where appropriate. However, such contrasts
are not highlighted as much as they had been in the
previous edition of this Handbook (1982), prepared when
the industrial price program was still in transition.2
Description of Survey
Universe
The Producer Price Index universe consists of the out
put of all industries in the goods-producing sectors of the
American economy—mining, manufacturing, agricul
ture, fishing, and forestry—as well as gas and electricity
and goods competitive with those made in the producing
sectors, such as waste and scrap materials. The output
of the services sector is also within the theoretical p p i
universe; although coverage of services currently is
minimal, planning is well underway for considerable
expansion in this area. Imports are no longer included
within the p p i universe; however, the b l s International
Price Program currently publishes price indexes for both
imports and exports. (See chapter 17.) Domestic pro
duction of specifically military goods is now included,
as are goods shipped between establishments owned by
the same company (termed interplant or intracompany
transfers).
Prices
One of the most crucial tasks in preparing any price
index is to define what constitutes the “ price” whose
changes are to be measured. A seemingly simple ques
tion such as “ What is the price of steel?” is unanswerable
until it is made more specific.
For the purposes of the industrial price program, a
price is defined as the net revenue accruing to a specified
producing establishment from a specified kind of buyer
for a specified product shipped under specified transac
tion terms on a specified day of the month. This defini
tion points up the several price-determining variables that
must be clarified before a cooperating business estab
lishment can report a meaningful price for any of its
2 Bureau of Labor Statistics, BLS H a n d b o o k o f M eth o d s, Vol. 1,
Bulletin 2134-1 (1982).
products to b l s . For example: If a company charges
more for a red widget than a white one, color is one of
the price-determining variables; if all widgets sell for the
same price regardless of color, color is not a price
determining variable.
Because the p p i is meant to measure changes in net
revenues received by producers, changes in excise taxes—
revenues received by the government—are not reflected.
But changes in rebate programs, low-interest financing
plans, and other sales promotion techniques are reflected
to the extent that these policies affect the net proceeds
ultimately realized by the producer for a unit sale. If an
auto manufacturer offers retail customers a rebate of
$500, the manufacturer’s net proceeds are reduced by
$500, and the p p i for new cars would reflect a lower
price. (Conversely, termination of a rebate program
would be treated as a price increase.) But if a retail
car dealer offers retail customers an additional rebate
whose cost is absorbed by the dealer rather than the
manufacturer, such a rebate would not affect the p p i .
(The Consumer Price Index, of course, would reflect a
customer rebate regardless of whether it was sponsored
by the manufacturer or the dealer.)
The statistical accuracy of Producer Price Indexes
depends heavily on the quality of the information volun
tarily provided by respondents, b l s emphasizes to
cooperating businesses the need for reports of realistic
transaction prices, including all discounts, premiums,
rebates, allowances, etc., rather than fictitious list or book
prices. The use of list prices in the industrial price pro
gram has been the exception, not the rule. Even before
the conversion to the methodology of the Producer Price
Index Revision, a b l s survey showed that only about 20
percent of traditional commodity indexes were based on
list prices. Inasmuch as the PPIR methodology is more
systematic than the traditional methodology in concen
trating on actual transaction prices, the use of list prices
is even less frequent now.
Neither order prices nor “ futures” prices are ordinarily
included, because the p p i tries to capture the selling
price for output being shipped in that same month, not
some time in the future. Changes in transportation fees
will be reflected in industry price indexes only when the
producing company (rather than a commercial shipper
or a contractor) receives revenues for delivering products
to the buyer.
Most prices refer to one particular day of the month,
viz., the Tuesday of the week containing the 13th of the
month; this pricing date can range between the 9th and
the 15th. There are some exceptions, however. Prices for
a number of farm products are for a day of the week
other than Tuesday. Prices for some refined petroleum
products are commonly an average of prices during the
first half of the month rather than the prices received by
oil refineries on a given day. Price indexes for natural
gas and some industrial chemicals are still based on data
126
for the calendar month as a whole and, therefore, lag 1
month behind other indexes. A natural gas index for
November, for example, would reflect price changes that
actually occurred in October.
Although most prices reported to the Bureau are the
selling prices of selected producers, free on board (f.o.b.)
point of production, some prices are those quoted on
organized commodity exchanges or at central markets;
this practice is most often found among farm products.
Product change and quality adjustment
Although the same product usually is priced month
after month, it is necessary to provide a means for bridg
ing over changes in detailed specifications so that only
real price change will be measured. An adjustment is
especially important when one product is replaced by a
new one. Even when companies report their selling prices
based on altered transaction selling terms (e.g., price per
1,000 sold instead of price per 100), or when there is a
change in the number or identity of companies reporting
to b l s , routine steps can be taken to ensure that only true
price changes influence the index.
When a company respondent reports a price that
reflects a physical change in a product, the Bureau uses
one of several quality adjustment methods. The direct
comparison method is used when the change in the
physical specification is so minor that no product cost
differences result; in this instance, the new price is directly
compared to the last reported price under the former
specifications, and the affected index reflects any price
difference.
When changes in physical characteristics of a product
cause product cost differences, however, the Bureau
attempts to make an accurate assessment of real price
change by taking systematic account of quality dif
ferences. The explicit quality adjustment method is crucial
for automobiles, machinery, and other types of goods
that undergo periodic model changes. The usual method
for quality adjustment involves the collection of data
from reporting companies on the costs they have incurred
in connection with the quality change. If the selling price
of a new model car is $500 more than the previous model
year’s version, but $200 of that increase is due to the extra
product cost and normal margin associated with the addi
tion of government-mandated safety equipment, then the
real price has only risen by $300; the change in the
passenger car index will reflect only that amount, not the
nominal price rise of $500. Admittedly, there are several
problems in applying this procedure in an economically
meaningful fashion; for example, some improvements in
quality and performance may actually cost the producer
less than the technology of an older, inferior product did.
The Bureau has been actively exploring the use of alter
native, hedonic quality adjustment procedures, but with
limited results to date. Any inability to reflect technical
change em bodied in new products im plicitly im parts a
bias o f unknow n m agnitude and direction to p p i data.
Unfortunately, it is not always possible to obtain a
value for quality adjustment if, for example, the respond
ent is unable to estimate the production cost difference
between an old item and a new one, or if an explicit com
parison between an entirely new product and a previous
product is not feasible. In such cases, the Bureau may
have to assume that any difference in price between the
old and the new items is entirely due to quality adjust
ment; the Bureau, therefore, employs the “ overlap”
method (if possible). Under this method, the Bureau col
lects prices for both the old and the new item over a
period of time and chooses 1 month as the overlap month.
In this overlap month, any difference between the price
levels of the two items is factored out. For purposes of
calculating the official price index, the Bureau uses price
changes for the old item through the overlap month but
thereafter follows price changes only for the new item.
Data Sources and Collection Methods
One of the fundamental differences between the tradi
tional p p i methodology and the p p ir methodology is the
switch from judgmental selection of companies and prod
ucts to probability sampling. Under the traditional
methodology, bls would normally ask the largest com
panies producing a given type of commodity to report
prices for their best-selling products. The p p i was, there
fore, too heavily composed of volume-selling products
made by major producers. This selection system, while
convenient, missed much of the economy’s flexibility and
dynamism by overlooking the behavior of medium- and
small-sized firms and the strategies for pricing mediumand small-volume products.
Under the sample design procedures of the Producer
Price Index Revision, the industry as a whole is the basic
starting point for sampling. Each industry has an
individually designed and tailored sample. The first step
in selecting a sample is to construct a universe frame of
establishments classified within that industry. The
primary source for compiling this universe of establish
ments is the data from the Unemployment Insurance
System, because virtually every employer is legally
required to be a member. Supplementary information
from multiple, publicly available lists is used to refine the
industry’s frame of establishments, e.g., by eliminating
firms that have gone out of business.
An establishment is defined as a production entity in
a single location. Two establishments may occupy the
same or adjacent space if they are separable by physical
identification, recordkeeping, or both. Establishments are
the units for which production and employment data are
collected; however, establishments may frequently not be
the appropriate unit for the collection of producer price
127
data. An establishment may be one of several owned
by a single business firm and operated as a cluster,
constituting a profit center; in such cases, the business
maximizes profit over the cluster as a whole rather than
for any one establishment.
The second step in constructing an industry sample
consists of clustering establishments into price-forming
units. Each member of a price-forming unit must belong
to the same industry; establishments in a profit center that
belong to other industries must be excluded in this step.
Once a list of price-forming units in an industry has
been constructed, the list must be stratified by variables
appropriate for that industry. The criterion for identify
ing the sampling strata is whether price trends may be
different for different values of a variable. For example,
the size of the production unit may cause differences in
production technologies and, thus, different responses to
changes in demand or input costs. Some industries may
be characterized by geographically independent markets,
which should become strata. Within each stratum, units
are usually ordered by size to ensure a proportionate
distribution of the sample.
The fourth step is to assign the number of units to be
selected in each stratum. Normally, this assignment is in
direct proportion to the value of shipments by units in
each stratum. However, if there is evidence that some
strata have more heterogeneity in price change, these
strata will be assigned a greater proportion of the total
sample than their simple shipment values would require.
Each price-forming unit is selected systematically with a
probability proportionate to its size. Ideally, the proper
measure of size would be the total revenue of the unit;
however, in practice, employment is used as a proxy
because employment information is usually more readily
available.
Once an establishment or cluster of establishments is
selected for pricing, a b l s field economist visits the unit
to solicit its cooperation. The officials of the unit are
assured that their assistance is completely voluntary, and
that any information they agree to provide to b l s will
be safeguarded under the strictest guarantees of confiden
tiality. If the officials agree to participate in the Producer
Price Index program, the BLS field economist proceeds
to select those transactions to be priced through time from
among all the unit’s revenue-producing activities. A prob
ability technique called disaggregation is used to select
those transactions. The disaggregation procedure assigns
to each category of items shipped, and to each category
of other types of receipts, a probability of selection pro
portionate to its value within the reporting unit. The
categories selected are broken into additional detail in
subsequent stages until unique items, or unique types of
other receipts, are identified.
Even after a physically unique item has been deter
mined, it may be necessary to disaggregate further. If the
same physical item is sold at more than one price, then
the conditions that determine that price—such as the size
of the order, the type of customer, etc.—must also be
selected on the basis of probability. This method for iden
tification of terms of sale (or transaction terms) both
ensures that the same type of transaction is priced over
time and eliminates any bias in the selection of the sales
terms.
To minimize the reporting burden on cooperating com
panies, the disaggregation process described above usually
is completed within 2 hours in the initiation interview.
Subsequently, reporting companies agree to supply prices
for those items selected on an agreed-upon schedule,
usually monthly but sometimes less often, b l s Form
473P, shown at the end of this chapter, is used for re
porting producer prices; it generally takes less than 30
minutes to fill out these repricing forms. Cooperation
generally remains high, although some companies decline
to participate from the beginning and others may drop
out of the program.
The BLS sample of each industry’s producers and out
put must be updated every few years to take account of
changing market conditions. This procedure, called
“ resampling,” takes place relatively often for industries
marked by dynamic changes in production technology or
industry structure. More stable industries need to undergo
resampling less frequently. In practice, many of the
reporting companies may be the same both before and
after resampling; likewise, some individual products in
the sample may also be the same. The resampling pro
cess gives the Bureau the systematic opportunity to keep
the p p i system as up to date and useful as possible.
Estimating Procedures
W eights
If the Producer Price Index system were composed
merely of indexes for individual products, with no group
ing or summarization, there would be no need to devise
a comprehensive weight structure. However, given the
desire for numerous indexes for groupings of individual
products, there is a need for a weight system that will
let more important products have a greater impact on
movements of groupings. Without a weighting structure,
a 10-percent rise in automobile prices would have the
same significance as a 10-percent rise in apple prices.
This section first describes the weighting policies for
the industry and product indexes of the Producer Price
Index Revision. The remainder of the section discusses
the weighting structure of the traditional commodity
groupings portion of the p p i family of indexes.
Item and product aggregation weights. A price index
for even the most finely detailed product (usually termed
a “ cell index” ) cannot be calculated without applying a
128
the input-output tables compiled by the Bureau of
Economic Analysis of the U.S. Department of Commerce,
and other detailed industry data. Currently, industry price
indexes continue to be calculated primarily with 1977 net
output weights and input-output relationships.
policy for weighting the individual price reports received
by b l s for each item. Under the current p p i r method
ology, reports of some establishments are given more
weight than those from others in calculating each cell
index. Item weights are assigned by b l s on the basis of
data on shipment values provided to b l s field representa
tives during the initiation interviews with cooperating
establishments, adjusted by b l s probability selection tech
niques. (Prior to 1978, almost all price reports used to
calculate any given cell index would implicitly be weighted
equally, regardless of any differences in size among the
reporters; if, for example, five companies provided prices
for a certain commodity each month, each of these five
reports would have had a weight of one-fifth.)
To calculate price indexes for levels of aggregation
above the cell index, b l s compiles weights based on
values of shipments for those aggregations of products
made within the same industry; thus, shipment values for
the same products made in other industries do not enter
the weighting structure. Data on values of shipments are
derived from information provided by the Bureau of the
Census and a few other sources.3 The total value of
shipments for each industry is distributed among the
products or other revenue sources produced by that
industry, thereby eliminating the need for any indirect
imputations of weight, a common practice under the pre1978 methodology of the Producer Price Index.
Industry net output weights. In compiling price indexes
for 4-digit sic industries, as well as for even more highly
aggregated industry group indexes, b l s employs net out
put values of shipments as weights. Net output shipment
values refer to the value of shipments from establishments
in one industry to establishments classified in another
industry. By definition, then, net output shipment values
differ from gross shipment values by excluding shipments
among establishments within the same industry, even if
those establishments are owned by separate and independ
ent firms. The meaning of “net output” depends on the
context of the index grouping. The net output for total
manufacturing, for example, would be the value of manu
factured output shipped outside the entire manufacturing
sector, e.g., to the construction sector or to consumers.
In addition to the value of shipments data supplied by
the Census o f Manufactures, b l s also constructs appro
priate net output price indexes through the use of infor
mation on the value of materials consumed (also from
the Census Bureau), data on detailed industry flows from
3 Information currently used for calculating weights throughout the
PPI family of indexes is largely taken from the following censuses con
ducted by the Bureau of the Census of the U. S. Department of Com
merce: (1) the Census o f M anufactures-, (2) the Census o f M ineral
In dustries (which includes oil and gas production); (3) the C ensus o f
Agriculture-, and (4) the Census o f W holesale Trade. Other current
weight sources include the Edison Electric Institute and the National
Marine Fisheries Service.
Weights for traditional commodity groupings. Weights
for individual commodity price indexes, and in turn for
commodity grouping price indexes, are based on gross
value of shipments data, as compiled by the Bureau of
the Census and a few other sources. These weights repre
sent the total net selling value of goods produced or proc
essed in the United States, f.o.b. production point,
exclusive of excise taxes. Since January 1987, shipment
values between establishments owned by the same com
pany (termed interplant transfers) have been included in
commodity and commodity grouping weights; inter plant
transfers had been excluded from the weight structure
before then.
Commodity and commodity grouping weights are
updated periodically to take into account changing pro
duction patterns. Since January 1987, these weights have
been derived from the total net selling value of com
modities reported in the 1982 economic censuses. From
January 1976 through December 1986, 1972 shipment
values formed the foundation for commodity and com
modity grouping weights. Updated weights are incor
porated into the p p i system in a manner that does not
require recalculation of indexes for earlier periods.
b l s does not publish the actual values used as weights,
but does publish what is called a relative importance for
each commodity and commodity grouping. The relative
importance of an item represents its basic value weight,
including any imputations, multiplied by the relative of
price change from the weight date to the date of the
relative importance calculation. The result is expressed
as a percentage of the total for all commodities. Data
showing the relative importance of commodity groupings
with respect to the three major stage-of-processing group
ings are also available.
b l s calculates relative importance data each December,
so that the impact of any additions or deletions to the
sample can be reflected. Except when entirely new weights
are introduced from the latest industrial censuses, or when
there are sample changes affecting a given grouping at
midyear, relative importance data usually change from
one December to another because of relative price move
ments. A commodity whose price rises faster than the all
commodities index from one December to the next will
have a higher relative importance (abstracting from any
sample changes); conversely, a commodity whose price
falls or rises less than the all commodities index will show
a smaller relative importance. Relative importance data
are not used, however, as fixed inputs by the Bureau to
calculate monthly price indexes. Rather, each com
modity’s actual weight value fluctuates each month in
129
accordance with its price movements. Theoretically, the
Bureau could calculate and publish a new set of relative
importance data every month. Relative importance data
for any given commodity grouping also change when its
components are subjected to a sample change.
Index calculation
In concept, the Producer Price Index is calculated
according to a modified Laspeyres formula:
Ii = (E Q aP/£ QaPo)
X 100
where:
PQ is the price of a commodity in the comparison period;
Pj is its price currently; and
Qa represents the quantity shipped during the weight-base
period.
An alternative formula more closely approximates the
actual computation procedure:
Ii = [(£ Q .P„ < P / P „ )) / E QaP„] x '00
In this form, the index is the weighted average of price
relatives, i.e., price ratios for each item (Pj /P D). The
expression (Qa P 0) represents the weights in value form,
and the P and Q elements (both of which originally relate
to period “ a” but are adjusted for price change to period
“ o” ) are not derived separately. When specifications or
samples change, the item relatives must be computed by
linking (multiplying) the relatives for the separate periods
for which the data are precisely comparable.
Analysis and Presentation
Classification
The Producer Price Index family of indexes consists
of several major classification systems, each with its own
structure, history, and uses. However, indexes in all
classification systems now draw from the same pool of
price information provided to BLS by cooperating com
pany reporters, and virtually all indexes are now calcu
lated consistent with the methodology of the Producer
Price Index Revision. The three most important classifica
tion structures are: (1) industry; (2) commodity; and (3)
stage of processing.
Industry classification. A Producer Price Index for an
industry is a measure of changes in prices received for
the industry’s output sold outside the industry (that is,
its net output). As previously stated, the sic 4-digit
industry code is the basis for the industry price index
system. Price indexes have also been available since 1985
for many more highly aggregated industry series at the
3- and 2-digit levels, as well as for total mining industries
and total manufacturing industries.
From the beginning of the transition to the p p i r
methodology nearly every 4-digit industry price index has
been accompanied by detailed indexes representing price
movements for the various products made in that
industry. Code numbers for these indexes at the 5-digit
(product class) and the 7-digit (individual product) levels
often follow the codes and titles established by the Census
Bureau as of 1977 as extensions of the sic structure.
Sometimes, however, b l s assigns its own codes and titles.
In general, there may be as many as three kinds of
product price indexes for a given industry. Every industry
has primary product indexes to show changes in prices
received by establishments classified in the industry for
products made primarily, but not necessarily exclusively,
by that industry. The industry under which an establish
ment is classified is determined by those products
accounting for the largest share of its total value of
shipments. In addition, most industries have secondary
product indexes to show changes in prices received by
establishments classified in the industry for products
chiefly made in some other industry. Finally, some
industries may have miscellaneous receipts indexes to
show price changes in other sources of revenue received
by establishments within the industry.
Commodity classification. The commodity classifica
tion structure of the Producer Price Index organizes pro
ducts by similarity of end use or material composition,
regardless of whether these products are classified as
primary or secondary in their industry of origin. This
system is unique to the PPI and does not match any other
standard coding structure such as the SIC or the United
Nations Standard International Trade Classification.
Historical continuity, the needs of index users, and a
variety of ad hoc factors were important in developing
the p p i commodity classifications.
Fifteen major commodity groupings (2-digit level)
make up the all commodities index. Of these, 2 major
commodity groupings form the index for farm prod
ucts and processed foods and feeds, while the other
13 are grouped into the industrial commodities price
index. Each major commodity grouping includes (in
descending order of aggregation) subgroups (3-digit),
product classes (4-digit), subproduct classes (6-digit), and
individual items (8-digit). The structure of the traditional
commodity classification system thus follows a strict, con
sistent hierarchy.
Corresponding indexes. Nearly all 8-digit commodities
under the traditional commodity coding system are now
derived from corresponding industry-classified product
indexes. In such instances, movements in the traditional
commodity price indexes are identical to movements of
their counterparts. Although most traditional commodity
price indexes continue to be published on their own
original base period, the corresponding industry product
130
commodities that have been processed that still require
further processing. Examples of such semifinished goods
include flour, cotton yarn, steel mill products, and lum
ber. The intermediate goods category also encompasses
nondurable, physically complete goods purchased by
business firms as inputs for their operations. Examples
include diesel fuel, belts and belting, paper boxes, and
fertilizers.
Crude materials for further processing are defined as
unprocessed commodities not sold directly to consumers.
Crude foodstuffs and feedstuffs include items such as
grains and livestock. The crude energy goods category
consists of crude petroleum, natural gas, and coal.
Examples of crude nonfood materials other than energy
include raw cotton, construction sand and gravel, and
iron and steel scrap.
The value-weight of a single subproduct class may be
allocated among several different s o p categories to reflect
different classes of buyers. For example, a portion of the
value-weight of the citrus fruits index has been assigned
to the index for crude foodstuffs and feedstuffs to repre
sent the proportion of citrus fruit sold to food processors;
most of the rest of the value-weight for this grouping has
been assigned to the index for finished consumer foods.
The value-weights are the same as those for the subprod
uct classes within the commodity classification scheme.
The allocations of these value-weights to various SOP
categories are currently based on input-output studies for
1972 conducted by the Bureau of Economic Analysis.
Many major stage-of-processing price indexes exist
continuously back to 1947. However, some special group
ings within the s o p system (such as finished goods less
foods and energy) were first calculated in the 1970’s and
have no historical record before then.
price indexes are published on a base of the month of
their introduction. Therefore, monthly percent changes
for corresponding indexes will be virtually identical even
though their respective index levels may differ.
Specifications for products priced under the current
methodology follow Census Bureau definitions and are
considerably broader than those formerly used for tradi
tional commodity indexes. Because companies are now
reporting prices for a broader range of commodity and
transaction-term specifications within a given commodity
index, it is no longer feasible to publish meaningful
average prices for individual commodities, as was some
times possible with the traditional methodology. Price
indexes are now usually calculated by constructing an
index for each reporting establishment’s price and then
averaging these indexes, with appropriate establishment
weights, to derive the commodity index. Under the former
methodology, an average price could be computed
directly from individual company prices. But despite the
broadening of specifications, industry-classified product
indexes are now available in much greater abundance and
detail than was the case with the traditional commodity
price indexes calculated before the conversion to the new
methodology.
Stage-of-processing classification. Stage-of-processing
price indexes regroup commodities at the sub
product class (6-digit) level according to: (1) the class of
buyer; and (2) the amount of physical processing or
assembling the products have undergone.
Within the stage-of-processing system, finished goods
are defined as commodities that are ready for sale to the
final-demand user, either an individual consumer or a
business firm. In national income accounting terminol
ogy, the Finished Goods Price Index roughly measures
changes in prices received by producers for two portions
of the gross national product: (1) personal consumption
expenditures on goods, and (2) capital investment
expenditures on equipment.4 Within the Finished Goods
Price Index, the consumer foods category includes
unprocessed foods, such as eggs and fresh fruits, as well
as processed foods, such as bakery products and meats.
The finished energy goods component includes those
types of energy to be sold to households—primarily
gasoline, home heating oil, and natural gas. The category
for consumer goods other than foods and energy includes
durables such as passenger cars and household furniture,
and nondurables such as apparel and prescription drugs.
The capital equipment index measures changes in prices
received by producers of durable investment goods such
as heavy motor trucks, tractors, and machine tools.
The stage-of-processing category for intermediate
materials, supplies, and components consists partly of
(SOP)
4 The Producer Price Index universe excludes the consumer services
portion of personal consumption expenditures and the structures portion
of investment expenditures.
Other. There are several additional classification struc
tures within the p p i family of indexes. For example,
Producer Price Indexes are available by durability of
product. Allocation of individual commodities to durability-of-product categories (such as durable manufac
tured goods and total nondurable goods) is based on the
Census Bureau definition: Products with an expected
lifetime of less than 3 years are classified as nondurable,
while products with a longer life expectancy are con
sidered durable goods. Special commodity grouping
indexes (such as fabricated metal products and selected
textile mill products) rearrange p p i commodity data into
different combinations of price series. In 1986, BLS began
publication of indexes measuring changes in prices of
material inputs to construction industries.
Most Producer Price Indexes, whether commodityoriented or industry-oriented, are based on a national
sample of producers because most output is destined for
a national market. Differences in transportation costs to
buyers in different parts of the country are normally
excluded by definition. However, regional price indexes
131
are published for a few selected items, such as electric
power, coal, sand and gravel, scrap metals, and cement,
where regional markets are the rule rather than the
exception.
Analysis
In 1978, as the transition from the traditional method
ology to the methodology of the Producer Price Index
Revision began, b l s decided to shift its analytical focus.
Prior to that time, the Bureau’s economic analysis had
focused on the all commodities index, the industrial com
modities index, and other highly aggregated major com
modity groupings. During the 1970’s, however, it became
clear that these indexes are subject to a major defect: The
multiple counting of price changes. This problem is com
mon among highly aggregated traditional commodity
groupings because they are calculated from price changes
of commodities at several stages of processing, where each
individual price change is weighted by its total gross value
of shipments in the weight-base year.
To illustrate the multiple-counting problem, suppose
that the price of cotton rises sharply. If this price increase
is passed through by spinners of cotton yarn, then by
weavers of gray cotton fabric, then by producers of
finished cotton fabric, and finally by shirt manufacturers,
the single price increase for the raw material cotton would
have been included five times in the all commodities index
and four times in both the industrial commodities
category and in the major commodity group for textile
products and apparel. As long as prices for all items at
all stages of processing are changing at about the same
rate, this multiple counting will not lead to any major
distortions. But if, as is more usually the case, prices are
rising at different rates, multiple counting can result in
rates of change for aggregated price indexes that are
highly misleading, because material prices tend to be more
volatile than finished goods prices are, and because gross
output values are used as weights for major commodity
groups. The rate of increase indicated by the all com
modities index would probably be exaggerated upwards
during inflationary times. When prices are falling, the rate
of decrease for that index would probably be similarly
off-target. In addition, at any given time, there will be
many items showing price increases while other items are
registering price declines; both kinds of changes probably
will be exaggerated (by different degrees) in the all com
modities index. Thus, the net effect of these many dif
ferent biases will be difficult to discern when the economy
is characterized by mixed price movements. (Less
aggregated commodity grouping indexes that cover only
a single stage of processing are not affected by this
multiple-counting defect.)
Stage-of-processing indexes have, therefore, become
the central classification structure used by the Bureau for
analyzing price trends in the general economy. In
particular, the single most important index now stressed
by the Bureau is the Finished Goods Price Index. This
index is crucial because it measures inflation in consumer
and capital goods, upon which demand for materials and
other inputs depends. Both this index and the index of
Crude Materials for Further Processing are largely free
of multiple-counting problems because they are rather
strictly defined. The index for Intermediate Materials,
Supplies, and Components, however, is a residual,
encompassing everything that cannot fit into one of the
other two major stage-of-processing categories. This
index, therefore, includes several different stages of proc
essing (three such stages in the shirt example above) and
is affected by the multiple-counting problem.
The Bureau is focusing more on the price indexes for
the net output of industries as a better solution to the
problems inherent in aggregated price indexes based upon
a weighting structure using gross shipment values.
Presentation
Producer Price Indexes are usually issued on the second
or third Friday of the month following the reference
month. The monthly p p i news release, available without
charge from the Bureau, shows the most recent originally
released and revised data for all stage-of-processing
indexes and for selected major commodity groupings that
comprise the bulk of the s o p indexes. All indexes in the
news release are presented as not seasonally adjusted, but
seasonally adjusted monthly percent changes are shown
for many series as well; price changes over the last 12
months are also included. Even though the news release
can display only a limited number of p p i series, all Pro
ducer Price Indexes are available and considered officially
published at the time of the release.
The monthly detailed report, Producer Price Indexes,
is printed 3 or 4 weeks after the news release date and
is available to the public from the Superintendent of
Documents, U.S. Government Printing Office, on a
subscription basis. The monthly detailed report currently
includes every not seasonally adjusted index within the
p p i family that is publishable, along with some monthly
and annual percent change calculations. Some seasonally
adjusted indexes and monthly percent changes are also
shown. The report also contains a narrative section
explaining the most significant price movements within
major stage-of-processing and industry groups for that
month. When appropriate, additional narratives explain
the latest sample changes (usually effective in January and
July), updates in seasonal adjustment factors or weights,
or other changes in methodology or presentation. Occa
sionally, a longer narrative section delves more deeply
into the economic background underlying recently
observed price movements. A subscription to this
periodical also includes an annual supplement. This sup
plement, commonly mailed to subscribers in the summer
132
the year after the reference year, provides all publishable
indexes and their annual averages for the calendar year,
as well as tables of relative importance data effective for
December of that year. Neither the monthly periodical
nor the annual supplement includes information on actual
dollar prices for any item.
Printouts of tables of historical price indexes for any
PPI series are available, usually without charge, from the
Bureau on request. Two computer tapes are available at
cost; one shows complete historical tables for all
individual commodities and commodity groupings, stageof-processing groupings, durability-of-product groupings,
and other indexes from traditional p p i structures, and the
other shows complete historical records for industry and
product indexes classified according to the sic and the
Census product codes. Complete historical records are
also available on microfiche at cost, bls has recently
made available to the public monthly diskettes showing
the latest monthly values and the previous 12 months of
data for most series included within the p p i news release.
p p i data may now also be accessed electronically on an
on-line basis through the bls Electronic News Service or
through a variety of data bases maintained by private
firms.
Seasonally adjusted data. Because price data are used
for different purposes by different groups, BLS publishes
seasonally adjusted as well as unadjusted data each
month. For analyzing general price trends in the
economy, seasonally adjusted data are usually preferred
because they are designed to eliminate the effect of
changes that normally occur at about the same time and
in about the same magnitude each year—such as price
movements resulting from normal weather patterns,
regular production and marketing cycles, model
changeovers, seasonal discounts, and holidays. Seasonally
adjusted data, therefore, reveal more clearly the underly
ing cyclical trends or unusual disturbances in normal
seasonal patterns (such as severe weather conditions).
Data that are not seasonally adjusted are of primary
interest to those who need information which can be more
readily related to the dollar values of transactions. For
example, unadjusted data are normally used in price
escalation clauses of long-term sales or purchase
contracts.
Producer Price Indexes may be seasonally adjusted at
various levels of aggregation if statistical tests show there
is a significant pattern of seasonal price changes, and
there is a genuine economic rationale supporting the
perceived seasonality. The Bureau’s economic analysis of
the p p i is normally based on seasonally adjusted data,
although unadjusted data are used when tests show an
absence of significant seasonality. Seasonal adjustment
factors are recalculated when January indexes are released
each February, as data for the most recent calendar year
are reflected for the first time and data for more distant
periods are disregarded. This recalculation of seasonal
factors leads to the revision of all seasonally adjusted
indexes for the immediately preceding 5 years. BLS uses
the X - ll seasonal adjustment method to compute
seasonal factors.
Because seasonal adjustment is a tool for enhancing
economic analysis, some indexes that are affected by the
multiple-counting problem described earlier (such as the
all commodities index and the indexes for the major com
modity groups) are deliberately not calculated on a
seasonally adjusted basis.
Revised data. All Producer Price Indexes are routinely
subject to revision only once, 4 months after original
publication, to reflect late reports and corrections by com
pany respondents. Once revised, indexes are considered
final. The Bureau does not use the term “ preliminary”
to describe the originally released p p i numbers, because
“ preliminary” usually describes data that are based on
a small sample of information and that are typically sub
ject to large revisions. When Producer Price Indexes are
first released, they are typically based on a substantial
portion of returns from respondents; hence, subsequent
revisions are normally minor, especially at the more
highly aggregated grouping levels. Either “ first pub
lished” or “ originally released” would be a more appro
priate term than “ preliminary.” Changes in previously
published data caused by a processing error are so indi
cated in a subsequent news release and/or detailed report.
Calculating index changes. Movements of price indexes
from one month to another should usually be expressed
as percent changes rather than as changes in index points
because index point changes are affected by the level of
the index in relation to its base period, while percent
changes are not. Each index measures price changes from
a reference period which is defined to equal 100.0; at this
writing, 1967 is the standard base period for most p p i
series, but many indexes that began after 1967 are based
on the month of their introduction. The following tabula
tion shows the computation of index point and percent
changes.
Index point change
Finished Goods Price Index........................................ 288.5
Less previous index .................................................... 285.0
Equals index point change......................................... 3.5
Index percent change
Index point change...................................................... 3.5
Divided by previous index ..........................................285.0
Equals..........................................................................0.012
Results multiplied by 100................................ 0.012 x 100
Equals percent change................................................ 1.2
An increase of 188.5 percent from the reference base
period in the Finished Goods Price Index, for example,
is shown as 288.5. This change can be expressed in
133
dollars as follows: Prices received by domestic producers
of a systematic sample of finished goods have risen from
$100 in 1967 to $288.50 today. Likewise, a current index
of 300.0 would indicate that prices received by producers
of finished goods today are triple what they were in 1967.
From time to time, the Bureau updates its standard
reference base period. The switch to the 1967 = 100
base occurred in January 1971; before that, the years
1957-59 were used as the standard reference base. For
reasons explained above, any switch of standard reference
base periods does not affect calculations of percent
change for any index. However, care must be taken to
ensure that indexes on one base period are not being in
correctly compared against indexes for the same series
expressed on a different base period. In 1988, the new
standard reference base period for the Producer Price
Index family of indexes is scheduled to become
1982= 100. Index series that began after January 1982,
however, will continue to be based on the month of their
introduction.
Uses and Limitations
Producer Price Indexes are used for many purposes
by government, business, labor, universities, and other
kinds of organizations, as well as by members of the
general public.
The Finished Goods Price Index is one of the Nation’s
most closely watched indicators of economic health.
Movements in this index are often considered to presage
similar changes in inflation rates for retail markets, as
measured by the Bureau’s Consumer Price Index. While
this may sometimes be the case, there are many reasons
why short-term movements in the p p i and the CPI may
diverge. For example, the Finished Goods Price Index
by definition excludes services, which constitute a major
portion of the c p i . The Producer Price Index does not
measure changes in prices for imported goods, but the
Consumer Price Index does include imports. Conversely,
the c p i does not capture changes in capital equipment
prices, a major component of the Finished Goods Price
Index. Large swings in producer prices for foods and
other items may be considerably dampened by the time
retail prices are measured, as retailers or other distributors
absorb price volatility rather than pass through wide price
variations to consumers.
Other stage-of-processing price indexes besides the
Finished Goods Price Index are used for general economic
analysis. Because prices for food and energy have tended
to be so erratic in recent years, some economists prefer
to focus attention on an index such as finished goods
other than foods and energy as a better measure of the
so-called “ underlying rate of inflation.’’ The index for
Intermediate Materials, Supplies, and Components is
closely followed as an indicator of material cost pressures
that may later appear in the Finished Goods Price Index
and/or the c p i . The index for crude materials other than
foods and energy is quite sensitive to shifts in total
demand and can be a leading indicator of the state of
the economy; its limited scope, however, makes it less
reliable as an indicator of the future status of inflation
in general. The stage-of-processing structure is especially
well suited for facilitating economic analysis of the infla
tion transmission process. Hence, it can be used for
analyzing the impact of government stabilization policies.
Producer Price Index data for capital equipment are
used by the U.S. Department of Commerce to help calcu
late the gross national product ( g n p ) deflator and many
of its component deflators, p p i data at all levels of indus
try and commodity aggregation can be used to deflate
dollar values expressed in current dollars to constantdollar values for a variety of economic time series, such
as inventories, sales, shipments, and capital equipment
replacement costs. To illustrate the deflation concept,
suppose that nominal shipment values for a given industry
have doubled over a 10-year span. If the Producer Price
Index for that same industry has tripled over the same
span, then the “ real” (i.e., inflation-adjusted) value of
shipments for that industry has actually declined; higher
prices more than account for the doubling of dollar ship
ment values, and physical volume has implicitly fallen.
Private business firms use p p i data to assist their
operations in a variety of ways, in addition to using these
figures for general economic analysis or deflation as
discussed above. Producer Price Indexes are frequently
cited in price escalation clauses of long-term sales or pur
chase contracts as a means to protect both the buyer and
the seller from unanticipated inflation or deflation.
Typically, an escalation clause will specify that the price
for x number of widgets being sold by company A to
company B each year will go up or down by a specified
fraction of the percentage of change in material costs,
as measured by one or more specified Producer Price
Indexes (often in conjunction with the change in an
average hourly earnings indicator, used to measure labor
costs). Hundreds of billions of dollars in contract values
are tied to Producer Price Indexes through these price
escalation clauses; such clauses are common in both
government and private sector contracts.
Private companies can also use p p i data to compare
changes in material costs they incur against changes in
the p p i for that material. By the same token, they can
compare changes in the selling prices they charge for their
own output to changes in the p p i for the same kind of
product, p p i inform ation may be employed in
econometric models, in forecasting, in market analysis,
and in academic research, p p i ’s are frequently used in
l if o (Last-In, First-Out) inventory accounting systems
by firms wishing to avoid the kind of “ phantom profits”
that might appear on their books with a FIFO (First-In,
First-Out) system.
134
Those wishing to follow p p i data for a particular
series over a prolonged time span should be aware that
highly detailed indexes are more vulnerable to being
discontinued by b l s than aggregated indexes. During the
industry resampling process described earlier, for exam
ple, the industry index (4-digit level) is commonly kept
continuous before and after the resampling process is
completed, while indexes for detailed products within
that industry may be discontinued and replaced by
items that are new or that had not been selected for pric
ing before. Finely detailed indexes are also vulnerable to
temporary suspension of publication. The Bureau’s rules
against disclosure of confidential information preclude
publication of indexes when fewer than three companies
report prices for a given product. Even if there are three
reported prices for a given product in any given month,
the Bureau will ordinarily publish that index only if at
least two of those prices are considered good (i.e., not
estimated) and if a single reporter does not account for
more than half of the market for that product. When a
detailed index disappears either temporarily or per
manently, the Bureau routinely recommends that users
who had been following that index either choose another
detailed index within the same product grouping or else
switch their attention to a more highly aggregated group
ing index.
Technical References
Archibald, Robert B. “On the Theory of Industrial Price
Measurement: Output Price Indexes,” Annals of Eco
nomic and Social Measurement, Winter 1977.
Clem, Andrew, and Thomas, William. “ New Weight Struc
ture Being Used in Producer Price Index,” Monthly
Labor Review, August 1987.
Council on Wage and Price Stability.
Index, June 1977.
The Wholesale Price
Early, John F. “ Improving the Measurement of Producer
Price Change,” Monthly Labor Review, April 1978.
Gousen, Sarah; Monk, Kathy; and Gerduk, Irwin. Pro
ducer Price Measurement: Concepts and Methods. U.S.
Department of Labor, Bureau of Labor Statistics,
June 1986.
Howell, Craig, and Thomas, William. Escalation and Pro
ducer Price Indexes: A Guide for Contracting Parties,
Report 570. U.S. Department of Labor, Bureau of Labor
Statistics, 1979.
Popkin, Joel. “Integration of a System of Price and Quantity
Statistics with Data on Related Variables,” Review of
Income and Wealth, March 1978.
Tibbetts, Thomas R. “An Industrial Price Measurement
Structure: The Universe Matrix of Producers and
Products,” 1978 Proceedings of the Section on Survey
Research Methods. American Statistical Association,
Washington, DC, 1979.
135
U, S Department of Labor
Bureau of Labor Statistics
Information for the Producer Price Indexes
T h is rep o rt is authorised by law , 2 9 U. S. C. 2 . Your
vo lun tary cooperation is needed to m ake th e m oults
o f th is s u r v e y com prehensive, accurate, and tim ely.
The Inform a tion collected on th is form by the Bureau of
Labor S tatistics w ill be held in the strictest confidence
and w ill be used fo r statistica l purposes only.
Form Aporoved
.0 M. 8 No 1220-0003
INSTRUCTIONS
IT E M DESCRIPTION
Please d e te rm in e if th e in fo rm a tio n yo u r
c o m p a n y p r e v i o u s l y p r o v t d a d is
c u rre n tly ap p lica b le o r re q u ire s update,
if re v is io n is re q u ire d , please in d ica te
th e cha n g e s in th e open areas
If yo u re vise th e de scr i p tio n, In d ic a te th e
d a te o n w h ic h th e c h a n g e b e c a m e
e ffe c tiv e a n d th e e s tim a te d v a lu e o f th e
ch a n g e (ch ange in cost p lu s sta n d a rd
m a rkup.)
P lease re v ie w th e A d ju s tm e n ts to P rice
are a to d e te rm in e if th e a d ju s tm e n ts
a n d re la te d te rm s a re c u rre n t as s h o w n .
A d d itio n a l in s tru c tio n s ap pear on th e
re ve rse side o f th is fo rm .
If yo u ha ve any q u e s tio n s c o n c e rn in g
co m p le tio n o f th is fo rm , plea se call:
;
»
B»s
■" R
e
M
»
-mu
'
'
sf
.
m
mmim
- I l l
"
A
R
f" -i-
k
s
■
PREVIOUS PRICE IN FO R M AT IO N — P lease re v ie w th e P re v io u s P ric e In fo rm a tio i
th a t a re s h o w n . N et tra n s a c tio n p ric e s a re th e m o s t d e s ira b le type o f p ric e , i f in c o rre c t o r i f y t
ty p e o f p rice .
Type o f price yo u re p o rt^
in g p ric e s if a v a ila b le . 'C orrect a n y in c o rre c t p ric e s
e to a d iffe re n t ty p e o f p ric e , in d ic a te th e c u r re n t
immsm
:
H ave y o u m a d e a n y ch a n g e s to th e to m
A s c r i p t i o n c r t h e P r e v io u s P r ic e
In fo rm a tio n de scrib e d above?
D id th e p ric e cha nge b e tw e e n
Please e n te r th e p ric e in th e bo xes b e lo w if th e re ha s be en a cha nge. U se black
only, w ritin g th e n u m b e r a s sh o w n . Please do rtqt type. D O N O T U S E B L U E . I*
a s h ip m e n t, e s tim a te th e e ric s th a t w o u ld have be en ch a rg e d on
I 2 3 4- 5 b 7 8 R 0
C h e c k if th is
is a
CLOSEOUT pr ice on
a n ite m w h ic h
b e in g p h ased ou t.
136
is
Dear Respondent,
T h a n k y o u fo r y o u r c o n tin u in g p a rtic ip a tio n in t h e P ro d u c e r P ric e In d e x (P P I) p ro g ra m . T h e d a ta w h ic h y o u p ro v id e a r e u s e d in c o m p u tin g th e P ro d u c e r
P ric e In d e x e s a n d c o n s titu te th e b a s is fo r a n a ly zin g in d u s tria l price c h a n g e s .
P le a s e u s e th e e n c lo s e d p o s ta g e fre e e n v e lo p e to re tu rn th e p ric in g fo r m s . Y o u r c o n tin u e d c o o p e ra tio n is g re a tly a p p re c ia te d .
C o m m is s io n e r o f L a b o r S ta tis tic s
Instructions for completing a PPI pricing form:
T h e in fo rm a tio n c o n ta in e d o n th is fo rm w a s fu r n is h e d b y y o u r firm in p re v io u s p ric in g p e rio d s . R e v ie w th e in fo r m a tio n c a r e fu lly to v e r ify th a t it re m a in s
c u rre n t. C ro s s o u t a n y in c o rre c t in fo rm a tio n a n d w r it e in all c o rre c tio n s a n d a d d itio n s th a t a re n e c e s s a ry . A n y in fo rm a tio n c o n c e rn in g th e ite m w h ic h
e x c e e d s th e sp a c e lim ita tio n im p o s e d b y th e fo rm is c o n tin u e d o n th e s u b s e q u e n t p a g e a n d s h o u ld als o b e ve rifie d .
Item and Transaction Description
if th e Ite m D e s c rip tio n o r th e T ra n s a c tio n T e r m s , o r b o th , n o lo n g e r a p p ly , a s u b s titu te ite m o r s u b s titu te tra n s a c tio n te r m s s h o u ld b e selected by y o u .
Ite m s u b s titu tio n s h o u ld o n ly o c c u r w h e n th e ite m p re v io u s ly re p o rte d is n o lo n g e r a v a ila b le b e c a u s e it is b e in g o r has b e e n permanently
d is c o n tin u e d . T h e s u b s titu te ite m s h o u ld b e a s s im ila r a s p o s s ib le to th e c u r r e n t ite m a n d s h o u ld b e e x p e c te d to re m a in a v a ila b le fo r sometime. The
s u b s titu te tra n s a c tio n te rm s s h o u ld lik e w is e b e a s sim ila r a s p o ss ib le to th e d is c o n tin u e d tra n s a c tio n te r m s .
R e p o rt th e s e c h a n g e s in th e c lo se s t o p e n a re a a n d p ro vid e c u rre n t price in fo rm a tio n .
Adjustments to Price
F o llo w in g is a list o f th e m o re c o m m o n a d ju s tm e n ts to p ric e . T h e s p e c ific a d ju s tm e n ts o n th e p ric in g fo r m w e r e s e le c te d o r ig in a lly a n d s h o u ld b e
c h a n g e d o n ly w h e n e ith e r th e level o f a n e x is tin g a d ju s tm e n t c h a n g e s o r a n e w a d ju s tm e n t b e c o m e s a p p lic a b le to th e p ro d u c t a n d tra n s a c tio n
d e s c rib e d .
Deductions from price in c lu d e :
1.
2.
3.
4.
S ta n d a rd d is c o u n ts ( C a s h , S e a s o n a l, C u m u la tiv e V o lu m e , Q u a n t ity , a n d T ra d e )
R e b a te s
O t h e r re c u rrin g d is c o u n ts
O th e r n o n re c u rrin g d is c o u n ts (C o m p e titiv e a n d N e g o tia te d )
A d d it i o n s t o p ric e in clu d e :
1 . S u rc h a rg e s
2 . O t h e r re c u rrin g c h a rg e s a d d e d to price
3 . O th e r n o n re c u rrin g c h a rg e s a d d e d to price
T a x e s sh o u ld a lw a y s be e xc lu d e d fro m th e p ric e . If th is e xc lu s io n is n o t p o s s ib le , n o te th is in R e m a rk s .
F r e i g h t c h a rg e s s h o u ld b e e x c lu d e d fro m th e p ric e u n le s s d e liv e ry w a s se le c te d o rig in a lly as p a rt o f th e p ro d u c t. M a k e c h a n g e s if th e c u rre n tly
d e sc rib e d fre ig h t te rm s n o lo n g e r e x is t.
QUESTIONS
A n s w e r w h e th e r y o u h a v e ( " Y E S " ) or h a v e n o t ( " N O " ) m a d e a n y c h a n g e s o r e n trie s to th e ite m d e s c rip tio n o r p re v io u s p ric e in fo rm a tio n .
A n s w e r Y E S or N O d e p e n d in g o n w h e th e r th e s h ip m e n t price c ' th e ite m d e sc rib e d c h a n g e d ( " Y E S " ) b e tw e e n th e t w o d a te s listed o r w h e th e r th e
s h ip m e n t price did n o t c h a n g e ("W O ") d u rin g th is tim e period.
If th e a n s w e r is N O , th e fo rm h a s b e e n c o m p le te d a n d is re a d y fo r m a ilin g . D O N O T E N T E R A P R I C E IF T H E P R I C E H A S N O T C H A N G E D . If the answer is
Y E S , p le a se e n te r th e n e w p rice .
If th e re h a s b e e n a c h a n g e in th e n a m e o r a d d re s s to w h o m th is fo r m s h o u ld be s e n t in th e fu t u r e , e n te r a n (X ) in th e b o x . It is n e c e s s a ry to make the name
a n d a d d re s s c h a n g e o n o n ly o n e fo rm .
Please complete and return within S business days all of the pricing forms that are mailed to you even if there are no changes.
If you are enticipeting a change in any of the information that you provide, please indicate in Remarka. List the anticipated change and when it will occur.
Any questions that you have regarding the pricing forms or their completion may be resolved by calling the Bureeu of Labor Statistics, Division of Industrial
Prices and Price Indexes. Washington, D.C.; by using the telephone number at the top of the pricing forms.
REMARKS:
137
Chapter 17. International
Price Indexes
Background
The International Price Program (IPP) provides, as its
primary output, price indexes for the U.S. merchandise
foreign trade sector. Additional outputs include interna
tional service sector indexes and international trade com
parison measures. They play a crucial role in the analysis
of trends in U.S. trade and are also used in the analysis
of domestic price levels.
Although the Bureau has collected international price
data for some time, the complete series has only been
published since the early 1980’s. The first collection of
import prices took place nearly 100 years ago as part of
a one-time study of U.S. price competitiveness. The next
effort occurred immediately following World War II. To
meet the need for accurate measures of price change
within the expanding U.S. foreign trade sector, b l s
initiated the limited development of export and import
price indexes. However, budget restrictions in 1948 led
to the termination of this effort. As a result, the develop
ment of foreign trade sector price indexes lay dormant
until 1967 when BLS began research on the feasibility of
producing U.S. export and import price indexes. This
research was among the early major products of the Divi
sion of Price and Index Number Research established in
1966 on the recom m endation o f the Price Statistics
Review Committee of the National Bureau of Economic
Research. This research resulted in congressional alloca
tion of funds for i p p in fiscal year 1970.
A limited number of export price indexes were first
published in June 1971 and selected import indexes
followed in 1972. Each of these introductory index series
presented annual values until 1974 when collection and
publication of quarterly price data began.
During the remainder of the 1970’s and into the early
1980’s, i p p steadily expanded its product coverage and
the publication of export-import price indexes. The IPP
goal of 100-percent coverage of the value of exported and
imported goods was reached with the publication of an
index covering all imports in February 1983 and an all
export index in February 1984.
In addition to its coverage of product classification,
i p p has embarked on a project to track price movement
in the services area of the foreign trade sector. Late 1986
saw the initial publication of service sector indexes
representing export air passenger fares and import
electricity. Further expansion is planned in this area.
Import tanker freight indexes were first available in 1987,
and development is underway on measuring price change
in the import air freight, import liner freight, and import
air passenger markets.
The indexes for both exports and imports are reflec
tive of the market basket of internationally traded goods
in the base year 1980. With the release of indexes repre
senting the first quarter of 1988, weights will be updated
to reflect the market basket of goods traded in 1985.
Description of Survey
Concepts
The U.S. export and import price indexes are generalpurpose indexes that measure changes in price levels
within the foreign trade sector. The all-export index pro
vides a measure of price change for domestically pro
duced U.S. products shipped to other countries; the all
import index measures price change of goods purchased
from other countries by U.S. residents. Since the indexes
are intended to be used as a deflator in the National
Income and Product Accounts ( n i p A ), the Bureau has
used balance-of-payments definitions, where appropriate,
in its methodology.
In addition to the all-export and all-import indexes,
indexes are published for a wide variety of product
categories with several levels of detail. Currently, the
Bureau publishes indexes for product categories which
recorded at least $400 million of imports or exports in
1980. Beginning in 1988, the export trade minimum will
be raised to cover product categories with a 1985 value
of $500 million for exports and $700 million for imports.
Indexes that represent areas of trade with smaller dollar
values are incorporated into the calculation of higher level
indexes but are generally not published separately.
Product universe
The product universe of the export-import price indexes
consists of all commodities exchanged between U.S.
residents and foreign residents. (“ Residents” refers to the
n i p a definition; it includes corporations, businesses,
138
and individuals but does not require either specific owner
ship or citizenship.) The export universe consists of prod
ucts sold by U.S. residents to foreign buyers. The import
universe covers products purchased from abroad by U.S.
residents. The universe in each case includes raw
materials, agricultural products, semifinished manufac
tures, and finished manufactures, including both capital
goods (electrical machinery, agricultural equipment, tex
tile equipment, etc.) and consumer goods (appliances,
electronic equipment, clothing, etc.).
Ideally, the total breadth of U.S. trade would be
represented in the product universe. However, items that
present difficulties in obtaining consistent time series
(works of art, commercial aircraft, ships, etc.) for com
parable products are not included. Also, military goods
are not priced in the indexes except to the extent that some
products are purchased on the open market for military
use; e.g., automobiles, clothing, nonspecialized hard
ware, fuel, etc.
Pricing basis
Most prices are collected directly from cooperating
exporters and importers (reporters). To the extent possi
ble, they are prices at the U.S. border for exports and
at both the foreign border and the U.S. border for
imports. The targeted reference period for transaction
prices in the survey is the third month of each calendar
quarter; e.g., March, June, September, and December.
Generally, the Bureau is seeking the first transaction in
the appropriate month, and the vast majority of prices
in the survey fall within the first week of the reference
month. If a firm is unable to provide a price quote for
the targeted reference period, prices for a transaction
from the previous month may be substituted.
Reporters are requested on the price reporting forms
to indicate all discounts, allowances, rebates, etc.,
applicable to the reported prices so that the price used
in the calculation of the indexes is the actual transaction
(or net) price for which the product was bought or sold.
(See copy of the form at the end of this chapter.) During
the quarterly repricing process, respondents are reminded
of this requirement in their repricing mail package and,
if necessary, by a phone contact from an industry analyst
on the i p p staff.
All prices collected are representative of actual trans
actions in the foreign trade market. Average prices are
not accepted in the IPP survey, with the exception of
selected commodities where secondary source data is
used.
i p p ’ s preferred pricing basis for exports is f.a.s. (free
alongside ship) U.S. port of exportation. When reporters
supply export prices that are on an f.o.b. (free on board)
factory basis, information on production point and
freight is collected to enable the Bureau to calculate a
shipment cost to the specified port of U.S. exportation.
This information includes location of production point
and port of exportation, size and weight of shipment,
name of carrier, and routing. This shipment cost can then
be appended to the reported f.o.b. factory price in order
to create a “ net” price that is usable in an f.a.s. index.
These adjustments are reviewed quarterly for significant
changes in transport costs. In addition, to the extent that
f.o.b. factory price data are collected and if shipping costs
are deemed negligible by a freight analyst, f.o.b. factory
price data can be included in an f.a.s. index.
In some product areas, for instance finished manufac
tures, reporters frequently provide prices only on an
f.o.b. factory basis and are unable to supply freight infor
mation. As a result, many of the export price indexes for
these products can only be calculated and published on
an f.o.b. factory basis. Consequently, more aggregated
series may be constructed from detailed categories with
differing price bases.
For imports, i p p ’s preferred pricing basis has been
c.i.f. (cost, insurance, freight) at the U.S. port of impor
tation. However, in product areas where c.i.f. prices are
not generally available, indexes are published on an f.o.b.
foreign port basis. As with exports, aggregated import
series will frequently be produced using detailed groups
with differing price bases, i p p does attempt to collect
import price data on both the c.i.f. and the f.o.b. foreign
port bases. The price on a c.i.f. basis consists of the
foreign selling price plus the other costs (insurance and
freight) associated with bringing the product to the U.S.
border. F.o.b. foreign port prices represent the cost of
the item at the foreign port of exportation, and are con
sistent with the basis for valuation of imports in n i p a ,
where insurance and freight costs are tabulated separately
in the services accounts. Import index calculation is per
formed on a duty excluded basis. Import duty informa
tion, if necessary, is collected separately and, where
appropriate, is deducted from the reported price.
Price index calculation requires that collected price data
reflect the same item from period to period. To ensure
this, the specifications for each product in the i p p survey
include detailed physical and functional characteristics
as well as the terms of transaction; e.g., number of units
bought or sold, class of buyer or seller, etc. Any change
in a product’s specifications or terms of trade is appraised
to ascertain the significance of these changes. If the
changes are cosmetic, product substitution is effected by
direct price comparison, and any reported price change
is reflected in the index. If the changes are substantive,
product substitution is made by linking which ensures that
the index reflects only actual or “ pure” price changes and
is not affected by quality changes. The following
simplified example illustrates the principle of linking: The
June reference period price for a specific imported
automobile was $9,250.88. In the September reference
period, a price of $10,108.77 was supplied for a new
model of this automobile. It was determined that the new
139
model was essentially comparable with the old, except
that it had a bumper assembly that could only withstand
a 2 ‘/2 m.p.h. impact without structural damage to the
vehicle, whereas the former model was equipped with a
5 m.p.h. bumper. For linking purposes, the object is to
determine the “ appropriate” price that the new model
would have commanded in the marketplace in the
previous reference period. In this example, the new model
was estimated to have a June price of $8,955.50
($9,250.88, the June price of the former model, minus
a $295.38 decrease in the value of the bumper assembly).
The price comparison between June and September was
based on the estimated June price of $8,955.50 and the
reported September price of 10,108.77. Thus, a 12.9percent increase was reflected in the September index, but
the price change including the quality decrease (poorer
quality bumper) was not reflected.
Linking is also used when products are added to or
deleted from an index. When a completely new product
series is added to a commodity grouping, the linking pro
cedure discussed above is not feasible. Instead, the
relative importance of each item in the commodity group
is redistributed to include the new item, and the historical
movement of the index is attributed to the new product.
A change in the relative importance of a product also
occurs when products are dropped from an index without
being replaced.
Classification
Export-import price indexes are based on the nomen
clature of Revision 2 of the Standard International Trade
Classification ( s it c ) of the United Nations.1 The sit c is
made up of 10 sections at the 1-digit level, 63 divisions
at the 2-digit level, 233 groups at the 3-digit level, and
786 subgroups at the 4-digit level. Additional subsidiary
classes are available, raising to 1,924 the number of
“ basic terms” in the s i t c .
The lowest level of publication is usually the 4-digit
level across the spectrum of the SITC. Currently, the BLS
publishes 209 export indexes and 236 import indexes.
Although primary indexes are calculated under the
s i t c , prices collected for the survey are classified by the
basic product classification systems used for recording
U.S. foreign trade: For exports—the 7-digit Schedule B
classification system of the U.S. Department of Com
merce;2 for imports—the 7-digit Tariff Schedule of the
United States A nnotated ( t s u s a ) . 3 Concordance
schemes are used for classifying Schedule B and t s u s a
categories into the appropriate s i t c ’s.4
By maintaining these detailed 7-digit product classifica
tions, BLS is able to prepare alternative indexes for group
ings other than those afforded by the SITC. Alternative
groupings in published indexes include the industryoriented Standard Industrial Classification based (sicbased) system and the Bureau of Economic Analysis’
end-use system. These additional indexes offer a wider
variety o f analytical uses for the international price index
data.
Data Sources and Calculation Methods
Collection
Normally, price data are collected by mail question
naire; in some cases the information is collected by
telephone. Most prices are collected quarterly; however,
where a product has a seasonal pattern, collection is
adjusted accordingly. In the overwhelming majority of
cases, prices are collected directly from the exporter or
importer, although in a few cases prices are obtained from
secondary sources.
Price reporting by firms is initiated by a visit from a
Bureau representative. At this time, the reporting
requirements are explained verbally and in writing, and
the selection of products for which the firm will report
price information is made. Information initially provided
by a firm usually contains data for the current and
previous quarter.
Subsequent repricing is conducted quarterly by mail.
However, if data clarification is required, the individual
at the firm who is responsible for providing the price
information is contacted by telephone.
Sampling
The objective of the i p p sample design is to provide
an unbiased measure of price change in each published
index. A multistage survey design is employed to provide
a sample of exporters and importers for specific product
strata as well as specific items which can be repriced over
time. The survey design is responsive to the constraints
of both cost and the burden on reporters. The cost con
straints impose limits on the number of establishments
selected in a sample, while the number of items priced
in each establishment is controlled to limit respondent
burden.
The two universes for the IPP are all exporters and all
importers (and their respective products). A sampling
frame for each universe is constructed from all documents
filed during a specified reference period (generally 1 year).
In the case of exports, these are the Shipper’s Export
1 United Nations Statistical Office, Standard International Trade
Classification, Revised, Statistical Papers, Series M, No. 34/Rev. 2 (New
York, United Nations, 1975).
2 U.S. Bureau of the Census, Statistical Classification o f Domestic
and Foreign Commodities Exported From the United States, Schedule
B, January 1, 1978, edition and revisions.
3 U.S. International Trade Commission, Tariff Schedule o f the
United States Annotated, 1980 edition and revisions.
4 U.S. Bureau of the Census, U.S. Foreign Trade Statistics Classi
fications and Cross Classifications, 1980.
140
Declarations (SED’s) and, in the case of imports, the
Consumption Entry Documents (c e d ). These documents
contain brief product descriptions, 7-digit product
classification codes, value, quantity (where required),
date, origin or destination, company name and address,
and an establishment identification code.
The availability of an establishment identification code
on both export and import records makes it possible to
incorporate frequency of trade (consistency) into the sam
ple design. Companies can be designated as either con
sistent or inconsistent exporters/importers of particular
products. This information is used in each stage of sam
pling with the result being an increased yield of usable
time series prices.
The import frame, obtained from the Customs Bureau,
contains a record of every import transaction. Corre
spondingly, the export frame, provided by the Bureau of
the Census, contains a record of every export transaction.
The sample design for both exports and imports con
sists of three stages. The first stage is the selection of
establishments. The second stage is the selection of Entry
Level Items (ELPs—commodity classes within a sampling
stratum). The third stage is the selection of specific items
(products) in the e l i . The system is identical for both
exports and imports unless otherwise noted below.
The first step is to generate the measure of size (maxprob) for an establishment, as follows. The dollar value
on each document is aggregated to company-ELi,
company-sample stratum, and company levels. It is also
aggregated within an e l i , and within a stratum across all
companies. A proportion is then calculated for each
company-stratum by dividing the aggregated companystratum dollar value by the aggregated dollar value within
the stratum. This “ company-stratum prob” is the pro
portion of dollar value that the company contributes to
the particular stratum. The max-prob for each company
is the maximum “ company-stratum prob” for that com
pany over all strata. In addition to a max-prob, a maxprob stratum (the stratum associated with the max-prob)
is assigned to a company. The companies are then
implicitly stratified by max-prob strata, and a systematic
probability proportionate to size (PPS) selection of com
panies is made using the max-prob as the company’s
measure of size. The principal advantage of max-prob
here is that a company’s chance of selection is based on
the product category for which it is most important; this
is desirable since the indexes are calculated and published
by product category.
When a company is selected in the first stage, it is
selected for all its products, including those outside its
max-prob stratum. In order to lessen the burden on
respondents, it is then necessary to select a second-stage
sample of e l i ’s within each company. The first step of
this second stage is to ensure that publication
requirements are met by selecting company e l i ’s with
certainty in some strata. The remaining e l i ’s in each
com pany are then sam pled using a system atic p p s techni
que. The m easure o f size is the ELi-prob (the com panystratum probability distributed am ong the e l i ’s in the
com pany in p roportion to their dollar value contribu
tion s). This constitutes the sam ple o f respondents and
their selected e l i ’s .
After the sample of companies and e l i ’s has been
selected, further sampling is needed to obtain a specific
item for repricing. Beginning in 1982, a probability selec
tion method, referred to as disaggregation, was
introduced. Under this method, the ELI is partitioned into
subclasses, and a p p s selection is made among the sub
classes using their proportion of trade in the establish
ment as the measure of size.5 The process continues
through the successive subdivisions of each selected
subclass until an identifiable item that can be priced over
time is obtained.
Probability sampling techniques have been used in ip p
since 1976. Prior to that time, samples were based on
nonstatistical, judgment samples of commodity groups.
The judgmental criteria required that items selected in
each company be repriceable and that their price move
ment be representative of the respondent’s other products
in that same ELI. Approximately 10 percent of the cur
rently repriced items in the ip p survey are from this
earlier sampling era; eventually, they will be replaced by
probability-selected products.
Estimating Procedures
Formula
The export and import price indexes are weighted
indexes of the Laspeyres type. Individual product price
relatives are assigned equal importance within each
weight-group category, and the weight-group relatives are
then aggregated to the successive stratum index levels.
E. E.1
j
p l
p°
n i
where:
x = sitc group for which index is calculated
j = the weight categories within x (they are the
Schedule B categories for exports, and the tsusa
categories for imports)
i = product within j
n = number of price relatives within j
t = time
w. = share of the value of jth category in group x in the
base year
P}/Pf = price relative of product i in year t to base year o
5 This “proportion of trade” estimate is provided by the respondent.
141
Weights
The ip p weights represent the total dollar value of U.S.
foreign trade in a designated base year and are distributed
among several thousand 7-digit weight-group categories.
Values assigned to each of these weight-group categories
are based on trade value figures compiled by the Bureau
of the Census for the base year.6 In the case of the
export price index, the 7-digit weight-group category is
the Schedule B, and, in the case of the import price index,
it is the t s u s a . Currently, i p p calculates indexes based
upon 1980 trade figures.
Although earlier published indexes used weights from
either 1969, 1970, 1973, or 1975, all historical data were
revised in February 1982 to reflect 1980 trade weights.
Beginning with the release of the first quarter 1988
indexes, the base period will be updated to 1985, but most
historical data will continue to use 1980 weights.
Each 7-digit weight-group category is considered to be a
relatively homogeneous commodity classification. There
fore, each product, or group of products, selected for the
ip p survey within a weight-group category represents not
only itself, but all products that fall within that weightgroup. Due to budgetary limitations, not all weight-group
categories are represented. Consequently, index trends of
the represented categories are used to impute the move
ment of weight groups which are not covered.
As the i p p product base is strengthened, new weightgroup categories can be introduced into an index. When
this occurs, the new groups are linked to the earlier indexes.
Presentation
All reporting is voluntary and confidential, and, there
fore, no index is published in such a way as to reveal the
name, price, or price behavior of any respondent. No
index is published when fewer than three companies pro
vide data; for the vast majority of indexes there are con
siderably more.
The export and import price indexes are published
quarterly in b l s news releases 5 weeks after the reference
month. The release includes a narrative summary as well
as the SITC end-use product indexes, service sector indexes,
and, beginning with the fourth quarter 1987 release, the
new average rate and nominal currency indexes. In addi
tion, comprehensive and specific historical tables are avail
able for export-import price index data in its SITC form
6 The value data for 7-digit categories are compiled by the Bureau
of the Census using Shipper’s Export Declarations and Consumption
Entry Documents. This information is available on magnetic tape and
can be found in the following Bureau of the Census publications:
Exports:
U.S. Exports—Schedule
B
Commodity by Country,
Report FT-410, December each year.
Imports:
U.S. Imports For Consumption and General Imports—
T S U S A Commodity by Country of Origin, Report FT246, annual.
as well as in its end-use and sic-based counterparts. Sum
mary tables are also published in the Monthly Labor
Review as well as in machine-readable form and on data
diskettes. Where possible, the index base is 1977 = 100.
However, in numerous cases where the price data are not
supportive of a 1977 index base period, the indexes are
presented with a more recent period. With the switch to
1985 weights in 1988, the index base will be shifted to
1985 = 100 for most series.
In addition to the export and import price indexes, a
quarterly report is prepared that updates BLS Bulletin
2046, Comparisons o f United States, German, and Japan
ese Export Indexes. These data present index comparisons
between the United States and the Federal Republic of
Germany, and between the United States and Japan.
These comparisons measure price movements of U.S.
exports relative to those of its major world competitors
for similar commodities.
Uses and Limitations
The indexes published in this program are the only
indexes of prices related to the U.S. foreign trade sector.
They provide quarterly measures of the price trend of U.S.
products sold abroad and of products imported to the
United States from other countries. The series enables ana
lysts and policymakers to assess the effect of export and
import price changes on the U.S. economy and its industrial
sectors, as well as to analyze the effects of price changes
on the balance of payments. The price measures provide
a basis for calculating changes in the volume of exports
and imports in the aggregate and for product groups.
The export price indexes provide a basis for measuring
changes in the prices of U.S. products in relation to price
trends of comparable products of other major industrial
countries with which the United States competes for mar
kets, and for assessing changes in U.S. price competitive
ness. However, prices collected for exports do not include
overseas transportation, foreign duties, or other charges
outside the United States. Presented in their end-use form,
the indexes are used as a deflator for the National Income
and Product Accounts, while their sic-based form serves
as a tool in conducting industry comparison analyses.
Since the import price indexes only measure the value
of a product at a port (either domestic or foreign), special
care must be taken in using this data to assess the impact
of import prices on domestic inflation levels. First, an f.o.b.
foreign port series excludes international freight charges.
Second, both an f.o.b. foreign port and a c.i.f. U.S. port
price series exclude duty as well as costs associated with
domestic middlemen (e.g., wholesalers and retailers). All
of these factors may affect the final selling price.
It should also be noted that the indexes are not
seasonally adjusted. Consequently, price trends for com
modities with seasonal patterns may require longer time
spans for proper analysis.
142
Technical References
Pratt, Richard J., and Ferguson, Gwyn R. “Alternative
Sample Designs in the International Price Program,” Pro
U.S. Department of Labor, Bureau of Labor Statistics. Com
parisons of United States, German, and Japanese Export
Price Indexes, Bulletin 2046, February 1980.
ceedings of the Section on Survey Research Methods.
Washington, American Statistical Association, 1980.
143
Bureau of Labor Statistics
U S. Import Product Information
U.S. Department of Labor
The Informal*-! collected on this form by Iha Bureau of
Labor Statistics wIM bo hold In tha strictas. confidence and
will be used for ataflatlcaf purposes only.
TM* report te a u th o rize d by lam 1 9 U.S.C. 2. Y our vo lu n ta ry
cooperation la needed ro make the reoutta o f thta eon ay
comprehensive, accurate, and fltne/y.
It you nave any questions concerning the comptatton of this form OR
PtlASE 0 0 NOT WRITE IN THE B LU E
SHADE 0 AREAS - BIS OFFICE
If tha product described below I* no longer traded, please C A L L
MR. BECKER
collect
AT < 2 0 2 ) 2 7 2 - 5 0 3 4
21 2 555 1212
u seo n lv
©
10002
©
BIKES UNLIMITED
MR. JERRY EVERS
V . P. SALES
367 48TH STREET
NEW YORK/ 4Y
10011
'
Form Approved
O .M .B. No. 1220-0026
Approval expires: 11/30185
©
P R O D U C T D E S C R IP T IO N
________________________________
If any change* are necessary, cross out the incorrect portions, write the new
information in tha opan apace below and see the met ructions for Colum n 7 In Section II.
BICYCLE CHAIN/ MODEL Hf 1 2 3 / FOR
GENERAL BICYCLE APPLICATIONS/
PITCH 1 / 2 INCH/ WIDTH W4 INCH
WEIGHT*. ^ LBS. / FT.
<20
SIZE OF ORDER: PRICE RELATED/* 1 0 / 0 0 0 FEET
CLASS OF SELLER: PRICE NOT RELATED
II
im p o r t
p r ic e
:
Cols 3 - 6 -E n te r the import prices closest to the first day of the
reference month and the appropriate currency. It the product has not been
Imported since the last report, enter ‘N A ‘ (no activity) in lieu ol tha price
p
5
R E FE R E N C E
M O N TH
C IF U S P O R T PRICE
(21
(3)
M ARa
6 2
*2
S E P *
*2
3 5 0 * | i '° :
m
82
m
*
MAR.
63
F O B F O R E IG N PO R T PRICE
CURRENCY OF
C IF PRICE
(4)
JU N E
; .
Cot. 7 • If appticable. enter your coat difference (plu* or minus)
caused by any change in the quality or specifications and
describe those changes in the ‘Remarks' below
« e f *
3 0 0 * 9 0
D O LLA R
» e
M
.
i u
r
M
f
i p
p
*
c o l l a r
1" " ,,n w ........
i
CURRENCY OF
F O B PRICE
16)
(5)
m
m
r n
CO ST OF O U A U TY
C H A N G E (♦ « - >
(7)
R tA S C E
FR A N C
FR A N C E
FR A N C
FR A N C E
FR A N C
' " Y //
V
i i
•
• , jj
i S
v
x
/
7
U.S.
* y ? £ , o o
O O U -4 J S .
2
7 O V -
F £ . F A a a jc
------------------ j t . ----------------------------
o o
P / a . o o
‘
VIA R K £
CA^H
i - r /ZqvcTD f> i£ <Qu#L/ Ty
A/O
L o NC eXL f y p p i l p
b/scoaNT
.
Continue remarks on reverse
Please indicate any cha n ge i in the price factors listed below. If unchanged, check the N O C H A N G E ’ box
III P R IC E F A C T O R S
PRICE F A C TO R S
1P R EV IO US LY
R E P O R TE D D A TA
NO
CHANGE
tAEN
U N IT PRICED
■
-’Kf
C O U N TR Y FROM
W H IC H P R O D U C T
W AS IM PO R TED
| FRANCE
M O S T F R E Q U E N TL Y
US ED U S P O R T O F
E N TR Y
| NEN YORK/ NY
/
I ''.
L ..
.s....... ........... ../ . . v
, •,,y i *t i 11ifffiit111*1il11iimiliri'iti
*
t i f
1
fCoB* I f i l a r m
***
y
/
C
IS
1D U TY
t.
...................... 3
r _________ H H i
E X C L U O E D FR O M C IF PRICE
CZ] P R O D U C T
1 1N O N E
|____|
_ _ _ 3 ._ _
_ _ _ S _ _
IS D U TY FREE
-d t
29
n ... *1 .....i
* ,
!
A L R E A D Y D E O U C TE D T O
ARR IVE A T T H E PRICE IN
S E C TIO N 117
YE S
NO
%D IS TR IB U TO R
H CASH
% O U A N T IT Y
I \/\ I
•A O T H E R (specify)
l
]
-
;-,t
.. *......... *■■■■!■
■■■*
i.. J.—
J-.1 ....L-..1.
.
I (
—
***
SS
;— —
fV l C J
I l l I
PLEASE RETURN WITHIN TWO' WEEKS
144
. . . i■221
Amount
*•1
........ 1-.......... i
■
i ] D U TY IN C L U D E D IN C IF PRICE
j
___________ ,_______
.
(9)
■A
%^ p
i PRODUCT IS DUTY FREE >
D IS C O U N TS
AP P LIC AB LE
TO PRODUCT
S P EC IF IE D
ABOVE
... __ j
-V -
D U TY (complete
only if cil price
reported above)
smottitiE
I
Qfiuf
. ,1
C U R R E N T D A TA (il changed)
»■—
i
E
~
Explanation
of
Footnotes
1.
Industry
yp U h a v e
A n a l y s t - P l e a s e do not h e s i t a t e to call
any qu es t i o n s , p r o b l e m s , or c o m m e n t s .
2.
C o m p a n y na me , r e p r e s e n t a t i v e ,
if n e c e s s a r y .
address
3.
Company
review
4.
P r o d u c t D e s c r i p t i o n - Any c o r r e c t i o n s to this p r o d u c t d e s c r i p t i o n s h o u l d be ma de to
th e right oT tne c U r r e n t d e s c r i p t i o n .
If the p r o d u c t d e s c r i b e d is no l o n g e r tr a d e d ,
p l e a s e call the I n d u s t r y A n a l y s t g i v e n in the t o p - l e f t c o r n e r of the fo rm for
i n f o r m a t i o n ab ou t c h o o s i n g a r e p l a c e m e n t p r o d u c t .
phone
number
- Please
- Please
this
your
re view
information
Industry
this
and
Analyst
information
co rrect
(collect)
and
if
correct
if n e c e s s a r y .
If the SI Z E OF O R D E R an d C L A S S OF S E L L E R we re p r e v i o u s l y r e p o r t e d , the r e s p o n s e
will be p r e p r i n t e d on the r e p r i c i n g form. If they are not p r e p r i n t e d , or the
r e s p o n s e has c h a n g e d , p l e a s e c o m p l e t e t h e s e items a c c o r d i n g to the f o l l o w i n g
guidelines:
If the Si ze of O r d e r has no b e a r i n g on the r e p o r t e d im p o r t pr ic e, en t e r " P r i c e
not r e l a t e d " .
If the im p o r t p r i c e c h a n g e s in a c c o r d a n c e wi th c h a n g e s in the
q u a n t i t y o r d e r e d , e n t e r " P r i c e r e l a t e d " and i n d i c a t e the q u a n t i t y ra ng e that
a p p l i e s to the im po rt p r i c e p r o v i d e d in S e c t i o n II.
C l a s s of s e l l e r r e f e r s to the r e l a t i o n s h i p (if any) e x i s t i n g b e t w e e n your c o m p a n y
and the s e l l e r of the r e p o r t e d p r o d u c t if that r e l a t i o n s h i p has any e f f e c t on the
p r i c e of the p r o d u c t .
If t h e r e is no r e l a t i o n s h i p or if the r e l a t i o n s h i p does not
a f f e c t the p r ic e, e n t e r " P r i c e not r e l a t e d " .
If t h e r e is a r e l a t i o n s h i p and the
s e l l i n g firm g i v e s you a p r i c e di'scount b e c a u s e of this r e l a t i o n s h i p , en t e r " P r i c e
r e l a t e d " and i n d i c a t e w h e t h e r the f o r e i g n s e l l e r is a s u b s i d i a r y , pa r e n t , l i c e n s e e ,
l i c e n s o r , c o n t r a c t o r , etc.
5.
Im po rt P r i c e s - N o t e that bo th the c i f U.S. P o r t P r i c e and the fob F o r e i g n Port
P r i c e s h o u l d be g i v e n if a v a i l a b l e . Wh e n p o s s i b l e , e x c l u d e a p p l i c a b l e d u t i e s wh en
quoting prices.
If the p r o d u c t was not t r a d e d d u r i n g the q u a r t e r (the r e f e r e n c e
m o n t h pl us the two p r e c e d i n g m o n t h s ) , p l e a s e e n t e r "NA" for "no a c t i v i t y " .
6.
C u r r e n c y - P l e a s e re p o r t the c u r r e n c y in w h i c h
a n d (5) are r e p o r t e d .
I n c l u d e the na m e of the
7.
Cost of Q u a l i t y C h a n g e - This c o l u m n s h o u l d be c o m p l e t e d if th e r e has been a
p h y s i c a l c h a n g e in the p r o d u c t a f f e c t i n g q u a l i t y - r e 1 a t e d a r e a s such as p e r f o r m a n c e ,
d u r a b i l i t y and r a ng e of c a p a b i l i t i e s .
Me re s t y l i n g or m o d e l c h a n g e s are not n e c e s
sarily quality changes.
R a t h e r , c h a n g e s such as the h o r s e p o w e r of an e n g i n e or
the a d d i t i o n of c e r t a i n o p e r a t i o n s in a c a l c u l a t o r r e p r e s e n t q u a l i t y c h a n g e s for
w h i c h we n e e d to k n o w the cost.
A d e t a i l e d d e s c r i p t i o n of the c h a n g e s h o u l d be
e n t e r e d in the " R e m a r k s " s e c t i o n at the b o t t o m of S e c t i o n II.
P l e a s e no te that this
c o l u m n s h o u l d not be us e d to re po rt s i m p l e pr i c e d i f f e r e n c e s from p e r i o d to p e ri od .
8.
Remarks
9.
P r i c e f a c t o r s - C h e c k the NO C H A N G E box for each
T T a p a r t i c u l a r p r i c e f a c t o r has c h a n g e d , p l e a s e
that f a c t o r in the C U R R E N T D A T A area.
NOTES:
- Enter
a.
b.
any
remarks
or
comments
the cif
country
and fob p r i c e s in
wi t h the c u r r e n c y
columns
name.
(3)
here.
p r i c e f a c t o r that has not c h a n g e d .
p r o v i d e the new i n f o r m a t i o n for
C o m p l e t e the du ty s e c t i o n on ly if a cif p r i c e is r e p o r t e d in S e c t i o n II.
If any part oT Fne d i s c o u n t p a c k a g e has c h a n g e d , p l e a s e e n t e r in the C U R R E N T
DATA area
all of the d i s c o u n t s w h i c h are now i n c l u d e d in the p a c k a g e .
In
a d d i t i o n , i n d i c a t e if ea ch d i s c o u n t in the new p a c k a g e is a l r e a d y d e d u c t e d to
a r r i v e at the c u r r e n t p r i c e p r o v i d e d in S e c t i o n II.
145
U.S. Departm ent of Labor
Bureau of Labor Statistics
U.S. Export Product Information
The information collected on this form by the Bureau of
T h is r e p o r t is a u t h o r i z e d b y l a * 2 9 U S C 2
c o o p e r a t i o n i s n e e d e d l o m a k e t h e r e s u lt s o f t h i s s u r v e y
c o m p re h e n s iv e , a c c u r a te , e n d tim e ly
If you have eny questions concerning the completion of thl» form Ofl
If the product described below is no longer traded, please CALL
MR. YAGER
c o llec t
(/ T1 V)
V 1J
F o rm A p p ro v e d
O .M .B . N o . 1 2 2 0 -0 0 2 5
A p p ro v a l E x p ire s 1 1 /3 0 /8 8
Y o u r v o lu n t a r y
Labor Statistics w ill be held in the etrtctest confidence and
w ill be used for statistical purpoees only
PLEASE DO NOT WRITE IN THE BLUE
SHADED AREAS - BLS OFFICE
AT ( 2 0 2 ) 2 7 2 - 5 0 3 4
m
E! II
l | Q B
UMrOHLY 31 2 555 1 21 2
J00°2
JONES COMPANY
MRS. JANE FREDERICK
/ o A
V. P. INTERNATIONAL SALES A 1 )
987 EUCLID STREET
\~ J
DECATUR/ IL
62521
f 1 ]
VO )
V I/
a m ass oi ta w s
• m o *TtJ6iooa
m w zzm xsgam
Information m the open space below and see the Instructions for Column S In Section II
PLUG FUSE/ MODEL # T 6 S 1 » 7 UL APPROVED/
15 AMP/ VISUAL DISPLAY/ HEAT RESISTANT
GLASS/ MAGNIFYING GLASS/ 500 PER
-e^
CONTAINER (BULK PACKAGE)/ SHP WT»30 LB
^765
v
f
A
j
SIZE OF ORDER: PRICE NOT RELATED
V I/
CLASS OF BUYER: PRICE RELATED) DISTRIBUTOR
EXPORT PRICE
REFERENCE
(3)
(2)
WAR# 62
JUNE 62
SEP. 62
Col 4 ■Enter a code for the appropriate price basis
1 - fas l fob port I fob border
2 - fob factory / fob distribution center / fob mill / fob mine
3 - fly awey factory (faf) l fob factory, freight allowed to port
or border
4 - other (if code 4 is entered, specify ------------------------------------------ )
w
.
613.00
613.00
ttu m '
613 • 25
* 2
s
Col 5 - if applicable, enter your cost difference (plus or minus)
caused by any change in the quality or specifications and
describe those changes in the Remarks' below
o & (Zcc-Ttrh ^ n o p e r.
PRICE
BASIS
Ki
MONTH
Col 3 - Enter the export price closest to the first day of the
reference month (fas price preferred) If the product has not been
exported since the last report, enter 'NA' (no activity) in Col. 3.
c o s t
o f o u a l it y
CHANGE (+ or - )
(5)
§j H
2
z
7
*/3.SO
MAR. 63
___ (G J
•
/ i/ u / m q o S.
Continue remarks on reverse
'■ease indicate any changes in the price factors listed II unchanged, check the NO CHANGE box
PRICE FACTORS
PREVIOUSLY REPORTED DATA
NO
OFFICE USE
ONLY
CURRENT DATA (If changed)
change
COUNTRY OR AREA
OF DESTINATION
UNIT PRICEO
DISCOUNTS
APPLICABLE
TO PRODUCT
SPECIFIED
ABOVE
H
''
m m *
■ i
*4
/
m
□
NONE
ALREADY DEDUCTED TO
ARRIVE AT THE PRICE IN
SECTION II?
no
----- - •/. DISTRIBUTOR
----- ---------- % CASH
---------------- •/. QUANTITY
---------------- */. OTHER (specify)
... 4
..................
i_____
\
lift
.i. .
i,
m
t
i..t ................... ill.
ii
ill
shown m Co> 2 If the d*td are the same for the current pricing period write SAME in Col 3 If the data in Col 2 are no longer applicable tor Coi 2 is blank),
please supply the new data in Col 3
(1) SHIPPING INFORMATION
FREIGHT CHARGES OF A TYPICAL SHIPMENT
FROM FACTORY OR DISTRIBUTION CENTER TO
U S PORT OR BORDER (IF KNOWN!
(3) CURRENT DATA
(2) PREVIOUSLY REPORTED OATA
e ta
m iP K iiif
m
NUMBER OF UNITS IN A TYPICAL SHIPMENT
f ill
WEIGHT OF A TYPICAL SHIPMENT
NAME OF THE ORIGINAL CARRIER (IF KNOWN)
MODE OF TRANSPORTATION TO PORT
(Sm Trentoonation corns w o * i
METHOD TYPICALLY USED TO SHIP PRODUCT
(Sw ShipptnQ codes 0. 1 0 . 1
LOCATION (CITY. STATE. ZIP) OF FACTORY
OR DISTRIBUTION CENTER
U S PORT (CITY STATE ZIP) MOST FREQUENTLY
USED FOR EXPORT OF THIS PRODUCT
TRANSPORTATION CODES
1 • motor 5 freight forwerder
2 rail
6 UPS
3 an
7 US Parcel Post
4 barge
t pipeline
9 other (specify)
[f|g
26
1.>
| i t
fc
I S r a„ L
291
W
W t
!
1...... t ......
U
-\.
wmrnmm
£
^
ta u e k
l o a d
• if
6 « ||r U * #
C H IC A G O #
SHIPPING CODES
t - car load lot
4 less than truck
load lot
2 less than car
load lot
S air freight
3 truck load
6 container
other (specify)
l
i
IL
62521
6 06 07
n o
is
....................
*3 7 )6 ?
---------------------- _______________»'Pi
mi
4 *»4 f r e g h i Code
i
i
i
i
!...... .
4?:
P f'Cii
B. 5 <
l
1963
***
PLEASE RETURN WITHIN TWC WEEKS
146
***
zu
Explanation
of
Footnotes
1.
In du st ry A n a l y s t - P l e a s e do not h e s i t a t e to call
you ha ve any quest i o n s , p r o b l e m s , or c o m m e n t s .
2.
C o m p a n y name, r e p r e s e n t a t i v e ,
TT necessary.
address
3.
Company
review
4.
P r o d u c t D e s c r i p t i o n - Any c o r r e c t i o n s to this p r o d u c t d e s c r i p t i o n s h o u l d be made to
the right oT tne cu r r e n t d e s c r i p t i o n .
If the pr o d u c t d e s c r i b e d is no lo n g e r tr aded,
p l e a s e call the I n d u s t r y A n a l y s t g i v e n in the t o p - l e f t c o r n e r of the form for
i n f o r m a t i o n ab ou t c h o o s i n g a r e p l a c e m e n t p r o d u c t .
phone
number
- Please
- Pl ease
this
your
review
information
Industry
this
and
An alyst
information
correct
(collect)
and
if
correct
if n e c e s s a r y .
If the SIZE OF O R D E R and C L A S S OF B U Y E R we re p r e v i o u s l y r e p o r t e d , the r e s p o n s e
will be p r e p r i n t e d on the r e p r i c i n g form. If they are not p r e p r i n t e d , or the
r e s p o n s e has c h a n g e d , p l e a s e c o m p l e t e t h e s e items a c c o r d i n g to the f o l l o w i n g
guidelines:
If the Size of O r d e r has no b e a r i n g on the r e p o r t e d ex po rt pr i c e , enter " P r i c e
not r e l a t e d " .
If the e x p o r t p r ic e c h a n g e s in a c c o r d a n c e with c h a n g e s in the
q u a n t i t y o r d e r e d , e n t e r " P r i c e r e l a t e d " and i n d i c a t e the q u a n t i t y range that
a p p l i e s to the e x p o r t p r i c e p r o v i d e d in S e c t i o n II.
C l a s s of B u y e r r e f e r s to the r e l a t i o n s h i p (if any) e x i s t i n g b e t w e e n your c o m p a n y
and the b u y e r of the r e p o r t e d p r o d u c t if that r e l a t i o n s h i p has any ef fe ct on the
p r i c e of the p r o d u c t .
If t h e r e is no r e l a t i o n s h i p o_r if the r e l a t i o n s h i p does not
af f e c t the pr i c e , e n t e r " P r i c e not r e l a t e d " .
If th e r e is a r e l a t i o n s h i p and the
b u y i n g firm r e c e i v e s a p r i c e d i s c o u n t b e c a u s e of this r e l a t i o n s h i p , en te r " P r i c e
r e l a t e d " and i n d i c a t e w h e t h e r the f o r e i g n bu ye r is a s u b s i d i a r y , p a r e n t , l i c e n s e e ,
l i c e n s o r , c o n t r a c t o r , etc.
5.
E x p o r t p r i c e and p r i c e b a s i s - E n t e r the c u r r e n t q u a r t e r p r i c e and pr i c e b a s i s .
The
p r i c e e n t e r e d s h o u l d r e f l e c t a t r a n s a c t i o n p r i c e c l o s e s t to the first day of the
reference month.
If the p r o d u c t was not t r a d e d d u r i n g the q u a r t e r (the r e f e r e n c e
m o n t h pl us the two p r e c e d i n g m o n t h s ) , p l e a s e e n te r "NA" for "no a c t i v i t y " .
6.
Cost of Q u a l i t y C h a n g e - This c o l u m n s h o u l d be c o m p l e t e d if t h e r e has been a
p h y s i c a l c h a n g e Tn the p r o d u c t a f f e c t i n g q u a l i t y - r e l a t e d ar e a s such as p e r f o r m a n c e ,
d u r a b i l i t y and ra ng e of c a p a b i l i t i e s .
M e r e s t y l i n g or m o d e l c h a n g e s are not n e c e s
sarily quality changes.
R a t h e r , c h a n g e s su ch as the h o r s e p o w e r of an e n g i n e or
the a d d i t i o n of c e r t a i n o p e r a t i o n s in a c a l c u l a t o r r e p r e s e n t q u a l i t y c h a n g e s for
w h i c h we n e e d to kn ow the co st .
A d e t a i l e d d e s c r i p t i o n of the c h a n g e s h o u l d be
e n t e r e d in the " R e m a r k s " s e c t i o n at the b o t t o m of S e c t i o n II.
P l e a s e no te that this
c o l u m n s h o u l d not be u s e d to re po rt s i m p l e pr i c e d i f f e r e n c e s from p e r i o d to period.
7.
Remarks
8.
Price
- Enter
factors
any
remarks
- Check
the
or
NO
comments
CHANGE
box
here.
for
ea ch
price
factor
that
has
not
changed.
TT a p a r t i c u l a r p r i c e f a c t o r has c h a n g e d , p l e a s e p r o v i d e the new i n f o r m a t i o n for
that
factor
NOTES:
9.
in the
CURRENT
DATA
area.
a. P r o v i d e da ta on s a l e s to the same c o u n t r y or area each q u a r t e r , if
possible.
b. If you p r e v i o u s l y i n d i c a t e d that the p r i c e of the p r o d u c t d e s c r i b e d in
S e c t i o n I was the sa me to all b u y e r s , r e g a r d l e s s of c o u n t r y , the e n t r y
" W O R L D - W I D E " wi ll be p r e p r i n t e d in the P R E V I O U S L Y R E P O R T E D D A T A area for
the c o u n t r y or area of d e s t i n a t i o n .
If this p r i c i n g p o l i c y has c h a n g e d ,
p l e a s e e n t e r in the C U R R E N T D A T A area the na me of the c o u n t r y or area
to w h i c h most e x p o r t s of the p r o d u c t are c u r r e n t l y so ld .
S e l e c t any
c o u n t r y if your e x p o r t s of the p r o d u c t are e q u a l l y d i s t r i b u t e d to s e ve ra l
count r i e s .
c. If any part of the d i s c o u n t p a c k a g e has c h a n g e d , p l e a s e e n t e r in the
C U R R E N T D A T A a r e a all of TFTe d i s c o u n t s w h i c h are now i n c l u d e d in the
package.
In add it i o n , i n d i c a t e if each d i s c o u n t in the new p a c k a g e is
a l r e a d y d e d u c t e d to a r r i v e at the c u r r e n t pr i c e p r o v i d e d in S e c t i o n II.
If the p r i c e p r o v i d e d »in S e c t i o n II do es not i n c l u d e s h i p p i n g c o s t s to th e U.S.
po i n t of e x p o r t a t i o n , p l e a s e c o m p l e t e S e c t i o n IV.
P l e a s e note that if the
P R E V I O U S L Y R E P O R T E D D A T A ( p r i n t e d in c o l u m n 2) have not c h a n g e d , the word ''Same"
s h o u l d be e n t e r e d in the a p p r o p r i a t e C U R R E N T DA T A b o x e s ( c o l u m n 3).
If the curr en t
d a t a d i f f e r , p l e a s e p r o v i d e the new i n f o r m a t i o n in c o l u m n (3).
147
Chapter 18. Consumer
Expenditures and Income
Consumer expenditure surveys are specialized studies
in which the primary emphasis is on collecting data
relating to family expenditures for goods and services
used in day-to-day living. Expenditure surveys of the
Bureau of Labor Statistics (b ls ) also collect information
on the amount and sources of family income, changes
in savings and debts, and major demographic and
economic characteristics of family members.
Background
The Bureau’s studies of family living conditions rank
among its oldest data-collecting functions. The first na
tionwide expenditure survey was conducted in 1888-91
to study workers’ spending patterns as elements of pro
duction costs. With special reference to competition in
foreign trade, it emphasized the worker’s role as a pro
ducer rather than as a consumer. In response to rapid
price changes prior to the turn of the century, a second
survey was conducted in 1901. These data provided the
weights for an index of prices of food purchased by
workers, which was used as a deflator for workers’
incomes and expenditures for all kinds of goods until
World War I. A third survey, spanning 1917-19, provided
weights for computing a cost-of-living index, now known
as the Consumer Price Index (CPI). The next major
survey, covering only urban wage earners and clerical
workers, was conducted in 1934-36, primarily to revise
these weights.
During the economic depression of the 1930’s, the use
of consumer surveys extended from the study of the wel
fare of selected groups to more general economic analysis.
Concurrent with its 1934-36 investigation, the Bureau
cooperated with four other Federal agencies in a fifth
survey, the 1935-36 study of consumer purchases, which
presented consumption estimates for both urban and rural
segments of the population. The sixth survey, in 1950,
was an abbreviated version of the 1935-36 study, cover
ing only urban consumers. The seventh survey, the 196061 Survey of Consumer Expenditures, which once again
included both urban and rural families, provided the basis
for revising the CPI weights, and also supplied material
for broader economic, social, and market analysis.
The next major survey to collect information on ex
penditures of householders in the United States was
conducted in 1972-73. That survey, while providing con
tinuity with the content of the Bureau’s previous survey,
departed from the past in its collection techniques. Unlike
earlier surveys, the Bureau of the Census, under contract
with b l s , conducted all sample selection and field work.
Another significant change was the use of two independ
ent surveys, a Diary Survey and an Interview Panel
Survey, to collect the information. A third major change
was the switch from an annual recall to a quarterly recall
(Interview Survey) and a daily recall (Diary Survey) of
expenditures. These data were again used to revise the
c pi weights.
It had been apparent for a long time that there was
a need for more timely data than could be supplied by
surveys conducted every 10-12 years. The rapidly chang
ing economic conditions of the 1970’s intensified this
requirement. The new continuing survey that was initiated
in 1979 extended the bls tradition of providing data
describing the consumption behavior of American
families.
Description of the Ongoing Survey
Unlike previous surveys, the latest survey, initiated in
late 1979, is ongoing. Data are, therefore, available at
least annually and possibly more frequently as the survey
continues. The collection of data is carried out by the
Bureau of the Census under contract with b l s . The
objectives of the survey remain the same: To provide the
basis for revising the weights and associated pricing
samples for the c p i and to meet the need for timely and
detailed information on consumption patterns of dif
ferent types of families.
Like the 1972-73 survey, the ongoing survey consists
of two separate surveys, each with a different data col
lection technique and sample. In the Interview Survey,
each consumer unit (CU) in the sample is interviewed
every 3 months over five calendar quarters. The sample
for each quarter is divided into three panels, with c u ’s
being interviewed every 3 months in the same panel of
every quarter. The Diary (or recordkeeping) Survey is
completed at home by the respondent family for two con
secutive 1-week periods.
The sample housing unit is notified in advance by a
letter informing the occupants about the purpose of the
148
survey and the upcoming visit by the interviewer. Both
surveys are conducted by personal visits with telephone
usage limited to appointment scheduling. The interviewer
uses a structured questionnaire to collect both the demo
graphic and expenditure data in the Interview Survey. The
demographic data in the Diary Survey are collected by
the interviewer whereas the expenditure data are entered
on the diary form by the respondent. Both surveys accept
proxy responses from any eligible household member who
is at least 16 years old if an adult is not available after
a few attempts in contacting that person.
The unit for which expenditure reports are collected
is the set of eligible individuals constituting a consumer
unit, which is defined as (1) all members of a particular
housing unit who are related by blood, marriage, adop
tion, or some other legal arrangement, such as foster
children; and (2) a person living alone or sharing a house
hold with others, or living as a roomer in a private home,
lodging house, or in permanent living quarters in a hotel
or motel, but who is financially independent. In the on
going survey, students living in university-sponsored
housing are also included in the sample as separate c u ’s.
The Interview Survey collects detailed data on an
estimated 60 to 70 percent of total family expenditures.
In addition, global estimates, i.e., estimated average
expenditures for a 3-month period, are obtained for food
and other selected items. These global estimates account
for an additional 20 to 25 percent of total expenditures.
On the average, it takes approximately 90 to 120 minutes
to complete the interview.
In the Diary Survey, detailed data are collected on all
expenditures made by consumer units during their par
ticipation in the survey. It is estimated that it takes
approximately 20 minutes for the interviewer to collect
the demographic data and to instruct the respondent on
how to keep the diary. It is also estimated that it will take
the respondent about 90 to 105 minutes each week to
complete the diary.
There is a reinterview program established for the Con
sumer Expenditure Survey to provide quality control. It
provides a means of evaluating individual interviewer per
formance to determine how well the procedures are being
carried out in the field. The reinterview must be
conducted by a member of the supervisory staff. A
subsample of approximately 6 percent of households in
the Interview Survey and 17 percent in the Diary Survey
are reinterviewed on an ongoing basis.
All data collected in both surveys are subject to Census
and BLS confidentiality requirements which prevent the
disclosure of the respondents’ identities. All employees
have taken an oath to this effect.
Interview Survey
The Interview Survey is designed to collect data on the
types of expenditures which respondents can be expected
to recall for a period of 3 months or longer. In general,
expenses reported in the Interview Survey are either
relatively large, such as property, automobiles, or major
appliances, or are expenses which occur on a fairly regular
basis, such as rent, utility bills, or insurance premiums.
Each occupied sample unit is interviewed once per quarter
for five consecutive quarters. After the fifth interview,
the sample unit is dropped from the survey and replaced
by a new consumer unit. For the survey as a whole, 20
percent of the sample is dropped and a new group added
each quarter. New families are introduced into the sam
ple on a regular basis as other families complete their
participation. This rotating procedure is designed to
improve operational efficiency. Another feature of the
current survey is that data collected in each quarter are
considered independently, so that estimates are not de
pendent upon a family participating in the survey for a
full five quarters.
For the initial interview, information is collected on
demographic and family characteristics and on the in
ventory of major durable goods of each consumer unit.
Expenditure information is also collected in this inter
view, using a 1-month recall, but is used, along with the
inventory information, solely for bounding purposes;
i.e, to classify the unit for analysis and to prevent
duplicate reporting of expenditures in subsequent
interviews.
The second through fifth interviews use uniform ques
tionnaires to collect expenditure information in each
quarter. Data collected in these questionnaires, which are
arranged by major expenditure component (e.g., hous
ing, transportation, medical, education), form the basis
of the expenditure estimates derived from the Interview
Survey. In addition, information is obtained on the names
of establishments (or outlets) from which selected com
modities or services are purchased. Wage, salary, and
other information on the employment of each cu
member is also collected or updated in each of these inter
views. The expenditure data are collected via two major
types of questions asked. The first type of question asks
for the purchase month directly for each reported expen
diture. The second type of question asks for a quarterly
amount of expenditures. The usage of these two types
of questions varies depending on the types of expenditures
collected. Approximately 64 percent of the data were col
lected using the direct monthly method, whereas about
36 percent were collected using the quarterly recall
approach.
In the fifth and final interview, an annual supplement
is used to obtain a financial profile of the consumer unit.
This profile consists of information on the income of the
cu as a whole, including unemployment compensation;
income from royalties, dividends, and estates; alimony
and child support, etc. A 12-month recall period is used
to collect income and asset type data.
149
Diary Survey
The primary objective of the Diary Survey is to obtain
expenditure data on small, frequently purchased items
which are normally difficult to recall. These items include
detailed expenditures for food and beverages, both at
home and in eating places; housekeeping supplies and
services; nonprescription drugs; and personal care prod
ucts and services. The Diary Survey is not limited to these
types of expenditures, but rather, includes all expenses
which the consumer unit incurs during the survey week.
Expenses incurred by family members while away from
home overnight and for credit and installment plan
payments are excluded.
Two separate questionnaires are used to collect Diary
data: a Household Characteristics Questionnaire and a
Record of Daily Expenses. The Household Characteristics
Questionnaire is used to record information pertaining
to age, sex, race, marital status, and family composition,
as well as information on the work experience and earn
ings of each cu member. This socioeconomic informa
tion is used by bls to classify the consumer unit for
publication of statistical tables and for economic analysis.
Data on household characteristics also provide the link
in the integration of Diary expenditure data with Inter
view expenditure data for publishing a full profile of con
sumer expenditures by demographic characteristics.
The daily expense record is designed as a self-reporting,
product-oriented diary on which respondents record a
detailed description of all expenses for two consecutive
1-week periods. Data collected each week are considered
independently. The diary is divided by day of purchase
and by broad classifications of goods and services—a
breakdown designed to aid the respondent when record
ing daily purchases. The items reported are subsequently
coded by the Bureau of the Census so that bls can
aggregate individual purchases for representation in the
Consumer Price Index and for presentation in statistical
tables.
Processing
Due to differences in format and design, Diary and
Interview Survey data are processed separately. Diary
questionnaires are reviewed for completeness and con
sistency and are then transmitted to the Census Processing
Center in Washington, DC, where computer processing
is performed. In addition, missing or invalid data on
demographic or work experience are imputed. No
imputation is done for missing data on income. The
families are assigned weights so that estimates can be
derived that represent the total civilian noninstitutional
population. Finally, monthly Diary data tapes are
transmitted to the Bureau of Labor Statistics.
As the monthly Diary data tapes are received, bls
combines the tapes into separate data bases that form
calendar quarters. The data on these quarterly tapes are
screened selectively for invalid coding and inconsistent
relationships as well as for extreme values that may affect
the reasonableness of estimates after the data are
aggregated. All errors of coding or extreme value are cor
rected before further processing.
Selected portions of the Diary data are also adjusted
by automated imputation and allocation routines when
respondents report insufficient detail to meet publication
requirements. These procedures are performed on the
data annually. The imputation routines assign qualify
ing information to data items when there is clear evidence
of invalid nonresponse. For example, the qualifiers clas
sify food expenditures by type of processing (i.e., fresh
or frozen) and apparel expenditures by age and sex group
ings of the members in the consumer units. Allocation
routines are a means of transforming reports of non
specific items into specific ones. For example, when
respondents report expenditures for “ meat” rather than
beef or pork, allocations are performed using proportions
derived from specific reports in other completed diaries
to distribute the expenditure reported for “ meat” to the
specific items such as beef or pork.
Census processing of Interview Survey questionnaires
proceeds along similar lines. The questionnaires are com
pleted and returned to the regional offices, where codes
are applied to identify demographic characteristics,
expenditures, income and assets, and other items such as
make and model of automobile, and trip destination. In
addition, all outlets are coded uniquely by name. Upon
completion of the clerical processing, the data are keyed
and transmitted to Washington where they pass through
a detailed computer pre-edit. Inconsistencies, errors, and
identification of missing questionnaires are transmitted
back to the regional offices for reconciliation by the field
staff through office review or interviewer followup. Cor
rections are keyed and transmitted to Washington, and
again cycled through the computer pre-edit. This con
tinues until errors identified by the pre-edit no longer
appear. Once the pre-edit process is completed for a given
month, data necessary for bounding are transcribed to
the next quarter’s questionnaire. The current quarter’s
questionnaire is sent to a regional processing office for
microfilming and storage.
The data then go through a series of complex computer
edits and adjustments which include the identification and
correction of data irregularities and inconsistencies
throughout the questionnaire. Other adjustments convert
mortgage and vehicle payments into principal and interest
(given associated data on the interest rate and term of
the loan), eliminate business and other reimbursed
expenses, apply appropriate sales taxes, and derive
weights for individual questionnaires. In addition,
demographic and work-experience items (except income)
are imputed when missing or invalid.
The Bureau of Labor Statistics, upon receipt of the
150
data from the Bureau of Census, conducts an extensive
review to ensure that severe data aberrations are cor
rected. The review takes place in several stages: A review
of counts and means by region; a review of coding of
family relationships for inconsistencies; a review of
selected extreme values for expenditure and income
categories; and a verification of the various data transfor
mations performed by b l s . Cases of questionable data
values or relationships are investigated by looking up
questionnaires on microfilm. Any errors are corrected
prior to release of the data for public use.
Data imputation routines are carried out in the Inter
view Survey to account for missing or inconsistent entries.
The procedures are performed on the data quarterly. The
routines, which affect all fields in the data base except
income and assets, are intended to improve the estimates
derived from the survey. Imputation in the Interview
Survey is done at the cell level with cells defined by
variables such as income class, family size, region, and
so on. The methods used—hot deck, weighting class, and
percent distribution—depend on the types of expenditures.
In addition, allocation routines are applied to the Inter
view data in a fashion similar to that for the Diary data.
Sample Design
Selection of households
The Consumer Expenditure Survey is a national proba
bility sample of households designed to represent the total
civilian noninstitutional population. The selection of
households begins with primary sampling units ( p s u ’s),
which consist of counties (or parts thereof), groups of
counties, or independent cities. The set of sample p s u ’s
used for the survey is composed of 109 areas, of which
91 urban areas have also been selected by bls for the
Consumer Price Index program. These urban PSU’s are
classified according to the following four categories: 31
“ A ” certainty p s u ’s (i.e., they are self-representing) that
are Metropolitan Statistical Areas ( m s a ’s) with nonfarm
population greater than 1.2 million; 22 “ L” p s u ’s , which
are medium-sized MSA’s; 24 “ M” p s u ’s , which are small
m s a ’s ; and 14 “ R” p s u ’s , which are urban nonmetro
politan areas. The population break between L and M
p s u ’s is different by region. The breaks are:
North
east—-500,000; Midwest—360,000; South—450,000; and
West—330,000. Since these p s u ’s do not represent the
entire rural part of the United States, 18 additional p s u ’s
(denoted as “ T ” ) are selected to represent the rural
nonmetropolitan areas.
The sampling frame (i.e., the list from which housing
units are chosen) for this survey is now generated from
the 1980 census 100-percent detail file, which is aug
mented by a sample drawn from new construction per
mits and coverage improvement techniques to eliminate
recognized deficiencies in the census. In addition, the
sample for the Diary Survey is doubled during the last
6 weeks of the year to collect expenditures during the
peak shopping period of the Christmas and New Year
holidays.
The population of interest is the total U.S. civilian
population. Within this framework, the eligible popula
tion is composed of all civilian noninstitutional persons
(for example, those living in houses, condominiums, or
apartments) and all people residing in the following group
quarters:
Boarding houses; housing facilities for students and
workers; staff units in hospitals and homes for the aged,
infirm, or needy; permanent type living quarters in hotels
and motels; and mobile home parks. Patients and inmates
are eliminated before sampling. The remaining ineligi
ble persons in group quarters and housing units on
military reservations are eliminated by later screening.
These include: Camps; communes; convents; halfway
houses; homes for the aged, infirm, or needy; transient
quarters in hotels or motels; and missions.
The Bureau of the Census establishes a sample of
8,180 addresses that are requested to participate annually
in the Diary Survey. This results in an effective annual
sample size of 6,050 households, since many interviews
are not completed due to refusals, vacancies, or the
nonexistence of the household address. The actual
workload of interviews is spaced over the 52 weeks of
the year.
The Interview Survey is a rotating panel survey in which
approximately 9,150 addresses are contacted in each of
the five calendar quarters. Allowing for bounding inter
views, which are not included in the estimates, and for
nonresponse (including vacancies), the number of com
pleted interviews per quarter is targeted at 6,760. Each
month, one-fifth of the units interviewed are new to the
survey. Each panel is interviewed for five consecutive
quarters and then dropped from the survey.
Cooperation levels
The response data for the Consumer Expenditure
Survey are shown below for the Interview and Diary
Surveys. The results are based on 1986 data. For the Inter
view Survey, the total refers to housing units when a
unique housing unit address is interviewed once each
quarter for the year.
Response data for 1986
Housing
Type
units
Survey designated B or C
fo r the nonresponse
survey
Interview 39,916
9,288
Diary
151
6,970
1,636
c e
Survey
Eligible
Total
^ype A
Responded
nonresponse
32,946
7,652
4,708
791
28,238
6,861
Response
rate
(percent)
86
90
Type B or C nonresponses are housing units that are
vacant, nonexistent, or ineligible for interview. Type A
nonresponses are housing units which the interviewers
were unable to contact or the respondents refused to par
ticipate in the survey. These response rates are based on
the eligible housing units (i.e., the designated sample less
type B and type C nonresponses).
Weighting
Each family included in the Consumer Expenditure
Survey represents a given number of families in the U.S.
civilian population, which is the universe. The transla
tion of sample families into the universe of families is
known as weighting. Several factors are involved in deter
mining the weight for each consumer unit for which a
usable report is received. One factor in assigning weights
is the inverse of the probability of selection of the hous
ing unit and the adjustment for subsampling in the field.
For interviews which cannot be conducted in occupied
sample households because of refusals or the fact that
no one is home, a complex noninterview adjustment is
made. Additional factors include a national-ratio estimate
adjustment for age, sex, and race to known civilian
noninstitutional population controls and an adjustment
based upon c u family composition to determine a weight
for the consumer unit. In the case of the Diary, a preChristmas seasonal factor is also included in the proba
bility of selection factor.
Beginning with 1984 data, an additional step in the
weighting procedure was introduced to correct the incon
sistent demographic estimates between the Diary and
Interview Surveys. The new step is a ratio estimation pro
cedure using the method of generalized least squares. It
is used to narrow the difference between the consumer
unit counts of the two surveys for selected published
characteristics. While this new step equalizes the number
of consumer unit counts of the two surveys, the impact
on mean expenditures is small.
Presentation
Information from the ongoing Consumer Expenditure
Survey is available in bulletins, reports, analytical papers,
and on public-use tapes. The publications may be
obtained through the BLS Office of Publications, the
Chicago regional office, or from the Government Print
ing Office. Information on public-use tapes can be
obtained from the bls Division of Consumer Expend
iture Surveys.
Publications from the Consumer Expenditures Survey
generally include tabulations of average expenditures and
income arrayed by family characteristics. Data tabulated
for a given year are shown at a relatively aggregated level
due to the small sample size of the ongoing survey. As
the survey continues and more data become available,
however, estimates for several years may be combined
to provide greater expenditure detail and additional
classifications of families.
Data are currently published for the Interview and
Diary Surveys separately; data from the two surveys
must be combined to obtain a complete spending pic
ture. Integrated data will be published in the near future.
The public-use tapes contain the actual expenditure and
income reports of each family but prevent identification
of the family. By eliminating selected geographic detail,
the Bureau reduces the possibility that participating
families may even be indirectly identified.
Uses and Limitations
The survey data are of value to government and pri
vate agencies interested in studying the welfare of
particular segments of the population, such as the
elderly, low-income families, urban families, and those
receiving food stamps. The survey data are used by
economic policymakers interested in the effects of policy
changes on levels of living among diverse socioeconomic
groups. Econometricians find the data useful in construc
ting economic models. Market researchers find them
valuable in analyzing the demand for groups of goods
and services. The Department of Commerce uses the
survey data as a source of information for revising its
benchmark estimates of some of the personal consump
tion expenditure components of the gross national
product.
As in the past, the revision of the Consumer Price
Index remains a primary reason for undertaking such an
extensive survey. The results of the Consumer Expend
iture Survey have been used to select new market baskets
of goods and services for the index, to determine the
relative importance of components, and to derive new
cost weights for the baskets.
Sample surveys are subject to two types of errors, non
sampling and sampling. Nonsampling errors can be
attributed to many sources, such as definitional dif
ficulties, differences in the interpretation of questions,
inability or unwillingness of the respondent to provide
correct information, mistakes in coding or recording the
data obtained, and other errors of collection, response,
processing, coverage, estimation for missing data, and
interviewer variability.
For the Interview Survey, an analysis of time-in-sample
and recall effects was done on a macro level using the
1982-83 data. Minimal-to-moderate conditioning effects
were found in less than half of the published means.
However, the recall-length effects were widespread and
substantial among the expenditure classes for which the
expenditure month is collected. Research on nonsampling
error will continue.
152
Sampling errors occur because observations are not
taken from the entire population. The standard error,
which is the accepted measure for sampling error, is an
estimate of the difference between the sample data and
the data that would have been obtained from a complete
census. The methodology employed to calculate the
sampling variance is balanced half-sample replication.
Standard error tables applicable to published BLS data
can be obtained from the BLS Division of Consumer
Expenditure Surveys.
153
Chapter 19
The Consumer Price Index
Part I.
The Index in Brief
The Consumer Price Index ( c p i ) is a measure of the
average change in the prices paid by urban consumers for
a fixed market basket of goods and services. It is calcu
lated monthly for two population groups, one consisting
only of wage earners and clerical workers and the other
consisting of all urban families.1 The wage earner index
(CPl-w) is a continuation of the historic index that was
introduced well over a half-century ago for use in wage
negotiations. As new uses were developed for the c p i in
recent years, the need for a broader and more represent
ative index became apparent. The all urban index (c pi - u )
introduced in 1978 is representative of the buying habits
of about 80 percent of the noninstitutional population
of the United States, compared with 32 percent
represented in the older index. The methodology for pro
ducing the index is the same for both populations and
is described in detail in part II of this chapter.
History
The Consumer Price Index was initiated during World
War I, when rapid increases in prices, particularly in ship
building centers, made such an index essential for calcu
lating cost-of-living adjustments in wages. To provide
appropriate weighting patterns for the index, so that it
would reflect the relative importance of goods and serv
ices purchased by consumers, studies of family expend
itures were conducted in 92 industrial centers in 1917—
19. Periodic collection of prices was started, and, in 1919,
the Bureau of Labor Statistics began publication of
separate indexes for 32 cities. Regular publication of a
national index, the U.S. city average, began in 1921, and
indexes were estimated back to 1913.2
1 The all-urban-consumer population consists of all urban
households in Metropolitan Statistical Areas and in urban places of 2,500
inhabitants. Nonfarm families living in rural areas within MSA’s are
included, but the index excludes rural families and the military and
institutional population. The urban wage earner and clerical worker
population consists of consumer units with clerical workers, sales
workers, craft workers, operatives, service workers, or laborers. More
than one-half of the consumer unit’s income has to be earned from the
above occupations, and at least one of the members must be employed
for 37 weeks or more in an eligible occupation.
2 Collection of food prices back to 1890 had been initiated in 1903.
During the course of the 1917-19 expenditure survey, retail prices for
other articles were collected in 19 cities for December of each year back
to 1914 and in 13 other cities back to December 1917 only. Retail prices
of food and wholesale prices of other items were used to estimate price
change from 1914 back to 1913.
Because people’s buying habits had changed substan
tially, a new study was made covering expenditures in the
years 1934-36, which provided the basis for a comprehen
sively revised index introduced in 1940.
During World War II, when many commodities were
scarce and goods were rationed, the index weights were
adjusted temporarily to reflect these shortages. In 1951,
the Bureau again made interim adjustments, based on
surveys of consumer expenditures in seven cities between
1947 and 1949, to reflect the most important effects of
immediate postwar changes in buying patterns.3
The first comprehensive postwar revision of the index
was completed in January 1953, using weights from the
1950 expenditure survey.4 At that time, not only were
the weighting factors, list of items, and sources of price
data updated (appendix 1), but many improvements in
pricing and calculation methods were introduced.
Medium-size and small cities were added to the city
sample to make the index representative of prices paid
by all urban wage-earner and clerical-worker families.
Another revision, completed in 1964, introduced new
expenditure weights based on spending patterns in 196061 of single persons as well as families, and updated
samples of cities, goods and services, and retail stores and
service establishments.5
The 1978 revision reflected spending patterns based
upon the surveys of consumer expenditures conducted in
1972-74. A new and expanded 85-area sample was
selected based on the 1970 Census of Population. The
Point-of-Purchase Survey was introduced, which elimi
nated reliance on outdated secondary sources for screen
ing samples of establishments or outlets where prices are
collected. A new store-specific approach to the item selec
tion process was also introduced. In addition, it also
introduced a second index, the more broadly based c p i
for All Urban Consumers (CPI-U), which took into
account the buying patterns of professional and salaried
workers, part-time workers, the self-employed, the
unemployed, and retired people, in addition to wage
earners and clerical workers.6
3 Interim A d ju s tm e n t o f C on su m ers’ P rice In dex, Bulletin 1039
(Bureau of Labor Statistics, 1951).
4 C onsum er P rices in th e U n ited States, 1953-58, Bulletin 1256
(Bureau of Labor Statistics, 1959).
5 The Consum er P rice Index: H isto ry an d Techniques, Bulletin 1517
(Bureau of Labor Statistics, 1966).
6 The C onsum er P rice Index: C oncepts an d C ontent O ver the Years,
Report 517 (Bureau of Labor Statistics, 1978).
154
Concepts
In January 1983, the Bureau changed the way in which
homeownership costs were measured.7 A rental equiva
lence method replaced the asset-price approach to homeownership costs for the c p i - u . In January 1985, the same
change was made in the c p i - w . The central purpose of
the change was to separate shelter costs from the invest
ment component of homeownership so that the index
would reflect only the cost of shelter services provided
by owner-occupied homes.
The most recent revision o f the CPI, completed in
1987, further improved sampling, data collection, proc
essing, and statistical estimation. This revision stressed
techniques that would make the production and calcula
tion of the CPI more efficient, especially with respect to
design and allocation of the samples. The updated
samples of items, outlets, and areas were based upon data
from the Survey of Consumer Expenditures for the years
1982, 1983, and 1984, the 1980 Census of Population,
and the ongoing Point-of-Purchase Survey, which, begin
ning with 1985, reflected the new item and area design.
The new technique of rolling in the new area, item, and
outlet samples significantly reduced the cost of introduc
ing new samples. In addition, the housing survey was
redesigned to represent optimally both owners and
renters, which improved the estimation method for shelter
costs.
The improvements introduced over the years have
reflected not only the Bureau’s own experience and
research, but also the criticisms and investigations of
outsiders. A major study was conducted during World
War II by the President’s Committee on the Cost of
Living.8 The House Committee on Education and Labor
conducted a detailed examination of the index in 1951.9
A decade later, a study was made by the Price Statis
tics Review Committee, which was appointed by the
National Bureau of Economic Research, at the request
of the Office of Statistical Standards of the Bureau of
the Budget, to review all Government price statistics.10
A continuing flow of articles in professional journals
and books has also contributed to the assessment of the
CPI’s quality and of the ways in which it might be
improved.11
Several key concepts indicate the nature of the Con
sumer Price Index and the way in which it is calculated.
Prices and living costs
The c p i is based on a sample of prices of food,
clothing, shelter and fuels, transportation, medical serv
ices, and other goods and services that people buy for
day-to-day living. Price change is measured by repricing
essentially the same market basket of goods and services
at regular intervals and comparing aggregate costs with
the costs of the same market basket in a selected base
period.
A unifying framework for dealing with practical ques
tions that arise in construction of the c p i is provided by
the concept of the cost-of-living ( c o l ) index.12 As it per
tains to the c p i , the c o l index for the current month is
based on the answer to the following question: “ What
is the cost, at this month’s market prices, of achieving
the standard of living actually attained in the base
period?” This cost is a hypothetical expenditure—the
lowest expenditure level necessary at this month’s prices
to achieve the base-period’s living standard. The ratio of
this hypothetical cost to the actual cost of the base-period
consumption basket in the base period is the COL
index.13
The c o l index is a measure of price change (it com
pares current-period and base-period prices). However,
the concept is difficult to implement because it holds the
standard of living constant, and the living standard must
be estimated in some way.
7 “Changing the Homeownership Component of the Consumer
Price Index to Rental Equivalence,” CPI D eta iled R e p o rt, January
1983, pp. 7-13.
8R ep o rt o f The P re sid en t’s C o m m ittee on the C o st o f L iving
(Washington, Office of Economic Stabilization, 1945).
9 C onsum ers’ P rice Index, Report of a Special Subcommittee of the
Committee on Education and Labor, U.S. Congress, House of
Representatives, 82/1, Subcommittee Report No. 2 (Washington, U.S.
Government Printing Office, 1951).
10 G overn m en t P rice S tatistics, Hearings before the Subcommittee
on Economic Statistics, U.S. Congress, Joint Economic Committee,
871. Part I (Washington, U.S. Government Printing Office, Jan. 24,
1961).
11 For a list of published papers on the CPI, see Technical References
at the end of this chapter.
The c p i uses a fixed market basket to hold the baseperiod living standard constant. The c p i equals the ratio
of the cost of the base-period basket at this month’s prices
to the actual cost of the base-period basket in the base
period. The formula used for calculating the c p i is the
one known in price index literature as the Laspeyres
index. (See part II.) The c p i provides an approximation
to a COL index as a measure of consumption costs. It is
sometimes said that the c p i ’s Laspeyres formula provides
an “ upper bound” on the COL index.
Note that both the CPI and the COL index that were
defined above measure changes in expenditures. Neither
one measures the change in income required to maintain
the base-period living standard. For this reason, neither
the c o l index nor the c p i are affected by changes in
12 On the use of a cost-of-living index as a conceptual framework
for practical decisionmaking in putting together a price index, see Robert
Gillingham, “A Conceptual Framework for the Revised Consumer Price
Index,” P roceedin gs o f th e Business an d E con om ic S tatistics Section,
American Statistical Association, 1974, pp. 46-52.
13 For more information on the cost-of-living index concept, see
Technical References at the end of this chapter.
155
than average price increase (or smaller decrease) become
relatively more important. Conversely, items registering
a smaller than average price increase (or larger decrease)
become relatively less important. Thus, the relative
importance of medical care in the index for all urban con
sumers, which was 6.0 percent in December 1982, was
6.9 percent in December 1986. During the same period,
the relative importance of energy fell from 12.4 percent
to 8.9 percent. The published data on relative importance
are often used to answer such questions as: What was
the direct effect on the overall c p i of a particular price
change (e.g., gasoline prices) for a particular period? (See
appendix 2.)
income taxes, but both will include the effects of changes
in sales taxes and other indirect taxes.
For certain purposes, one might want to define price
indexes to include, rather than exclude, income taxes. One
could develop either a c o l index or a Laspeyres index
along these lines. Such indexes would provide an answer
to a different question from the one for which the pres
ent CPI is relevant, and would be appropriate for dif
ferent uses. For a research measure of a consumption
index inclusive of income taxes and Social Security con
tributions, see Gillingham and Greenlees.14
Sampling
Since it is not practical to obtain prices for all consumer
transactions in the United States, the c p i is estimated
from a set of samples. These samples are designed using
statistical procedures to make the c p i representative of
the prices paid for all goods and services purchased by
consumers in all urban areas of the United States. The
samples are:
Owners’ equivalent rent
• A sample of urban areas selected from all U.S.
urban areas,
• A sample of consumer units within each selected
urban area,
• A sample of outlets from which these consumer
units purchased goods and services,
• A sample of specific, unique items—goods and
services—purchased by these consumer units,
• A sam ple o f housin g units in each urban area for
the shelter com p onent o f the
c p i.
Weights and relative importance
The weight of an item in the c pi is derived from the
expenditure on that item as estimated by the Consumer
Expenditure Survey. This survey provides data on the
average expenditure on selected items, such as white
bread, gasoline, and so on, that were purchased by the
index population during the survey period. In a fixedweight index such as the c p i , the implicit quantity of any
item used in calculating the index remains the same from
month to month.
A related concept is the relative importance of an item.
The relative importance shows the share of total expend
iture that would occur if quantities consumed were unaf
fected by changes in relative prices and actually remained
constant. Although the implicit quantity weights remain
fixed, the relative importance changes over time, reflect
ing the effect of price changes. Items registering a greater
14 Robert F. Gillingham and John Greenlees, “The Impact of Direct
Taxes on the Cost of Living,” Journal o f P o litica l E co n o m y, 95(4),
August 1987.
The concept of owners’ equivalent rent used to meas
ure homeowner shelter costs was introduced in the
CPi-u in January 1983 and in the c p i - w in January 1985.
The owners’ equivalent rent index measures the change
in the cost of renting housing services equivalent to those
services provided by owner-occupied housing.
Prior to the introduction of owners’ equivalent rent,
homeowners’ shelter costs in the c pi were represented by
five elements: (1) house prices, (2) mortgage interest
costs, (3) property taxes, (4) homeowner insurance
charges, and (5) maintenance and repair costs. These
constitute the major costs associated with purchasing and
maintaining the physical asset of a house.
This “ asset price” approach to homeowner costs was
flawed because it failed to distinguish the investment
aspect of owning a home from the consumption aspect.
The basic concept of the c p i is as a measure of the
average change in the prices paid by consumers for con
sumption goods and services. Investment purchases, such
as stocks and bonds, are conceptually out of the scope
of the index and are excluded. A house is not consumed
at the time of purchase. It is a long-lived asset (invest
ment), but it also provides the owner with a flow of
shelter services over time. Thus, it is the cost of this shelter
service provided by the asset that is the conceptually
appropriate measurement for the CPI.
To implement the new concept, the old homeownership component was replaced with two items: (1) owners’
equivalent rent; and (2) household insurance, which con
tains those parts of homeowners’ insurance that do not
insure the structure.
In addition, the previous maintenance and repairs com
ponent was made a new component covering both renters’
expenses and owners’ expenses—exclusive of those
estimated to be part of owners’ equivalent rent. Also, the
weight for household appliances was reduced to remove
those expenses in homeowners’ cost for appliances in
cluded with the house.
156
Scope and Calculation
Prices for the goods and services used to calculate the
index are collected in 91 urban areas throughout the
country and from about 21,000 retail and service
establishments—supermarkets, department stores, filling
stations, hospitals, etc. In addition, data on rents are col
lected from about 40,000 tenants and 20,000 owneroccupied housing units.
Food, fuels, and a few other items are priced monthly
in all 91 locations. Prices of most other goods and serv
ices are collected monthly in the five largest urban areas
and bimonthly in the remaining areas. All price infor
mation is obtained through visits or calls by trained bls
field representatives.
To calculate the index, the price changes for all the
various items within each area are averaged together using
weights which represent the importance of the items in
the spending pattern of the appropriate population group
in that area. The U.S. city averages are obtained by
combining the local area data. Separate indexes are com
piled for: 14 groups cross-classified by region and
population size, 4 regions, 4 size classes, and 27 local
metropolitan statistical areas.
Movements of the indexes from one month to another
are usually expressed as percent changes rather than
changes in index points because index point changes are
affected by the level of the index in relation to its base
period while percent changes are not. The example in the
tabulation illustrates the computation of index points and
percent changes:
As an economic indicator. As the most widely used
measure of inflation, the c pi is an indicator of the effec
tiveness of government economic policy. The President,
Congress, and the Federal Reserve Board use trends in
the cpi to aid in formulating fiscal and monetary policies.
In addition, business executives, labor leaders, and other
private citizens use the index as a guide in making
economic decisions.
As a deflator o f other economic series. The c p i and its
components are used to adjust other economic series for
price changes and to translate these series into inflationfree dollars. Examples of series adjusted by the c pi
include retail sales, hourly and weekly earnings, and com
ponents of the gross national product.
As a means o f adjusting income payments. More than
3 million workers are covered by collective bargaining
agreements which tie wages to the c p i . The index affects
the income of more than 60 million persons as a result
of statutory action: 38 million Social Security benefici
aries, over 3 million military and Federal Civil Service
retirees and survivors, and about 19 million food stamp
recipients. Changes in the c pi also affect the cost of
lunches for the 24 million children who eat lunch at
school. Some private firms and individuals use the index
to keep rents, royalties, alimony, and child support
payments in line with changing prices. Finally, since 1985,
the c pi has been used to adjust the Federal income tax
structure to prevent inflation-induced increases in tax
rates.
Index point change
CPI ...............................................................
Less cpi for previous period ..............................
Equals index point change .........................
326.0
318.8
7.2
Percent change
Index point differen ce..........................................
7.2
Divided by the previous i n d e x ...............................
318.8
E q u a ls ...................................................................... 0.023
Results multiplied by 100 ...................... 0.023 X 100
Equals percent c h a n g e ..........................................
2.3
Percent changes for 3-month and 6-month periods can
be expressed as annual rates and are computed according
to the standard formula for compound growth rates.
These data indicate what the percent change would be
if the average rate for the 3-month or 6-month period
were maintained for a 12-month period.
Uses
Almost all Americans are affected by the Consumer
Price Index because of the many ways that it is used.
Three major uses are:
Analysis and Presentation
c pi data are issued initially in a news release about 3
weeks following the reference month. Summary tables
are sent to persons on the mailing list at that time. The
C P I Detailed Report, available about 3 weeks after the
initial release, provides detailed indexes and a monthly
analysis of U.S. price movements. The Monthly Labor
Review also contains much of the c pi data each month
and provides semiannual analyses of recent price move
ments as well as of long-term trends.
Seasonally adjusted data—primarily of use for current
economic analysis—are presented in addition to the unad
justed data. The purpose of seasonal adjustment is to
remove the estimated effect of changes that normally
occur at the same time and in about the same magnitude
every year (such as price movements resulting from
changing climatic conditions, production cycles, model
changeovers, holidays, sales, etc.). Seasonal factors used
in computing the seasonally adjusted indexes are derived
using the a r im a option of the X - ll variant of the
Census Method II Seasonal Adjustment Program. The
seasonal factors are updated annually, and seasonally
157
adjusted data that have been published earlier are sub
ject to revision for up to 5 years after their original
release. (See appendix A for an explanation of bls
seasonal adjustment methods.)
Limitations of the Index
It should be understood that the c pi may not be
applicable to all questions about price movements for all
population groups. For example, the indexes are designed
to represent the average movement of prices for the U.S.
urban population and, thus, are not precisely appropriate
for use by nonurban residents. Also, the CPI does not
provide data separately for the rate of inflation experi
enced by any particular demographic subgroup of the
population, such as the elderly.
In addition, the indexes cannot be used to determine
relative living costs. An individual geographic area index
measures how much prices have changed in that particular
area over a specific time period. It does not show whether
prices or living costs are higher or lower in that area
relative to another.
Part II.
A further limitation is that the c p i is not a complete
measure of price change. Because the index is estimated
from a sample of consumer purchases, the results may
deviate slightly from those which would be obtained if
all consumer transactions were covered. These estimating
or sampling errors are statistical limitations of the index.
A different kind of error in the c p i can occur when
a respondent provides bls field representatives with inac
curate or incomplete information. The Bureau attempts
to minimize these errors by obtaining prices by personal
observation wherever possible, and by correcting errors
immediately upon discovery. The field representatives,
technicians, and commodity specialists who collect, proc
ess, and analyze the data are trained to watch for devia
tions in reported prices which might be due to errors.
Also, an independent audit staff conducts a systematic
evaluation of all c pi collection and processing activities.
The goal is to develop long-term quality improvement in
all aspects of the index calculation.
A fuller discussion of the varieties and sources of possi
ble error in the index is presented in part III of this
chapter, “ Precision of Estimates.”
Construction of the Index
The construction of the Consumer Price Index is based
on a series of samples and on estimation procedures
described below.
Definition of the Index
The c pi is defined as a fixed-quantity price index, that
is, a measure of the price change in a fixed market basket
of consumption goods and services of constant quantity
and quality bought on average by urban consumers, either
for all urban consumers (CPI-U) or for urban wage
earners and clerical workers ( c p i -w ). It is a ratio of the
costs of purchasing a set of items (i) of constant quality
and constant quantity in two different time periods. We
denote the index by It 0, where t is the comparison period
for which a new index number is to be calculated and 0,
the reference period:
data, E Pib Qib, from the 1982-84 Consumer Expenditure
Survey (described later in this chapter) were updated for
relative price changes (Pip / Pib) to November or Decem
ber 1986, the respective pivot periods, p, when they were
introduced into the cpi. Expenditure data for index areas
priced bimonthly in the odd-numbered months were
updated to November 1986. Expenditure data for the
index areas priced monthly or bimonthly in the evennumbered months were updated to December 1986. Price
relatives from the midpoint (June 1983) of the Consumer
Expenditure Survey collection period to November or
December 1986 were calculated from the appropriate
detailed indexes for the period. The price relatives were
applied to the corresponding expenditure data, thereby
updating the expenditures to the end of 1986. Continuity
with the pre-1987 version of the cpi was maintained in
the published version by modifying the above formula to:
ho -
? p i, Q it
V
Ip, 1967
? PipQib
where:
where Ip>1967 is the 1967-based value of the
cpi
for the
Pjt is the
price for the ith itemin comparison periodtpivot month, November or December 1986.
PjO is the
price for the ith itemin reference period 0
Qjb is the
quantity of the ith item consumed in the ex N ote: The base period for the expenditure weights, 1982—
84, should not be confused with the numeric base period for
penditure base period b.
the index, 1967 = 100. A new base period for the index, cor
responding to the base period for the expenditure weights (1982When the expenditure base (b) and reference period (0)
84), will be established effective with the index for January 1988.
coincide, this becomes the Laspeyres price index formula.
Such changes in numeric bases are achieved by dividing all
For the 1987 revision of the c p i , however, they did not
indexes in a series by the appropriate average index on the old
coincide and the formula was modified. The expenditure
base for the time period of the new base.
158
E w hihip h it
Index Estimation
lez
4 iz t , t - l
For sampling and computational purposes, the set of
all retail consumer expenditures by the target population
for a given index area has been subdivided into 207 classes
of similar items called item strata. The item strata are
mutually exclusive and exhaustive of all consumer
expenditures. They are defined identically for both index
populations.
The CPI is computed by a chaining process in which
the index for the previous month, Ihzt _ 1 0, for each basic
area (h) and item stratum (z) is multiplied by an estimate
of the relative change in price from the previous month
to the current month to provide the current-month index
for that area and item stratum:
£lez w
P
w hi r hi,t-l
This is the ratio of the summation of weighted prices,
where the weights (Whi) reflect the probability of selec
tion of the item being priced and a noninterview adjust
ment. Noninterview adjustment is a statistical procedure
designed to adjust for nonresponse. This form of the
estimator is used in the rent and owners’ equivalent rent
item strata.
When the samples of outlets and items are selected with
probability proportional to expenditure, the estimator of
R hzt,t-1 *s *
E z w u p u .t / p hia
‘‘h z t. t- l
^hzt,0 “
^hzt - 1,0 X
where Rhzt>t _ t is an estimate of the one-period price
change in the hth basic area for the zth item stratum.
Basic area and item stratum indexes are then aggregated
using aggregation weights to form desired aggregate area
and item indexes, Ihzt>0:
EE
h z
^hzt,0
A hz ^hzt,0
AHZ
where Ahz is the aggregation weight for the hth basic area
and item stratum and AHZ is the aggregation weight for
the hth area aggregate and zth item aggregate.
The U.S. city average all items c pi is computed by
aggregating all basic component area and item aggregate
indexes:
E E
H z
where Phia is an estimate of the price of the selected item
in period a, corresponding to the expenditures used in
outlet sampling.
This is the ratio of the summation of weighted price
ratios, where the weights reflect the probability of selec
tion of the item being priced and noninterview adjust
ments. It is used for all commodity and service item
strata.
Thus, construction of the c p i is a twofold estimation
process. First, the aggregation weights, A, must be
estimated. These estimates are derived from the Con
sumer Expenditure Survey (CE) as explained in the next
section. Second, the one-period price changes, Rhzt t_lt
must be estimated for each pricing period. The method
ology for estimating price changes is explained in later
sections.
A hz ^hzt,0
where A is the aggregation weight for the U.S. city
average all items c p i .
Aggregation weights are calculated for a given area and
item combination as the expenditures for the pivot period
(Pip Qir) divided by the corresponding index in the pivot
period, that is,
A = Pip Qir / Ip
where p is November or December 1986.
The computation of the index using one-period price
relatives based on identical item specifications in adja
cent periods allows the requisite flexibility to update the
samples of outlets and specifications to reflect an updated
distribution of purchases within an item stratum within
a local area. The form of the estimator for a one-period
price change, Rhzt>t_j, depends on the procedures used
for selecting the samples of outlets and items. When the
samples are selected with each unit having a probability
proportional to quantity, the estimator of Rhzt t-1 is:
L WMPM,M / phla
R hzt,t - 1
Estimation of expenditure-population weights
(aggregation weights)
Within each of the index areas for each item stratum
for both the U and W populations, an estimate of expen
diture is needed to define and weight the market basket
of goods and services for which the index is computed.
Each expenditure-population weight is the product of
estimates of mean expenditures per consumer unit derived
from the 1982-84 CE Survey and estimates of the number
of consumer units obtained from a special tabulation of
the 1980 census sample detail files.
Calculation of mean expenditures consists of three
steps: (1) estimation of preliminary expenditures and
their relative importance, (2) estimation of relative im
portances, using a composite estimation procedure,
(3) estimation of final mean expenditures using a raking
process on mean expenditures derived from the relative
importance data determined in (2). The design criterion
for this estimation procedure is to minimize the average
mean square error of the relative importance of the
aggregation weights for the index areas.
159
Preliminary mean expenditures and relative importance.
Preliminary mean expenditures and their relative impor
tance are calculated for each item stratum and expend
iture class, for each population, index area, replicate,15
and major area by survey source—the CE Interview or
Diary Surveys. There were eight major geographic areas
(index area aggregates), each consisting of either the selfrepresenting or non-self-representing index areas16 in a
particular Census region. The mean expenditures are
estimated using information from the CE Survey. They
are the simple weighted averages of the expenditures for
the particular item stratum or expenditure class for all
consumer units in the population desired. The consumer
unit weights are those described in the section on the CE
Survey. The relative importance of an item stratum or
expenditure cla s is calculated as the proportion of expen
ditures that consumer units in a population (U or W),
index area (or major area), and replicate spend for that
item relative to their expenditures on all items.
Composite estimation. Composite estimation is a
method used to decrease the mean square error (MSE) of
index area, item stratum, and exenditure class (EC)
estimators by using data for the corresponding major
area. It is implemented separately for data from the Diary
and Interview Surveys. The composite estimator of
relative importance for a particular index area, item
stratum, or expenditure class is a weighted average of the
two preliminary relative importance estimators, Ri, for
the index area and major area.
To calculate the composite estimator, let RIj and
RIm be, respectively, the relative importance for the
particular item stratum (or EC) at the index area and
major area. The initial composite estimator, i c r i , is
calculated as:
ICRI = Bj * RIm + (1-Bj) * RIj
where:
Var RI. - Cov RL
B. = --------------------1
ESD RIj
where:
Var RI;,
is the estimated variance of the relative impor
tance for the ith index area,
Cov RI;, is the estimated covariance of the relative im
portance RI; and RIm, and
ESD RI;, is the estimated expected squared differences of
the relative importance RI; and RIm.
If Bj <0, then Bj is set equal to 0. If Bj >1.0, then Bj
is set equal to 1.0.
15 A single selection of entry level items and outlets for all item strata
assigned to a primary sampling unit is called a replicate.
16 Self-representing and non-self-representing index areas are defined
in the section on sample and publication areas.
The above composite estimation procedure defines a
“ shrinkage” estimator. A further refinement is employed
in defining the final composite estimator, c r i , to limit
the shrinkage. Let SDRIj be the square root of the
Var RIj. The final composite estimator is calculated as:
CRI; = ICRI;
CRIi = RIi - V
mg*
if ICRI; - RI; <
mg* SDRI;
SDRI; if RI; - ICRI; >
mg* SDRI;
CRI; = RI; +Km mg* SDRI; if ICRI; - RI; > Km> mg* SDRI;
where Km> mg is a limiting factor defined for each major
area, m, and major item group, mg. The parameter,
Km, mg’ was determined by testing alternative values on
the 1972—
73 CE data. The value which maximized the
reduction of the MSE and minimized the change for
expenditure estimates at the major group level was
selected.
Raking. After composite estimation of the relative
importance, initial total expenditure estimates, TE, for
each item stratum or expenditure class, population, index
area, and replicate for the Diary and Interview Surveys
were calculated as:
TEaj = CRIAi * ESa
where ESA is the sum of preliminary mean expenditures
across all item strata for index area A.
Similarly, total expenditure estimates were calculated
for each item stratum and EC at the major area level.
To reestablish data consistency between item strata and
EC levels and to reflect the special consumer unit weights,
an iterative ratio estimation procedure (raking) was per
formed. That is, the sum of the expenditures for all item
strata within an EC for an index area was forced to equal
the total expenditures for the EC in the index area. The
sum of the expenditures for a specific item stratum across
all index areas in a major area was forced to equal the
major area estimate of average expenditures for the item
stratum times the special consumer unit weights.
Expenditures from the Diary and Interview Surveys
were then integrated. Each item has as its expenditure
source either the Diary or Interview Survey. All of the
processing activities described above were performed
separately by survey. After raking and composite estima
tion were completed, expenditures from the two surveys
were combined to form the set of total expenditures.
The raked, composite estimated expenditures are con
verted to aggregation weights in a two-step process. First,
item strata expenditures were updated from the midpoint
of the CE, June 1983, to the pivot (November or Decem
ber 1986). Each expenditure weight was multiplied by a
long-term price change from the CPI for the time interval:
160
Ihi8612
E hi8612
Ehi8306 •
t
Ahi8306
where:
Ehi86i2 1S
updated expenditure for the item strata in
the hth area for December 1986,
Ehi8306 1S the ra^ed composite estimated expenditure for
the ith item strata in the hth area for June 1983,
Ihi86 i 2 is the c pi index for the ith item strata for the hth
area for December 1986, and
Ihi8306 is the c pi index for the ith item strata for the hth
area for June 1983.
Updated item strata expenditures were summed to
arrive at the required updated expenditures for aggregate
items and areas. In the second step, the updated expend
iture weights were divided by a corresponding index for
the pivot period, yielding the aggregation weight:
A hi
~
E hi8612
j
Ahi8612
Special expenditure weight procedures. As a result of the
1987 revision, the cost-weight definitions for new and used
vehicles and for medical care item strata were changed
significantly. In the former c p i , all expenditures for vehi
cles were valued at the net transaction price—the negoti
ated price less any trade-in value. In the revised CPI, the
treatment of trade-ins and outright sales of used vehicles
has been changed. Ttade-ins at their market value continue
to be netted out of the price of used vehicles. In addi
tion, the market value of trade-ins on new vehicles is net
ted from used vehicle purchases, rather than new vehicles.
Sales of vehicles from one consumer to another are netted
against the corresponding purchase because the transac
tion is really an intrapopulation exchange of wealth; there
is no net change in the consumption of used vehicles. In
medical care, there is a change in the way health insurance
premiums are represented in the cost weight. Health insur
ance represents only expenditures by consumers for pre
miums; employers’ contributions are not included, just
as other medical care expenditures are out-of-pocket
payments by consumers. Insurance premiums can be
viewed as purchasing two things: (1) the medical care for
which benefits are paid, and (2) the services of the
insurance carrier in administering the policy. This second
element has been labeled retained earnings and refers to
operating costs and any profit of the insurance carrier.
In the former c p i , the entire insurance premium was
classified as health insurance. However, within health
insurance, it was broken into many item strata for pric
ing—one for each type of benefit paid and one for the
retained earnings associated with each type of benefit.
The price movement for a health insurance benefit stra
tum (for example, insurance-paid hospital rooms) was the
same as the price movement for the corresponding medi
cal item in the c p i (hospital rooms). The price movement
for a retained earnings stratum was the combination of
price change for the relevant medical care items and an
estimate of changes in retained earnings as a proportion
of premiums. In the revised c p i , instead of using the
price change for hospital rooms (or any other medical
item) for both the hospital room index and the hospitalroom-paid-by-insurance index, the expenditures for the
two types of hospital payment are combined into a single
index. The cost-weight for each medical care item is the
combination of direct out-of-pocket expense for the item
and indirect out-of-pocket expense for the item paid from
consumer-purchased health insurance. The health
insurance cost weight is the sum of all retained earnings.
While this has no effect on the final index result and is
mathematically equivalent to the former procedure, it is
believed that the new structure provides a clearer picture
of the role of health insurance in the c p i .
Annual and semiannual average index estimation
Annual average c p i values are constructed using 12
successive months of c p i values as:
12
W
=
E W
i= 1
12
Semiannual average indexes are computed using 6 suc
cessive months of c p i values as:
6
1=1
where the value of each monthly index is real or imputed,
depending on availability.17 For bimonthly indexes, the
intermediate indexes are calculated using a geometric
mean of the values in the months adjacent to the one
being estimated.
Sampling: Areas, Items, and Outlets
Area sample
Pricing for the c p i is conducted in 94 primary sam
pling units ( p s u ’s) in 91 geographic areas. (The New
York area has three p s u ’s and the Los Angeles area has
two p s u ’s .) The area design18 and sampling are sum
marized as follows: The entire country was divided into
1088 p s u ’s . A p s u is a county or a group of contiguous
counties. The basis of the p s u definition was the geo
graphic areas defined by the Bureau of the Census for
17 To be published, a semiannual average must have at least two
nonimputed index values with sufficient samples. An annual average
must have at least four nonimputed index values with sufficient samples.
18 Cathryn S. Dippo and Curtis A. Jacobs, “Area Sample Redesign
for the Consumer Price Index,” P roceedin gs o f th e Su rvey Research
M eth o d s Section, American Statistical Association, August 1983.
161
the Current Population Survey in 1980 with population
estimates from the 1980 census. Each Consolidated
Metropolitan Statistical Area ( c m sa ) or Metropolitan
Statistical Area (MSA) as defined by the Office of
Management and Budget in 1983 is a p s u . bls grouped
the remaining non-MSA counties containing any urban
population to form p s u ’s . Rural areas of the non-MSA
counties were excluded. (See appendix 3.)
Ninety-one geographic strata were defined by combin
ing similar p s u ’s according to the following characteris
tics, which were found to be highly correlated with price
change:
a.
b.
c.
d.
e.
f.
g.
Region, population size, MSA vs. non-MSA,
Mean interest and dividend income per housing unit,
Mean wage and salary income per housing unit,
Percent of housing units heated by electricity,
Percent of housing units heated by fuel oil,
Percent black, and
Percent retired.
This area design resulted in 31 strata with 1 pricing area
per stratum (self-representing p s u ’s) and 60 non-selfrepresenting strata. (The three New York p s u ’s and two
Los Angeles p s u ’s are also self-representing.) One sam
ple p s u was selected from each non-self-representing
stratum. A controlled selection program was used to
insure th at the sample areas were distributed
geographically across the United States and to increase
the overlap between the old area sample and the new area
sample.
Although 39 of the 94 p s u ’s selected are new to the
CPI, the new area sample for the CPI is being introduced
over a 2-year period. Twenty new p s u ’s were initiated
during 1986 and have been used in c pi index calculations
since January 1987. Sixteen of the remaining new p s u ’s
were initiated during 1987, and the final three will be
initiated in 1988. Each of the 19 old p s u ’s will continue
to be priced until the new p s u which corresponds to it
is initiated and linked into the index.
The new area design defines 37 publication areas, that
is, areas for which a cpi is published. Twenty-seven of
the self-representing areas were defined as publication
areas. Eight additional publication areas were defined by
crossing the two city-size classes (non-self-representing
msa areas) by the four Census regions. The non-MSA
areas in the Midwest and South were also defined as
publication areas. Each of these region-by-size publica
tion areas has 4, 6, or 10 strata. Only two strata were
defined in the non-MSA areas in the West and Northwest,
which made them ineligible for publication. Indexes are
also published for the U.S. total as well as for region and
city-size class totals.
Indexes for the U.S. total, the 10 region-by-size class
areas, and the 5 largest local areas are published monthly.
Indexes for the 10 next largest areas are published bi
monthly, and indexes for the smaller local areas are
published only as semiannual and annual averages.
The c pi includes the pricing of 134 replicates every 2
months where one replicate is approximately 1,100 price
quotes for commodities and services and 390 housing
units for shelter. The allocation of the replicates is pro
portional to the population represented by the p s u , with
at least one replicate assigned to each PSU.
Item and outlet samples: Com m odities and serv
ices other than shelter
Item structure and sampling. The c pi item structure has
four levels of classification. The 7 major product groups
are composed of 69 expenditure classes (EC’s), which in
turn are divided into 207 item strata. Within each item
stratum, one or more substrata, called entry level items
(e l i ’s), are defined. There are 364 entry level items. (See
appendix 4 for a complete list of EC’s, item strata, and
ELl’s.) The e l i ’s are the ultimate sampling units for items
as selected by the bls national office. They are the level
of item definition at which the data collectors begin item
sampling within each sample outlet.
To enable the c p i to reflect changes in the market
place, item and outlet samples are selected each year for
20 percent of the p s u ’s on a rotating basis. Each year,
four regional item universes are tabulated from the two
most recent years of Consumer Expenditure Survey data.
An independent sample of e l i ’s is selected for each item
stratum for each PSU replicate scheduled for rotation that
year from the corresponding regional item universe. Thus,
for the complete 5-year cycle, 134 samples of e l i ’s are
selected nationally. Each eli sample is selected using a
systematic sampling procedure, with each eli within an
item stratum assigned a probability of selection propor
tional to the relative expenditures for the ELI within the
item stratum for the c p i - u population within the region.
Selection of outlet samples is described in the following
section. Linking of the new samples is presented in the
section “ Item substitution, quality adjustments, and
linking.”
Item and outlet sample design. Two separate sample
designs are employed in the c p i , one for commodities
and services, and one for rent and owners’ equivalent
rent. The methodology employed to determine the com
modities and services item and outlet sample design is
presented here. Those for the rent and owners’ equivalent
rent components are described later. For the development
of the sample design, all commodities and services item
strata were grouped into eight major groups:
Food and beverages
Fuels and utilities
Household services and
furnishings
Apparel and upkeep
Transportation
Medical care
Entertainment
Other commodities and
services
The objective of the sample design methodology was
to determine the number of e l i ’s to be sampled and the
162
number of outlets to be selected per p s u replicate by
major group. Four major activities were included in the
design project. First, a variance function was developed
to project the variance of price change as a function of
the above variables for the commodity and service com
ponents. Second, a cost function was formulated to
model the total annual cost of the commodities and serv
ices components of the c p i . Third, estimated values for
all coefficients of the two functions were developed
including estimates of outlet overlap. Fourth, nonlinear
programming techniques were used to determine approx
imately optimal values for the item and outlet sample sizes
under varying assumptions of annual price change and
cost constraints.
The variance and cost functions for the CPI were
modeled for 10 PSU groups:
1. New York City
2. New York, New Jersey
suburbs
3. Los Angeles City
4. Los Angeles suburbs
5. Chicago
6. Philadelphia
7. San Francisco, Detroit
8. Large self-representing
PSU’s
9. Small self-representing
psu ’s
10. Non-self-representing
psu ’s
A detailed discussion of the sample allocation
methodology is provided in appendix 5.
The solution of the design problem yielded the follow
ing number of item strata selections per PSU replicate by
major item group:
expenditure
categories
Item strata
selections (ELI’s)
cpi
Food and beverages
Fuel and utilities
Household services and furnishings
Apparel and upkeep
Transportation
Medical care
Entertainment
Other commodities and services
73
12
66
47
34
18
27
21
The number of outlets selected for each Point-ofPurchase-Survey category (see below) for each major item
group by p s u group is as follows:
expenditure
categories
pops
psu group
1 2 3 4 5 6 7 8 9
10
Food and beverages........
Fuel and utilities ............
Household services and
furnishings ......................
Apparel and upkeep
6 7 6 6 8 9 9 4 2 6
7 8 4 4 7 8 6 4 2 6
Transportation ................
Medical care ....................
E ntertainm ent..................
Other commodities and
services ..............................
2 4 3 3 3 4 4 3 1 3
3 3 3 3 3 5 3 3 1 4
1111111111
2 2 2 2 2 3 3 2 1 2
1111111111
111112 1111
The PSU groups are as defined earlier. With this alloca
tion, outlets and quotes will be initiated each year under
sample rotation. For ongoing pricing, there will be about
25,000 outlets visited each month, with prices collected
for 95,000 items.
Outlet and price surveys, bls field representatives col
lect prices monthly for food, energy items, rent, owners’
equivalent rent, and a small number of other commod
ity and service item strata in all 94 p s u ’s . Prices are col
lected monthly for all commodity and service item strata
in the five largest index areas (New York, Los Angeles,
Chicago, Philadelphia, and San Francisco). Prices are
collected bimonthly for the item strata not cited above
in the remaining index areas. A given p s u is assigned to
either the even- or odd-numbered months for pricing.
Food and other commodities and services, rent, and
owners’ equivalent rent each have separate pricing
surveys, each with its own sample design. Prices for health
insurance are obtained from sources outside of b l s . The
sample design for each component is described later.
Frame development and sampling methodologies for
these items are also described later.
Point-of-Purchase Survey. Since 1977, the Bureau of
the Census has conducted a Continuing Point-ofPurchase Survey (c po ps ) for b l s , to acquire current data
on outlets from which urban households made purchases
of defined groups of commodities and services. Data
from the survey provide the sampling frame of outlets
for food and most commodities and services to be priced
in the c p i . c p o p s is the source of the outlet sampling
frame for about two-thirds of the c p i items by expen
diture weight. (See appendix 6 for a list of CPOPS
categories.) Items not covered by the c p o p s include rent,
owners’ equivalent rent, natural gas, and electricity.
The Point-of-Purchase Survey conducted in 1974 was
the source of the outlet sampling frame in the 87 p s u ’s
defined for the 1977 c p i . It was based on the 1970
census. From 1977 to 1984, the Continuing Point-ofPurchase Survey was conducted in approximately onefifth of these p s u ’s on a rotating basis, so that the outlet
sample for any PSU was never more than 5 years old.
Since 1985, the survey has been based on the 1980 census
and covers the 94 p s u ’s defined for the 1987 CPI.
Various methods have been tested to determine the
sample of households to be interviewed in the survey. In
1974, a highly clustered sample of households was selected
on the assumption that, if families tended to buy in the
areas where they live, the outlets given as responses to
the survey would also be clustered. In order to increase
the expected chance of clustering the outlets, the
household clusters were formed (where possible) around
central business districts, shopping centers, and other
retail centers. These large clusters were called secondary
sampling units (ssu’s). Within a cluster of tracts, a sam
ple of Census enumeration districts (e d ’s) was selected,
and, within the selected e d ’s , the sampled households
163
were dispersed evenly. Five housing units were selected
in each ED, and, since the desired sample size per cluster
was 40 housing units, about 8 e d ’s were sampled from
each cluster. In areas that issue permits for new construc
tion, construction units were selected from the list of per
mits issued; in other areas, selection was made from area
segments. This sampling approach was used, with minor
modifications, in 1977 and 1978.
From 1979 through 1984, unclustered samples of
households were selected for the survey. Since 1985,
households have been selected on the basis of a non
compact clustering procedure which is modeled after the
sample design for the Consumer Expenditure Survey
(CE). There are five sample frames: Unit, special place,
area, block, and permit. The 1980 Census 100-Percent
Edited Detail File is used as the source for all frames,
except new construction in permit areas. For this frame,
an unclustered sample of units is chosen from the per
mits issued since January 1980. For the unit, special place,
area, and block frames, e d ’s are selected first. Then, a
systematic sample of four or five units from each chosen
e d is selected.
The Continuing Point-of-Purchase Survey is conducted
annually over a period of 4 to 6 weeks, usually begin
ning in April, in approximately one-fifth of the p s u ’s in
the c p i . The eligible population for the survey is the same
as for the CE Survey: All civilian, noninstitutional
persons, including persons residing in boarding houses,
housing facilities for students and workers, mobile home
parks, permanent-type living quarters in hotels and
motels, and staff residing in institutions. The interviews,
conducted in selected housing units, consist of two parts.
First, the interviewer elicits information on the demo
graphic and socioeconomic characteristics of the house
hold. This information is used to analyze the shopping
patterns of various segments of the population. It is also
used to determine how many consumer units reside in the
housing unit and should be interviewed. A consumer unit
(cu) consists of all members of a particular housing unit
or other type of living quarters who are related by blood,
marriage, adoption, or some other legal arrangement,
such as foster children, or who are unrelated but finan
cially dependent upon each other for major living
expenses, such as housing or food.
In the second part of the interview, the respondent is
asked whether or not the cu purchased categories of
goods and services within a specified recall period. Com
modities and services are grouped into sampling
categories called p o p s categories based on entry level
items as defined in the c p i classification structure. Some
p o p s categories consist of only one ELI, while others con
sist of combinations of e l i ’s . e l i ’s are combined into a
single POPS category when the commodities or services
are generally sold in the same outlets. For example, POPS
category 106, Meat and Poultry, consists of eight beef
e l i ’ s , six pork e l i ’s , four e l i ’s for other meats, and
three poultry e l i ’s . These e l i ’s are combined because an
outlet that sells beef also tends to sell other meats.
Recall periods for p o p s categories vary from 1 week
to 5 years. The recall period for a specific p o p s category
is defined to produce a sufficient, but not excessive,
number of outlets for sampling purposes. Since consumer
units tend to purchase food items, tobacco products, and
gasoline frequently throughout the year, a 1- or 2-week
recall period is used. In contrast, CU’s tend to purchase
cars, hard-surface flooring, and funeral services infre
quently; therefore, a 5-year recall period is assigned to
these categories.
In the 1987 survey design, there are 170 p o p s cate
gories. Two different checklists of p o p s categories are
used by interviewers—each checklist is used in one-half
of all sample households in the total nationwide sample.
Each checklist consists of a subsample of 147 p o p s cate
gories. Most categories are included on both checklists.
However, some of the short-recall-period categories are
included on only one checklist. Subsampling on two
checklists is used to control the expected number of
responses received from a household and to minimize
respondent burden. The combination of sample size and
reference period for a given p o p s category is designed
to generate 6 to 12, not necessarily unique, outlets
reported for a given p s u / p o p s category. For each p o p s
category on the designated checklist, the respondent for
a cu is asked whether purchases were made within the
stated recall period and, if so, the names and locations
of all places of purchase and the expenditure amounts.
From the results of the annual household survey, a new
sample of outlets is selected for approximately one-fifth
of the p s u ’s in the c p i . In the year following the survey,
BLS initiates these new outlets, selects items for pricing
from each, and replaces the former set of items in the
CPI from each surveyed city with the new outlets and
items.
Outlet sampling procedures. As indicated earlier, item
and outlet samples are selected each year for 20 percent
of the p s u ’s on a rotating basis. When a sampled ELI is
selected, a specific p o p s category is identified for outlet
selection. In self-representing p s u ’s , sample households
for the p o p s are divided into two or more independent
groups by the first-stage order of selection, which defines
two or more frames of outlets for outlet selection. The
independent groups or replicates are needed for variance
estimation. A single selection of e l i ’s and outlets for all
item strata assigned to a PSU is called a replicate. For a
given PSU, POPS category and replicate, the total expend
itures reported for a given outlet are edited. If a purchase
is reported for an outlet but the amount of expenditures
is not reported, then, to ensure a chance of selection for
the outlet, the mean expenditure for the PSU, replicate,
and category is assigned. If an outlet reports large expen
ditures, then the amount is restricted to 20 percent
164
of the total expenditure for the PSU, replicate, and
category.
Outlet samples are selected independently for each
PSU, replicate, and pops category using a systematic
sampling procedure. Each outlet on the frame has a prob
ability of selection proportional to the amount of expend
itures reported in the pops. All eli’s selected in the item
sampling process for a given psu replicate are assigned
for pricing in each sample outlet for the corresponding
PSU, replicate, and pops category. When multiple selec
tions of the sample outlet occur, a comparable increase
is made in the number of quotes priced for the outlet.
The designated sample size for a given pops category
within each major item group for each replicate within
a PSU group was presented in the section on item and
outlet sample designs. The designated sample size is the
number of outlet selections and not the number of unique
outlets. It also does not reflect nonresponse, which is
discussed in a later section. The number of replicates by
PSU group is presented below. There are a total of 134
replicates included in the cpi.
PSU group
1. New York City ......................................
2. New York, New Jerseysu b u rb s............
3. Los Angeles C ity ....................................
4. Los Angeles suburbs ............................
5. C h icag o ....................................................
6. Philadelphia ............................................
7. San Francisco, D e tro it..........................
8. Large self-representingp s u ’s ................
9. Small self-representing p s u ’s .................
10. Non-self-representing p s u ’s ..................
Number o f
replicates
4
4
4
2
4
2
4
18
32
60
Outlet sampling procedures fo r commodities and services
not included in the p o p s . Some commodity and service
items were excluded from the pops either because existing
sampling frames were adequate or it was apparent that
the pops would not yield an adequate sampling frame.
For each component, the sampling frame was either
acquired from another source or constructed by bls.
Each of these items has its own sample design. The frames
consist of all outlets providing the commodity or service
in each sample area. A measure of size was associated
with each outlet on the sampling frame. Ideally, this
measure of size was the amount of revenue generated by
the outlet for the item for the CPI-U population in the
sample area. Whenever revenue was not available, an al
ternative measure of size, such as employment, number
of customers, or sales volume, was substituted. All sam
ples were selected using systematic sampling techniques
with probability proportional to the measure of size.
The source of the sampling frame, the definition of the
sampling unit, the measure of size employed, the final
pricing unit, and the number of designated outlets and
quotes for each item are presented in appendix 7.
Merge o f item and outlet samples. Since the item and
outlet samples are selected in separate processes for each
PSU replicate, they must be merged before data collec
tion. A concordance was created mapping each ELI to a
corresponding Point-of-Purchase category. Each sample
ELI is assigned for price collection to the outlet sample
selected for the corresponding Point-of-Purchase
category. The number of price quotes collected for an
ELI in each outlet is equivalent to the number of times
the ELI was selected for the psu replicate in the item
sampling process.
The number of price quotes assigned for collection in
a sample outlet is determined through the item/outlet
sample merge. In the outlet sampling process, an outlet
may be selected more than once for a given pops
category, provided the expenditures reported for the
outlet are large. The outlet may also be selected for more
than one pops category. If an outlet is selected multiple
times for a given POPS category, the same multiple of
price quotes will be assigned for collection for each
sample ELI matching the category. If an outlet is selected
for more than one pops category, price quotes will be
assigned for collection for all sample eli’s matching the
categories.
Selection procedures within outlets. For each ELI
assigned for price collection in a sample outlet, a BLS
field representative selects a specific store item using
multistage probability selection techniques. The field
representative first identifies all of the items included in
the eli definition and offered for sale by the outlet. Items
are grouped by common characteristics, such as brand,
style, size, or type of packaging, etc. With the assistance
of the respondent for the outlet, probabilities of selec
tion are assigned to each group.
The probabilities of selection are proportional to the
sales of the items included in each group. The field repre
sentatives may use any of four alternative procedures for
determining the proportion of sales. In order of prefer
ence, they are:
a.
b.
c.
d.
Obtaining the proportions directly from a respondent;
Ranking the groups by importance of sales as indicated
by the respondent and then obtaining the proportions
directly or using preassigned proportions;
Using shelf space to estimate the proportions where
applicable; and
Using equal probability.
After assigning probabilities of selection, the field
representative uses a random number table to select a
group. All items included in the selected group are iden
tified. Further groups are formed based on the common
characteristics of the items. Probabilities are assigned to
each group, and a random number table is used for selec
tion. The process is repeated through successive stages
until a unique item is identified. The field representative
165
describes the selected item on a checklist which contains
the descriptive characteristics necessary to identify the
item and to determine or explain price differences for all
items defined within the eli.
These procedures make possible an objective probabil
ity sampling of items throughout the cpi. They also allow
broad definitions of eli’s so that the same tight specifica
tion need not be priced everywhere. The wide variety of
specific items greatly reduces the within-item component
of variance, reduces the correlation of price movement
between areas, and allows a substantial reduction in the
number of quotes required. A second important benefit
from the broader eli’s is a significantly higher probabil
ity of finding a priceable item within the definition of
the eli within the sample outlet.
This selection process is completed at the initial visit
to the outlet. Subsequent visits are made to obtain the
price for the selected item either monthly or bimonthly.
Data collection is generally done by personal visit, but
some pricing is completed by phone.
Item and outlet samples: Shelter
The CPI housing sample is the source of information
on price change for the two principal shelter indexes—
the residential rent index and the owners’ equivalent rent
index. The shelter indexes account for approximately 25
percent of the total cpi weight. The housing unit sam
ple is a stratified, systematic, multistage, cluster sample
that consists of approximately 40,000 rental units and
20,000 owner units, bls selected housing units con
structed before 1980 with data developed from the 1980
Census of Population and Housing. For housing con
structed since 1980, the Bureau of the Census supplies
an annual sample of new units from building permits data
and from a canvassing of an area sample developed for
areas not requiring permits.
Stratification, bls used two variables, average rent level
and tenure (percent renter occupied), to select the stratify
ing area clusters called segments; these variables corre
late with rent change. Using them for sampling stratifica
tion serves to make the sample sizes within clusters more
consistent, uniform, and homogeneous.
Stratification accomplished two goals. First, stratify
ing by variables associated with rent change insured
sample coverage for important characteristics that cor
relate with rent. Second, stratification by percent renter
occupied produced the clustering and the consistent sam
ple sizes of renter and owner housing units within clusters.
It is this geographic clustering that permits the assign
ment or “ matching” of renter-occupied units in the
sample to the owner units in the sample. Matching is the
mechanism that provides the basis for measuring price
change of owner housing that is used in the owners’
equivalent rent index.
To meet the stratification goals, bls stratified at the
lowest published Census areas within the 94 cpi psu ’s.
Before stratifying areas where the Census Bureau pro
vided data by block group and block, bls defined partial
block groups (pbg ’s). In the few cpi areas where the
Census Bureau provided data only by enumeration
district (ed), bls stratified at the ED level. Individual
blocks within a block group were established as indepen
dent pbg’s when they had a high percentage of renters
and a large enough number of housing units to stand as
an individual cluster. The balance of each original block
group was also designated in total as a pbg . Less than
5 percent of the block groups were affected by this pro
cess. However, the use of the process cut significantly the
cases where large numbers of renters were selected in a
single building. The resulting pbg ’s were generally far
more homogeneous in terms of percent renter occupied
and structure type than the original block groups.
Strata boundaries were defined, and the pbg ’s and
ed ’s were sorted into the strata. The two variables,
average rent level and tenure (percent renter occupied),
defined the strata. Eighteen strata were defined for each
PSU, using three rent ranges and six tenure ranges. An
important enhancement from previous cpi housing
samples was that strata boundaries were defined dif
ferently for each PSU to insure that each stratum con
tained roughly the same number of housing units and
allowed for between-PSU differences in rent levels and
housing characteristics, bls sampled PBG/ED’s within
each stratum, thereby insuring that the survey included
housing clusters of all rent and tenure levels. Stratifying
by tenure also permitted bls to vary the sample rate for
owners and renters in each cluster to obtain consistent
sample sizes by tenure within the clusters.
Sample allocation to p s u ’s and strata, bls allocated the
sample to minimize a value called Z, which is propor
tional to the sum of the variance of the rent and owners’
equivalent rent indexes. This value is expressed as:
s
z= E
i = 1
where:
S
Oi
Ri
L
°i
r;
_
number of of strata,
= number of owner units in the ith strata,
= number of renter units in the ith strata,
= the total unit variance,
= within cluster variance,
= number of owners allocated to ith strata,
= the number of renters allocated to ith strata.
bls determined the strata sample sizes, O; and
subject
to the sample size constraints, by finding the values for
166
occupied to pass screening and remain in the survey). With
the selection sheets, the field representatives designated
on the listing forms the units to be screened and their
desired tenure.
Survey statisticians prepared the selection sheets for
each segment taking into account each segment’s propor
tion that was renter occupied from the 1980 census to
determine how many units to screen, and how many to
require of each tenure to yield the optional sample of
renter and owner units for the CPI housing survey. The
two within-segment sampling rates determined how many
renters and owners should be in the final sample. The
survey statisticians determined them from the final desired
probability of selection for each tenure, the number of
renters and owners in each stratum, and the number of
selected segments and total segments in each stratum.
Before the field agents contacted survey respondents
for the first time, they had already designated the units
in each segment into one of four cases:
Oj and Tj that minimize Z .19 This produced an optimal
sample for the given resource constraints for the two
indexes.
BLS determined an initial allocation simultaneously
across all strata and p s u ’s based on a criterion requir
ing a minimum sample size for each published index. If
a publication p s u was not allocated the minimum sam
ple, it was assigned a designated number of units large
enough to meet publication standards. This minimum
sample was allocated applying the above formula for Z
among the strata within the PSU. A single process
reallocated the remaining sample to the remaining p s u ’s .
The budget for the CPI housing survey dictated a sam
ple of 10,000 clusters (called segments) and 100,000 pric
ings per year for the pre-1980 sample portion of the
survey, b l s added 900 more segments to compensate for
an expected 9-percent sample loss that results from dif
ferences in the Census Bureau and CPI definitions of
housing units. In contrast to the Census definition, the
CPI excludes public housing, institutional housing, and
military housing.
(1)
(2)
(3)
(4)
Sample selection within strata, b l s selected sample clus
ters in each stratum using a systematic probability-proportional-to-size (PPS) sample selection method. Each p b g
within a stratum was assigned a measure of size accord
ing to the total number of housing units, with controls
on the maximum and minimum percent renter occupied.
BLS sorted the p b g ’s geographically and, using the meas
ures of size, allocated the sample of p b g ’s systematically.
Next, b l s partitioned each selected p b g into a number
of clusters, depending on the p b g ’s size, and selected one
at random. When a single Census block contained more
than one cluster and one of these was a selected cluster,
b l s field representatives defined the individual cluster
following strict procedures derived from the sampling
plan. For example, suppose that the sample design deter
mined that a segment began 10.1 percent of the way into
a block and ended at 23.8 percent into the block. If the
block was not too big, the b l s representative would
enumerate the entire block and then define the segment.
In large blocks, the representative prescreened the blocks
and sent the information to the b l s national office, which
determined the segment.
A field screening determined the final selection of hous
ing units in the sample. In the first step of this process,
called listing, b l s field representatives enumerated in
order on listing forms every housing unit or potential
housing unit they saw in each segment. The national office
had prepared selection sheets that indicated the sequence
number of each unit to be screened and its “desired
tenure” (whether the unit needed to be owner or renter
19 For a full derivation of Z, see W. F. Lane and J. P. Sommers,
“Improved Measures of Shelter Costs!’ Proceedings o f the Business and
E conom ic Statistics Section, American Statistical Association, 1984.
Screen but initiate only if the unit is renter occupied.
Screen but initiate only if the unit is owner occupied.
Screen and initiate if either owner or renter occupied.
Do not contact for screening or initiation.
During screening, the field representatives contacted
an eligible respondent for each housing unit and deter
mined that the unit met the tenure criteria for the survey
as well as other criteria such as being a year-round hous
ing unit, built before 1980, that was someone’s primary
residence.
After they determined the desired tenure, the field
representatives contacted an eligible respondent for each
unit to be screened and obtained the actual tenure and
other eligibility data. Housing units were initiated into
the survey only if their actual tenure was the same as the
desired tenure, and they met the other eligibility criteria.
(Initiation is the process of collecting a first-time inter
view.) Units that did not pass screening were not initiated;
however, they may be recontacted in the future to aug
ment or to rotate the sample.
As planned, only about one-fourth of the units inter
viewed for screening met the tenure and other eligibility
criteria required to pass screening and be initiated into
the c p i housing survey. Because renters were allocated
according to total units and each cluster was allocated
an equal number of renters, most of the units contacted
in areas known from the Census Bureau to be mostly
owner occupied failed screening. However, this process
located the sparse renters in these areas for the survey
and added “ extra” renter units to the rent sample in
owner areas. Although they represent few renters in the
renter universe and, consequently, have very low weight
in the rent index, they serve as the main source of rental
units to match with owner units. It is their movement that
primarily drives the owners’ equivalent rent index.
167
Estimation of Price Change
Commodities and services other than shelter
At the end of each pricing period, the estimate of the
one-period (t-1 to t) price change (price relative) is com
puted for each item stratum and index area. Only price
quotes obtained in both the current and previous pricing
periods for the same or comparable items are used in the
estimate. Where appropriate, prices for food items are
converted to a price per ounce before they are used in
the estimation of price change. The same quote weights
are used both for the current- and previous-period price
quotes. The estimate of the one-period price change for
the hth index area for the zth item stratum for a given
market basket is computed as:
^hzt, a
^hzt,t-l — p
Khzt-1, a
|
W h i P h i , / P hia
E w hi P hiM / p hia
16Z
where:
Phit
is the price o f the ith quote in the current pricing
period, t, for item stratum z in index area h;
Phit_i
is the price of the ith quote in the previous pricing
period, t-1, for item stratum z in index area h;
P hia
is the estimated price for the ith quote for item
stratum z in the time period, a, o f the pops in
index area h;
W hi
is the quote weight for the ith quote for item
stratum z in index area h.
The quote weight, Whi, consists of the product of the
following factors: An estimate of the total daily expend
iture (E) for the p o p s category, for the index area
replicate and the CPI-U population; a duplication factor
(f) to reflect any special subsampling of outlets or quotes;
the percent of sales (a ) of the e l i to the total sales of
the p o p s category in the outlet; the proportion (B) the
expenditures for the selected e l i is of the total expend
itures for the item stratum in the region (the probability
of selection for the e l i ); a geographic factor (g) to reflect
the difference in coverage for the index area for the pre1987 area definitions to the 1987 area definitions; and
the number of usable quotes (M) for the e l i / p s u
replicate within the item stratum:
W hj = a E f g / M B
NOTE: The geographic factor is 1.000 for all samples
selected using the 1987 area definitions. See the section
on outlet sampling procedures for an explanation of sam
ple rotation.
Item substitution, quality adjustments, and linking
One of the more difficult conceptual problems faced
in compiling a price index is the accurate measurement
and treatment of quality change due to constantly chang
ing product specifications and consumption patterns. The
concept of the c p i requires measures of the cost of pur
chasing a fixed market basket of goods and services of
constant quality through time. In reality, products fre
quently disappear, products are replaced with new ver
sions, and new products emerge. Ideally, estimates would
be obtained for the dollar value of each quality change
resulting from a change in the model priced or a substitu
tion to a new item. This estimate would reflect how much
consumers value the quality change. The direct measure
ment of the value consumers place on quality change
when product substitution occurs, of course, is rarely
possible. As an approximation, b l s uses several methods
to adjust for quality change and to account for the change
in item specifications. These methods may be categorized
as 1) directly comparable, 2) direct quality adjustment,
3) linking with overlap price, and 4) linking without
overlap price. In all cases, it is necessary to estimate a
new base-period price in order to use the new item specif
ication in future, if not current, periods.
Directly comparable. If the new and old item specifica
tions are considered directly comparable, i.e., the charac
teristics that define the new specification are essentially
the same as the old item’s characteristics, the base-period
price for the new specification is set equal to the baseperiod price for the old specification, and the price
comparison between the items is used in the index. It is
assumed that no quality difference has occurred.
Direct quality adjustment. This is the most explicit
measure for dealing with specification changes. Direct
quality adjustments are frequently made for the food,
rent, and automobile components of the c p i . The con
version of food prices to price per ounce accounts for
some quality adjustment. If the net weight of an item
changes, then the method used in recording food prices
will take into account this type of change in quality.
Quality adjustments are also made to the cost of rental
housing used in the rent and owners’ equivalent rent
indexes. BLS collects the rent change plus a description
of major services and facilities provided by the landlord.
If the services and facilities differ between two collection
periods when rents are compared, the rent for the current
period is adjusted to reflect the differences in services
between the time periods. For instance, if the owner no
longer provides a certain utility, BLS calculates and adds
an estimate of the value of that utility to the current rent
in order to have an adjusted rent value. This adjusted
rent is the current cost of the same set of services pro
vided for the previous rent payment.
b l s used data from the Department of Energy’s Resi
dential Energy Survey to develop formulas to estimate
utility usage for various types and sizes of housing, in
168
various climates, with different types of heating and airconditioning, hot water, cooking stove, and so on. The
prices for the utilities come from the CPI average price
program. A similar, although simpler, formula estimates
water costs. Research is underway to determine how to
quality adjust major changes such as changes in the
number of rooms or bathrooms. Currently, when such
major changes occur, BLS omits these observations from
the calculation in estimating price change.
The most frequently cited example of direct quality
adjustment is the annual model changeover for new
automobiles. Direct quality adjustments are made for
changes in standard features between model years. This
estimate is based on all costs incurred in manufacturing
plus the established manufacturer and retail markup to
the selling price of passenger cars. This producer-cost
estimate applies to all new features that are installed as
standard equipment, that is, features on all cars in the
same or comparable series. Any former optional item that
becomes standard has a market price (i.e., the former
option price) which is the consumer value of that option
for those who bought it. For such items, the value of the
quality change is a weighted average of the former option
price and the producer cost. For all items that replace
or modify some previously existing feature, the estimate
is based on the difference in producer cost between the
old and the new feature, marked up to retail. In other
words, the estimate of total production cost for new items
is computed for both the new and the old feature. The
difference between these values is used as the estimate
of quality change.
Adjustments for quality change in the cpi new car
index include structural and engineering changes that
affect safety, environment, reliability, performance,
durability, economy, carrying capacity, maneuverability,
comfort, and convenience. Although antipollution equip
ment on automobiles does not directly increase the quality
of the automobile for the buyer, these devices do improve
the quality for consumers in general. Consequently,
quality adjustments are made for pollution controls to
automobiles on the assumption that, by legislative defini
tion, the cost of installing antipollution devices was no
more than the value derived from them.
Quality adjustments of new cars exclude changes in
style or appearance, such as chrome trim, unless these
features have been offered as options and purchased by
customers. Also, new technology sometimes results in bet
ter quality at the same or reduced cost. No satisfactory
value can usually be developed for such a change. In such
cases, it is ignored, and prices are compared directly.
In general, if the new item specification is similar to
the previous one but has changed one or more of its com
ponent parts, a quality adjustment may be made to
establish comparability between their prices. A synthetic
previous-period price for the new item (P*iM ) is
calculated as follows:
;
+ QA
P*:i,t-l = P i,t-l
where:
M ,t-i is the previous-period price of the old
specification,
and
QA is the dollar value of the quality change which may
be either positive or negative. After the above imputation
is made, the base-period price for the new item (P*i>a) is
computed as:
P i,a P V l
P*.i,a
P:i,t-l
where:
Pj
is the base-period price for the previous item.
Linking with overlap price. When a noncomparable
substitution occurs and a price is obtained for both the
old (Pj t) and new (P*i t) specifications in the same period
(overlap pricing), the estimation of the new base-period
price is based on the same-period price relationship of
the two specifications. The new base-period price (P*a)
is estimated as follows:
p i% = p *t (p u
/ p i.t)
The linking of quotes with overlap prices is done before
item relatives are compared. For the current month, the
price comparison used in the index is based on the old
variety. At the next pricing, the comparison will be made
on the new item. The quality difference is assumed to be
the difference in the observed prices of the old and new
varieties in the current month.
Linking without overlap price. For quotes which are not
comparable because of a change in specification (substitu
tion) and no quality adjustment or overlap price can be
obtained, the new specification price is not used in the
current-period estimate of the relative. Implicitly, this
means that the price change for that price quote is as
sumed to be the same as the average change of those
quotes in the same item stratum/index area/pricing cycle.
To execute the link, an estimate of the long-term
change for the previous pricing period ( R ^ g ) for the
item stratum and the current 1-month pricing period
relative (Rztt_j) is required. A new base-period price
(P*i a) for the specification is computed as follows:
Pzi,t
p*zi,a =
^ zt-l,a ^zt.t-l
The value of R ^ a is estimated at either the quote level
by using the ratio of the previous-period price to the
base price or at the item strata level by using the ratio
of indexes for the item stratum/index area/cycle, where:
Rzt-l,a
l z,t-l
z,a
169
Izt_j is the index of the previous pricing period for the
item stratum/index area/pricing cycle,
Linking without an overlap price could cause a bias
in the index due to the nature of price setting in retail
apparel stores. New items are sold at introductory prices
in the beginning of their selling season and then continu
ally have their prices reduced through repeated “ sales”
until the stock is depleted. Linking quotes would result
in no price comparison being made between items at the
time of substitution. Price change is imputed from those
quotes priced during the period, which include many yearround items and some new varieties that are deemed com
parable to the old item. In subsequent months, as price
reductions occur, the index will show continuous price
declines. The following example illustrates the problem:
Iz a is the base-period index for the cpops (or other frame
reference),
t_j is the one-period price change relative for the stratum.
The quality difference between the items in this case
is assumed to be the difference between the price of the
new variety and the imputed price for the old variety. (See
discussion of price imputation in the “ Estimation of price
change” section above.)
The incidence of substitution by major CPI category
during 1983 and 1984 is presented in table 1. The rate of
substitution has averaged about 3.9 percent of price occur
rences. The highest rates of substitution occur in the
apparel and upkeep category and the transportation cat
egory. Direct quality adjustments are made most fre
quently to new vehicles within the transportation category.
P rices f o r h yp o th etica l a p p a rel item
Jan.
Apparel. The pricing of apparel items causes a number of
problems for quality adjustment in the CPI. Many apparel
items are seasonal and subject to frequent style changes.
When new styles replace old ones, many substitutions are
deemed to be noncomparable. Adjusting for quality change
becomes complicated because information from manufac
turers to make direct quality adjustments is not available
or difficult to estimate. In addition, overlap prices are not
available since the old style has usually been sold out. The
only method available is to link the quote.
T a b le 1.
S u b s t it u t io n s a s a p e r c e n t o f o c c u r r e n c e s b y m a jo r
Feb.
M ar.
A p r.
Old item ..................
$100
$75
$50
—
New it e m ..................
—
—
—
$150
Percent ch an ge........
—
-25
-33
4
M ay
June
—
—
$125 $100
-17
-20
1 Imputed from other apparel items.
In January through March, the old item was priced and
showed large declines. In April, the old item is not avail
able and a substitution to a new item occurs. The two
are declared noncomparable and there is no overlap price.
If the new item is linked into the index, the price change
is imputed as the average of all other items in that index
area/stratum. Assume this to be 1 percent.
cpi
c a te g o ry , 1 9 8 3 a n d 1 9 84
Link
Category and year
Total
Directly
comparable
Total
With
overlap
Without
overlap
Direct
quality
adjustment
All:
1983 ...............................................................................................
1984 ...............................................................................................
3.85
3.95
1.56
1.70
1.97
1.95
0 23
.23
1 74
1 71
0.32
.30
Food and beverages:
1983 ...............................................................................................
1984 ...............................................................................................
1.81
1.93
.52
.52
1 29
1.41
.04
.08
1.25
1.33
.00
.00
Housing:
1983 ...............................................................................................
1984 ...............................................................................................
4.25
4.73
2.21
2.67
1.89
1.93
.22
.21
1.67
1.62
.15
.22
Apparel and upkeep:
1983 ...............................................................................................
1984 ...............................................................................................
17.34
17.59
7.15
7.80
10.15
9.70
2.69
2.43
7.46
7.27
.03
.09
Transportation:
1983 ...............................................................................................
1984 ...............................................................................................
6.72
5.80
3.13
3.02
1.41
1.93
.06
.07
1.35
.96
2.18
1.74
Medical care:
1983 ...............................................................................................
1984 ...............................................................................................
2.22
2.19
.65
.80
.95
1.02
.03
.03
.981
.99
.64
.38
Entertainment:
1983 ...............................................................................................
1984 ...............................................................................................
4.61
6.08
1.92
2.85
2.51
2.96
.23
.26
228
2.70
.18
.27
Other goods and services:
1983 ...............................................................................................
1984 ...............................................................................................
3.30
3.99
1.44
1.94
1.69
1.64
.06
.08
1.64
1.56
.17
.49
Note: Because of special pricing procedures the following items were excluded: Residential rent and owners' equivalent rent within housing; used
cars within Transportation; health insurance within Medical Care; and magazines, periodicals, and books within Entertainment.
170
In May’s pricing, the new item shows a decline of 17
percent, and, in June, there is another decline of 20 per
cent. Thus, the index would show an almost continual
downward movement. To overcome this inherent bias,
there is a procedural rule used that prohibits an item from
going out of the index on sale. When the item is no longer
available following a sale price, and a noncomparable sub
stitute is selected, an estimated overlap price is used. This
estimated price is the item’s last regular (nonsale) price.
In the above example, when a noncomparable substitute
occurred in April, an artificial price of $100 would be used
as an estimated overlap for computing the index so that
the March to April change would be 100 percent ($50 to
$100), thus bringing the index back to its presale level.
This procedure of estimated overlaps prevents a severe
downward bias from being introduced into the c p i when
items disappear from the index on sale and are replaced
by varieties with substantially different quality features.
This artificial overlap represents an estimate of the “non
sale” price of the old variety if it were still available.
Research is underway to evaluate alternative estimators
that may produce better unbiased estimated overlap prices
for this situation. Alternative imputation procedures for
noncomparable substitutes are also being studied.
Medical care. Another area in which quality adjustment
presents particular difficulties is medical care. Not all fac
tors affecting the quality of medical care services can be
accounted for in the description of the item being priced.
Quite often the respondent does not have knowledge of
many price-determining quality factors. For example,
hospital room modifications, changes in the nurse-topatient ratio, or the availability of new equipment are all
likely to contribute to determining the price level of the
room service priced. Such changes are normally reflected
as price movement because b l s either is not aware of the
changes or has no method available to deal with the
change. Improved technologies and procedures can lead
to quality changes that cannot necessarily be measured
by BLS. For instance, new advances in the development
of porous materials in the manufacturing of prosthetic
implants, such as in hip replacement surgery, allow the
bone to grow around the prosthesis. This is not the case
with the nonporous materials that have been commonly
used in hip replacement prosthetic implants. Many doc
tors view this porous implant as an improvement in the
results of hip replacement surgery. In pricing total hip
replacement surgery, the quality impact of shifting from
a nonporous to a porous implant would not be factored
out of the index, as b l s has no methodology to account
for the value of the quality difference.
There are, however, certain areas in medical care where
the quality difference can be measured and adjustments
made for changes in the quality of priced services. For
example, the CPI might be pricing a limited visit to a
physician’s office for treatment of a sore throat, and the
physician might have later changed the fee schedule so
that a throat culture would be included in the price of
the visit, whereas previously it had been separately billed
for by a laboratory. The addition of the throat culture
would be reflected as a quality change to the previously
described service. If a hospital introduces a separate
admitting charge that previously was included in the room
rate, b l s prorates the admission charge to a per-day basis
using an appropriate hospital-provided length-of-stay
measure. The prorated admission charge is then added
to the room rate priced to reflect the price movement in
the index.
Other price adjustments
Bonus merchandise adjustments. Sometimes products
are offered with free merchandise included with the
purchase of the original item. Such “ bonus” items may
provide additional satisfaction to consumers, and bls
will, therefore, make adjustments to the purchase price
to take into consideration the value of the bonus mer
chandise. The adjustment made depends on the type of
merchandise offered and the perceived value of the bonus
to the consumer.
If the bonus merchandise consists of more of the same
item, the adjustment is reflected in the unit price of the
item. For example, if a manufacturer offers 2 ounces of
toothpaste “ free with the purchase of the regular 6-ounce
tube,” then the item’s price will be reduced by 25 per
cent, reflecting the decline in the cost per ounce. When
the bonus is removed, the price per ounce will return to
its prior level, and a price increase will be recorded. In
this instance, the value to the consumer is assumed to be
one-third greater.
If the bonus merchandise consists of an item that has
some significant value to the consumer, and the item is
of a different genre, an adjustment will be made to
account for the value of the free item offered when the
original item selected for the c p i is priced. For example,
when a box of cereal is sold containing a free package
of candy, the item’s price will be reduced by the unit cost
of the candy to the manufacturer. In this instance, the
value to the consumer is assumed to be equal to the
manufacturer’s cost.
When bonus merchandise adjustments occur, base
prices are not adjusted since there is no difference in the
quantity or quality of the original item being priced.
Utility refunds. Sometimes public utility commissions
require that utilities such as telephone, natural gas, or elec
tricity companies make rebates to their customers. These
rebates may arise from a number of different causes. For
example, a utility may be permitted to use a new rate
schedule temporarily until a final determination is made.
If the final rates set by the commission are lower than
the temporary ones, the difference must be refunded for
171
consumption during the period. The utility bills priced
for the c p i will reflect the full amount of these refunds
in the month(s) they are credited to the customers.
Manufacturers’ rebate. When product manufacturers
offer cash rebates to consumers for purchases of items
priced in the c p i , b l s reflects these rebates as price reduc
tions in the index. The amount of the rebate adjustment
usually depends on the percentage of purchasers who take
advantage of the rebate offer. For example, when auto
manufacturers offer a $500 rebate on the purchase of a
new car or reduced-rate financing, the price of each car
eligible for the rebate is reduced by the proportion of cus
tomers who opt for the rebate. If 70 percent of customers
choose rebates for a particular model, then the price of
each quote for that model in the c p i will be reduced by
$350, and the index will reflect the price decline. The re
duced interest rates chosen by the remaining customers
will be reflected in the auto financing component of the
c p i . For mail-in rebate offers, an attempt is made to de
termine the proportion of customers who take advantage
of the rebate, and the reported prices are adjusted
accordingly.
Cents-off coupons. Generally, no adjustments are made
for coupons presented by customers as price reductions
at the time of payment. Research has indicated that less
than 10 percent of customers take advantage of these
coupons. One exception is when the coupons are attached
to the product for immediate redemption at the point of
purchase, b l s field representatives are instructed in this
latter situation to deduct the amount of the coupon from
the price of the item.
Seasonal items
Seasonal items are those commodities and services that
are not available year round but are available in a sea
sonal pattern. Heavyweight coats, tents, and fresh
peaches are examples of items that are often available
only during certain times of the year. Special procedures
are employed when selecting and pricing these types of
items to ensure they are appropriately represented in the
sample and price changes are correctly included in the
calculation of the c p i .
Although seasonal items can exist in any ELI, some
e l i ’s include an especially large percentage of seasonal
items and receive special treatment. These seasonal e l i ’s
include most apparel items, fresh fruit, and sports and
recreational equipment. The designation of an e l i as sea
sonal or nonseasonal was made at the regional level, using
the four geographic regions in the c p i design. It is not un
common for some items that are seasonal in the Northeast
region, for example, to be sold year round in the South.
After the samples for these seasonal e l i ’s are selected
following the normal sample selection procedure, the
number of quotes is doubled to ensure that, despite the
seasonal disappearance of a large number of quotes, a
large enough residual number of in-season quotes will
exist to calculate the index. The quotes in these e l i ’s are
paired; that is, for each original quote that is selected, a
second quote in the same e l i and outlet is initiated and
priced. In the fresh fruit e l i ’s, one quote of each pair is
designated January-June, and the other quote is
designated July-December. In all other seasonal e l i ’s,
one quote of each pair is designated fall/winter, and one
quote is designated spring/summer. The fall/winter and
spring/summer designations are used for the nonfood
quotes because these are the distinctions that are most
commonly used by the retailing industry to categorize
seasonal merchandise. The seasonal designations are used
to help establish the specific items eligible for each quote
so that year-round items and items from each season are
initiated in their proper proportions.
At initiation and each time an item is priced, the data
for in-season months are collected for every specific com
modity and service priced in the c p i , including year round
items. These data become a part of the item description
and are updated if there is a change. An attempt is made
to price every item every on-cycle month, even during
those months when the item may be out of its indicated
season. If the item is available, the price is collected and
used in the calculation of the c p i . If the item is unavail
able because it is out of season, no further action is taken,
and that item is not used in the calculation of the c p i that
month. Its price is imputed using standard imputation
procedures.
When an item becomes permanently unavailable, the
standard procedure is to substitute the most similar item.
In the case of a year-round item not in a seasonal ELI,
this process takes place as soon as the item is permanently
unavailable. However, for items in seasonal e l i ’s and
seasonal items in e l i ’s that are not designated seasonal,
the period during which a substitution can take place is
restricted to those months when a full selection of
appropriate seasonal merchandise is available.
These special initiation, pricing, and substitution pro
cedures are intended to ensure that an adequate sample
of items is available every month, and the correct balance
of seasonal and year-round items is maintained. As a
result, the estimates of price movement for the e l i ’s that
include seasonal items correctly reflect price changes for
the universe of items included in those e l i ’ s .
Special estimation procedures
A number of special estimation procedures are used
in compiling price information for selected categories of
items in the c p i .
New vehicles. Prices for new cars and trucks selected
for inclusion in the c p i pose a special problem since the
172
manufacturer’s suggested retail (sticker) price does not
represent the transaction price for most new vehicles.
Most automotive dealers offer customers concessions on
the sticker price and include certain dealer preparation
charges. In some instances where models are in high
demand, dealers will even charge an additional markup
beyond the sticker price. When pricing new vehicles, b l s
field representatives obtain separately the base price and
all the options on the selected vehicle. In addition, they
obtain from the dealer the average concession and/or
markup during the preceding 30 days. This enables b l s
to estimate the average price of the vehicle after concessions/markups.
Used cars. The only expenditures on used cars included
in the c p i market basket are those for previously owned
cars consumers purchased from the business or govern
ment sectors and the profit of dealers on the sale of used
cars. (See Special expenditure weight issues above.) The
used-car sample was selected from types of cars purchased
for use by businesses and governments. The sample con
sists of 1- through 5-year-old models. Average wholesale
prices of clean cars sold at used-car auctions are published
by the National Automotive Dealers Association. The
average of these prices is adjusted for depreciation using
the difference in prices between model years for the same
model car. The prices used in the index are a 3-month
moving average of the average wholesale price adjusted
for depreciation.
Natural gas. The energy value of natural gas varies
according to the quality of the gas supplied, b l s attempts
to price a constant amount of energy consumption for
natural gas. When natural gas is sold by volume—e.g.,
cubic feet—the amount of gas needed to produce a con
stant amount of energy will vary depending on the heating
value of the gas. To ensure that a constant amount of
energy is being priced, the amount of gas consumed is
adjusted each month based on the current heating value.
Thus, through time, a constant amount of energy is
priced. The current adjusted consumption is calculated
as follows:
Current adjusted consumption = original consump
tion X (original heat value/current heat value).
Health insurance. Health insurance is not directly priced
in the CPI. The price change is imputed from the price
movement of the various services that are covered by
health insurance and from the change in the ratio of
retained earnings to benefits paid by type of health
insurance carrier —Blue Cross/Blue Shield or other. (For
additional detail, see Special expenditure weight pro
cedures.) Thus, the price change for health insurance, by
type of carrier, is estimated each month by the product
of two relatives—one for the change in the various
assigned medical care items (e.g., physician services,
hospital rooms, etc.) and the other for the change in the
retained earnings ratio of the carrier.
Automobile finance charges. The price used in the CPI
for automobile finance charges represents the amount
paid for financing a loan with fixed characteristics such
as downpayment percentage, term of the loan, type and
model of car, etc. The price change is affected by two
items—the interest rate on the loan and the changes in
the amount financed due to price movement for new cars.
The automobile financing charges index is estimated each
month by the product of two relatives, one for changes
in interest rates charged on new-car loans and the other
for changes in new-car prices.
Quantity discounts. Many items in the c p i are sold both
individually and in quantity. When consumers are able
to purchase an amount greater than a single unit at a dis
counted price, the first multiple unit price is reported for
use in the CPI. For example, if the 12-ounce can of corn
being priced can be purchased at 25 cents for a single can,
three cans for 69 cents, or five cans for $1, the price used
in the c p i will be the per-ounce price of the three cans.
Unit-priced fruits and vegetables. When pricing fresh
fruits and vegetables that are sold on a unit basis, two
of the items are weighed to determine an average weight
for the item. This helps to reduce the variability in the
size that occurs among individual, loose-produce items
and is not overly burdensome for the data collection proc
ess. For example, if the item being priced is Red Delicious
apples and the price is 50 cents, the b l s field staff will
report the price of one apple and the weight of two Red
Delicious apples taken from the produce rack. In com
puting the price per ounce, the weight of the two apples
is divided by two and the price of an apple is divided by
this average weight.
Bottle deposits, b l s collects information on bottle
deposits for a variety of nonalcoholic and alcoholic
beverages in order to calculate the influence of changes
in bottle legislation on price change. Consumers who
purchase throwaway containers are considered to be pur
chasing the product itself and the convenience of throw
ing the container away. When a local jurisdiction enacts
bottle legislation and no longer allows stores to sell throw
away containers, those consumers who were previously
purchasing throwaway containers may experience a
change in price for the convenience. The price of the same
size bottle of product plus its deposit establishes an upper
bound of the price change since the consumer could retain
the former convenience by now purchasing returnables
and simply throwing them away. In similar fashion, infor
mation about deposits and the status of bottle legislation
173
can be used to estimate price change when a bottle bill
is repealed. Changes due to bottle bills are shown in the
month the legislation is effective.
Sales taxes. Conceptually, the cpi should include all
applicable taxes paid by consumers for consumption
items. A number of products and services are priced with
taxes included since this is the manner in which they are
sold. Many items are priced using their shelf prices and
with taxes subsequently added during compilation of the
c pi . Tax rates for these items are determined from sec
ondary sources based on the tax jurisdiction in which the
outlet is located and the entry level item (ELI) in which
the item is priced. There are instances where the majority
of items within an ELI in a tax jurisdiction are taxable
but some items are not. Taxes are applied to all items
in that ELi/tax jurisdiction, regardless. In other instances,
the majority of items in the eli are not taxable, but some
are. For these cases, no taxes are applied, bls is currently
evaluating the procedure of collecting data on sales taxes
for each individual item priced so that sales taxes may
be reflected more accurately in the index.
Shelter: Rent and owners’ equivalent rent
The rent and owners’ equivalent rent indexes measure
the change in the cost of shelter for renters and owners,
respectively. Price change data for these two indexes come
from the c p i housing survey. Each month, bls field
representatives gather information from renter units on
the rent for the current month and the previous month
and on what services are provided; from owners’ units,
they obtain an estimated or implicit rent; and from all
units, they collect information on characteristics of the
sample housing units and respondents.
Rent, bls estimates the monthly rent price indexes for
each market basket using the rent indexes for the previous
month and for 6 months earlier and 1-month and 6month measures of rent change estimated from the CPI
shelter survey. The estimate of the 1-month rent change
is the sum of the current month’s rents—weighted and
adjusted for 1 month of aging—divided by the previous
month’s sum of weighted rents. The estimate of the 6month rent change is the sum of the current month’s
rents—weighted and adjusted for 6 months of a g in g divided by the sum of weighted rents for the previous 6
months. The current month’s rent index is a weighted
average of the previous month’s rent index moved for
ward by the estimate of 1-month rent change and the rent
index from 6 months earlier advanced by the estimate of
6-month change.
To put this in the form of an equation, let S, be the
set of rental units interviewed in the shelter survey in time
t in a market basket with valid comparable rents in both
time t and in time t—1; and let S6 be the set of units
interviewed in t with valid comparable rent values in both
time t and time t-6. Vacant units that were previously
renter occupied are also included in S, and S6 and have
current (t) and previous (t-1) month’s rents assigned using
a vacancy imputation process. Let the rent, for rental unit
i in time t be rit, and let ait be a factor that adjusts for
the estimated small loss in quality due to the aging it
experienced between t-1 and t. The 1-month and 6-month
estimate of rent change, Rt M and Rt t_6, are calculated
where:
Wj, and Wi6 are the renter units probabilities of
selection adjusted for nonresponse.
Using Rt t_j and Rt t_6 and the indexes for the previous
month, It_p and for the 6 months previous, It_6, bls
computes two preliminary estimates (It_1Rt t-1 and
It_6Rt t_6) of the current month’s rent index, It, for each
market basket. The final rent index for month, t, for each
market basket is the weighted average of the two pre
liminary estimates:
It = A (IM V t > + d - A> dt-6 Rt,t-6)
where:
A = 0.65, the value that simulation studies deter
mined minimizes the mean squared error of the
estimate.20
Vacancy imputation. Vacant units which were previ
ously renter occupied are used in the calculation of Rt M
and Rt t_6. The vacancy imputation process incorporates
several assumptions about the unobserved rents of vacant
units. It is assumed that rents tend to change at a dif
ferent rate for units that become vacant (and are,
therefore, in the process of changing tenants) than for
other units. The vacancy imputation model assumes that,
after an initial lease period, expected rents change at a
steady rate until the old tenant moves out of the unit.
When there is a change in occupant or a unit becomes
vacant, its rent is assumed to “jum p” at some rate, re
ferred to as the “ jump rate.” In markets with generally
rising rents, this jump rate is usually greater than the
average rate of change for occupied units, bls estimates
the jump rate based on nonvacant sample units in the PSU
which have had a change in tenant between t and t-6.
Nonvacant units without a tenant change are used to
calculate the average continuous rate of change. These
20
For a derivation of the optimal value of A, see C. L. Kosary, J.
P. Sommers, and J. M. Branscome, “Evaluation Alternatives to the
Rent Estimator,” Proceedings o f the Business and Economic Statistics
Section, American Statistical Association, 1984.
174
values are used to impute rents for vacant units for
periods t and t-1 from their rent in t-6.21 In general, the
imputed rents, ri t and rj M, of the ith vacant rental unit
in t and t-1 are:
r i,t- i
=
r i , t —6 c 5 a n d r i (t =
where J is the jum p rate for the
is the steady rate o f change.
psu
r i,t- l J
calculated, and C
The imputation of vacant rents ensures that the
unobserved rent change that occurs when a unit becomes
vacant will be reflected in the final rent index. The 6month rent-change estimates will capture these changes
once the units become occupied. However, they will be
missed in 1-month rent-change estimates without vacancy
imputation. Because the final rent index is calculated
using both 1- and 6-month change estimates, omission
or misstatement of rent estimates for vacant units would
lead to BLS missing part of rent changes in the CPI.
Aging adjustment. The aging adjustment accounts for
the small loss in quality as housing units age (or
depreciate) between interviews. The aging adjustment fac
tors, ait can be thought of as l/( l- d ) where d is the
monthly rate of physical depreciation, bls computes
factors for each housing unit with regression-based for
mulas. The formulas account for the age of the unit and
a number of structural characteristics.22 The aging
adjustment procedure was introduced into the c pi in
1988.
The rent figures collected in the housing survey are the
amounts the tenants pay their landlords plus any rent
reductions tenants receive for performing services for the
landlord (sometimes called “ rent as pay” ) plus any sub
sidy payment (such as Section 8 payments) paid to the
landlord. If the rent is lower than prevailing market rents
because the tenant is related to the landlord, the unit is
not used in the calculation. Reductions for any other
reasons are not considered part of the rent.
The collected rents are “ contract” rents; they are the
payment for all services the landlord provides in exchange
for the rent. For example, if the landlord provides elec
tricity, it is part of the contract rent. The c pi item
expenditure weights also include the full contract rent
payment.
Quality adjustment. Quality adjustments made to the
cost of rental housing are used in the rent and owners’
equivalent rent indexes. BLS collects the rent charged plus
a description of major services and facilities provided by
21 For more information on vacancy imputation, see J. P. Sommers
and J. D. Rivers, “Vacancy Imputation Methology for Rents in the
CPI,” P roceedin gs o f the Business an d E con om ic S tatistics Section,
American Statistical Association, 1983.
22
For further information, see William C. Randolph, “ Housing
Depreciation and Aging Bias in the Consumer Price Index,” BLS Work
ing Paper 166, April 1987.
the landlord. If the services and facilities differ between
two collection periods when rents are compared, the rent
for the current period is adjusted to reflect the differences
in services between the time periods. For instance, if the
owner no longer provides a certain utility, bls calculates
an estimate of the value of that utility and adds it to the
current rent in order to have an adjusted rent value. This
adjusted rent is the current cost of the same set of ser
vices provided for the previous rent.
To make quality adjustments in costs of utilities and
water, BLS uses data from the Department of Energy’s
Residential Energy Survey to develop formulas to
estimate utility usage for various types and sizes of hous
ing, in various climates, with different types of heating
and air-conditioning, hot water, cooking stoves, and so
on. Prices for utilities come from the c p i average price
program. A similar, simpler formula is used to estimate
water costs. Research is underway to determine how to
adjust for major quality changes in housing such as in
the number of rooms or bathrooms. Currently, when
such major changes occur, bls omits these observations
from the calculation for estimation of price change.
Owners’ equivalent rent, bls estimates the owners’
equivalent rent index23 by estimating the owners’ implicit
rent, m^, for each owner unit, j, in the sample. In con
trast to the contract rent concept used in the rent index,
the implicit rent is a “ pure rent” ; that is, it excludes
payments for extra services such as utilities and furniture.
Once the implicit rents are estimated, the calculation of
the owners’ equivalent rent index essentially follows that
of the rent index.
The initial value derived for time t for nij t is an
estimate of the rent the owner-occupied housing units in
the housing survey would bring if they were rented. The
estimate is based on the answer to the question, “ If this
house or apartment were a rental unit, how much do you
think it would rent for monthly, including maintenance
but without utilities and furniture?” For owner-occupied
units whose owners are unable to estimate their unit’s
implicit rent, bls uses an imputation procedure that
assigns the implicit rent from a similar unit to any that
have missing values.
To get subsequent values of implicit rent each month,
bls assigns a set of renters, Qj( to each owner unit, j.
This assignment is done on the basis of location within
the PSU, structure type, and structural characteristics, bls
first tries to match owners with renters that fit for all
variables. For those owners for whom a matching set of
renters is not obtained at the first stage, bls relaxes the
23 Substantial changes in the method of measuring price change of
owner-occupied housing were introduced with the index for January
1983 (January 1985 for the CPI-W). For information on the change and
the old method, see “Changing the Homeownership Component of the
Consumer Price Index to Rental Equivalence,” CPI D eta iled R ep o rt,
January 1983.
175
constraints one at a time until a satisfactory set of renters
is found for all the owners. In general, a single renter
may be assigned to sets for estimating no more than three
owner equivalents. When several renters, say nj, are
assigned to owner j, this counts as only 1/nj toward each
renter’s maximum of three owners. However, renters are
only checked against their maximum after a round of
matching, so it is possible for a renter to move more than
three owner equivalents if the renter is matched to more
than three during one round. The sample selection pro
cess, which sampled renters in owner areas at a very high
rate, facilitated the matching of renters to owners. (See
the section on Item and outlet samples for shelter.) bls
estimates the pure rent, Pi, for all the rental units in Q^.
This pure rent estimate, P, is r minus an estimate of the
value of any utilities or furniture the landlord provides.
Qjj is the subset of the renters in Qj that have valid
comparable rents in both t and t-1. % is the subset with
valid rents in both t and t-6. Vacant, previously renteroccupied housing units are eligible for
and Qj6. The
implicit rent, mjt, for owner j in time t is estimated from
the implicit rent for t-6 and the average change in the
pure rent of the units in Qj6:
t = m; , ,
E
, ',‘6 *Qj6
(
Pi,t
l
P i,,-6
+ 6a; \ I ni#1
I 1 16
where pi t is the ith rental unit’s pure rent, is the aging
adjustment factor, and nj6 is the number of rental units
Qj6.
The 1-month previous implicit rent is the current
month’s implicit rent moved back 1 month with the pure
rents in Q-x:
P;i.t-1
E
m j,t-i m jt
ifQji
P i,t + aiP i,t-l
Once b l s obtains estimates of current, 1-month-ago, and
6-months-ago implicit rent for all owners, it proceeds to
estimate the owners’ equivalent rent index for the cur
rent month. The process is similar to that used for the
rent index. There is no problem here with missing price
change for vacant units or aging since the calculation of
implicit rent already adjusts for these considerations.
BLS makes 1-month and 6-month estimates of change
in the owners’ shelter cost for each market basket as
follows: Let S5 be the set of owner units with implicit
rents in both time t and times t-1 and S6 with implicit
rents in t and t-6. Note that owner units may not be in
Sj or S6 if their sets
or Q* are empty. The 1-month
and 6-month estimates of price change for owner units
in each market basket are:
L mjt Wjl
£ mj, Wj«
jeS,
Ri,-,=
VL —mj.t-1
ifS
leS6
Rl,.-6 =
" jl
ieS*
mlj,t—6
Wj6
where the w^’s and wj6’s are the owner units’ probabil
ity of selection adjusted for nonresponse.
As in the rent index, two preliminary estimates of the
current month’s price index for each market basket
(It_j Rt tl and It_6 Rt t_6) are averaged together to get the
final estimate It:
It = A (IM Rt M) + (1-A) (It_6Rt>t_6)
Again, A = 0.65 is the value that minimizes the mean
squared error.
Average Prices
Average prices are estimated from CPI data for selected
food items, gasoline, utility (piped) gas, electricity, and
fuel oil. Average prices for each food item for a specified
unit of size (i.e., pound, gallon, etc.) are published monthly
for the U.S. average and for four regions—Northeast,
Midwest, South, and West. The regional definitions are
those of the Bureau of the Census.
Average prices for gasoline, utility (piped) gas, elec
tricity, and fuel oil are published monthly for the U.S.
average, 4 regions, 14 region/population size-class cross
classifications, and the 15 largest index areas. For utility
(piped) gas, average prices per therm, per 40 therms, and
per 100 therms are published. For electricity, average prices
per kWh and per 500 kWh are published. For fuel oil and
gasoline, the average price per gallon is published. Average
prices for all types of gasoline, leaded regular, unleaded
regular, and unleaded premium are published.
Price quotes for 40 therms and 100 therms of utility
gas and for 500 kWh of electricity are collected in sam
ple outlets for use in the average price programs only.
Since they are for prespecified consumption amounts, they
are not used in the c p i . All other price quotes used for
average price estimation are regular CPI data.
With the exception of the 40 therms, 100 therms, and
500 kWh price quotes, all prices are converted to a priceper-normalized quantity. For example, prices for gallons,
quarts, or pints of milk are converted to prices per ounce.
All prices are then used to estimate a price for a defined
fixed quantity. That is, a price per ounce of milk is
estimated and multiplied by 64 to yield a price per half
gallon, the published quantity.
The average price for collection period, t, is estimated
as:
Ej w u p „ / Pia
Pt
W U / P ia
where Wit is the quote weight as defined in the estima
tion of price change modified to reflect the number of
quotes (M ') usable for average price estimation for the
ELi/PSU/replicate. (Imputed prices are used in estimating
average prices.)
In the equation, Wit is an expenditure weight. Dividing
the expenditure weight by the price, Pia, for a given quote
yields an implicit estimate of quantity. Thus, the average
price is, conceptually, a weighted average of prices where
the weights are quantity amounts.
176
Part III.
Precision of Estimates
An important advantage of probability sampling
methods is that a measure of the sampling error can be
computed directly from the sample data. The c p i sam
ple design accommodates error estimation by making two
or more selections (replicates) of items and outlets within
an index area. Therefore, two or more samples of quotes
in each self-representing p s u and one in each non-selfrepresenting p s u are available. Given this structure, which
reflects all stages of the sample design, variance estima
tion techniques using replicated samples are employed.
Different methods of variance estimation were used for
the c p i during the period 1978-86 than are used in the
current c p i . The sampling of areas, outlets, and items
for the c p i for both periods follows replicated sample
designs. The 1982-84 CE Survey also employs a replicated
sample design. However, the 1972-73 CE Survey, which
provided the expenditure weights for the 1978-86 c p i , did
not use a replicated sample design. To reflect the con
tribution to the variance from the CE during both time
periods, separate methods of total variance estimation
are employed. Both methods are described below.
1978-86
cpi
r(0
ht
r®ht-6
Variance Estimation
The method used to estimate c p i variances depends
upon the statistical independence of the estimated indexes
for individual market baskets. The independence is
violated somewhat by the fact that, for the 1978-86 time
period, a replicate of e l i ’s was used in more than one
PSU within a region. However, since the selection of
specific items to be priced involved considerable sub
sampling of the e l i ’s within outlets, the local area
indexes may be regarded as statistically independent for
variance estimation. For each index area, two indepen
dent estimates of the index were constructed using the
replicates specified in the design. This required calcula
tion of price relatives by replicate for each item stratum
for each time period, application of the replicate relatives
to the previous-period replicate index for a particular item
stratum, and aggregation across item strata to the index
area replicate level.
Squared differences of these indexes (properly scaled)
provide conditional estimates of the variance of the area
index. These variance estimates are conditional on the
values of the base expenditures estimated from the 1972—
73 CE Survey. Unconditional estimates of index vari
ances, including the expenditure weight component of
error, are derived analytically. These area index variances
are the building blocks upon which the variance of
regional and national indexes are based. The variance
estimates may be for the index for all items or for a subset
of items.
Price relatives were computed for each item stratum
for each index area of the c p i . Variance estimation
required that price relatives also be computed for each
item stratum for each index area and replicate. The
methodology for computing the price relatives was the
same for the full index area as for the replicates. All
replicate computations were for the c p i - u population
with sales and excise taxes included.
For commodities and services, each price quote was
sampled independently by index area replicate. For the
shelter survey, for each self-representing PSU, each rental
unit was assigned to replicate A or replicate B. For nonself-representing index areas, the index p s u determined
the replicate for a given rental unit.
For relative computation for rent and owners’ equiva
lent rent, artificial replicate cost weights for each index
area replicate were constructed to provide a basis for
weighting the 1-month and 6-month relatives together.
Let:
denote the ith replicate cost weight for index area h at
month t
denote the corresponding 6-month replicate (i) relative
Then A 0)
j,t is computed by:
,(i)
A<1h, =
,(0
r(i)
AUh,-, r'h. + ° - 3 5 A W h , - < i
The final shelter relatives, R ^ , are computed by:
<
- <
/ A(L
The c p i before 1987 was computed by a chaining proc
ess in which an estimate of expenditure for the previous
month in each item stratum was multiplied by the price
relative to provide an estimate of the current month’s
expenditure for the item stratum. The item stratum
expenditure values, called cost weights and denoted by
Czt, were then aggregated and compared to the total
expenditure in the reference period. Thus, for a single item
stratum index the sequence of computations would be:
^zt = c zt_i Rzt.t-i
and
Izt.o = 7 s - x 100.0
Cz0
Replicate indexes for variance estimation were similarly
constructed. A data base of replicate cost weights was
created, maintained, and updated monthly with corre
sponding replicate relatives. Each replicate index was
computed as the ratio of the updated replicate cost weight
divided by the replicate cost weight in the reference
period:
177
Ihat = T
v -----hap
^
p(>)
l(i)
=
zt,0
zt
x 100.0
r (i)
iap
^ zO
Estimates of the variance of the index, conditional on
the reference period cost weights, for all items or a subset
of items at the national, regional, or area level, were
calculated as follows:
Let the quantity Uht>0 be a relative importance com
puted as a ratio as follows:
a. The numerator is the sum of the total cost weights over
the item strata and pricing cycles being considered for index
area h at time t;
b. The denominator is the summation of all base-period cost
weights for the item strata and index areas being considered.
The corresponding quantities based on the replicate cost
weights for each index area replicate (instead of the total
cost weights) are denoted, respectively:
where p denotes the pivot month. The pivot month is the
month for which a revised index series is linked to the cor
responding unrevised index series, for the 1978-86 CPI,
the pivot month can generally be assumed to be December
1977.
The price change from period t-m to t can be expressed
as:
P C iat,t-m =
Let Var Ciat and Var Ciat_m be the estimated variances
of the cost weights at times t and t-m , respectively. Let
Cov (Ciat, Ciat_m) be the estimated total covariance of the
expenditure weights for t and t-m . The total variance of
the index for period t is:
Var (Iiat) =
U(A)ht,0
or
100 [diat / I ia t-J -1 ]
U(B)
u ht,0
I.iap 2
c
The price change for period t relative to a period m
months earlier is denoted by:
V t-m =
Vo
'
Ciapj
-2
100 2r
Var (PCiat>t_m) =
Cov(ItA It_m,„)= E ^ [ ( U <Ai,o -U ht,0)(U<Ai_m,0-U h,_mi0)
< w
“2 If,t-m C ° v dt,0 ’ It-m.O^]
The foregoing discussion has treated two replicates. The
method can be extended for those index areas with four
and six replicates.
Unconditional estimates of the total variance of the
index or of price change for all items or a subset of items
at the national, regional, or area level are calculated as
follows. Let the value, Ciat, be the cost weight at time t
for item or item aggregate i, for area or area aggregate
a. The index for time t can be denoted by:
/ G
mt
C„
C;lat
Glat-m
Var (Ciat.m)
Cov(G t,C: t )
In order to estimate the cost-weight variances and
covariances in the above expressions, estimates of two
more conditional covariances of the index must be given.
The conditional covariance of the index between items
j and j ' in item aggregate i, aggregate area a, and time
t is estimated by:
C ov
One variance estimator of Itt_m is estimated by the
Taylor series approximation:
c„
Var(Ciat)+
-2
In order to compute the estimate of the variance of
It t_m the covariance of the numerator and denominator
must be estimated. The estimated covariance of It 0 and
I t_m o is the sum over all index areas h being considered:
Cov (Ciap> Ciat)
The total variance of price change from period t-m to
t is:
lt-m,0
o2 < V J - T T 1- ? K<i.,o> + < W 2
rit-m.ol
Var (Ciap)
*ap.
A variance formula for the index It 0 is the sum over
all index areas, h, being considered:
- E - I T [(U<Ah.,o - u h.,0>2 + (u<Bh,,0 - Uh ,/ 1
CIM\.2
Var (Ciat) +
(Ijat, V a t)
=
1 /2 Id ja A t "
W
"1" djaBt “ Ijaft)
’
<V«At
~
Vaft>
dj'aBt “ Vaft))
Here, Ijaft represents the full sample index for the item
stratum, aggregate area, and time t and IjaAt and IjaBt
represent the indexes for the corresponding replicates.
Similarly, the conditional mixed covariance of the
index between times t and t-m , and item strata j and j '
in aggregate area a is estimated by:
C ° v d jat, lj'at-m ) =
1 / 2 IdjaAt “ Ijaft) ' d j'aA t-m “ V a ft-m )
"1”
d j a B t
~
Ija ft)
d j 'a B t-m
~
l j 'a f t - m ) )
Given these index variance and covariance estimators,
the cost-weight variances and covariances in the expres
sions for unconditional variance above can be estimated
in the following way:
178
An estimate of the cost-weight variance for item
aggregate i, area aggregate a, and time t is given by:
An estimate of the covariance of the cost-weights be
tween times t and t-m is given by:
Var Cial = <Ciap / Iiap)2 • Var Iial
Cov (Ciat, Ciat_m) = (Ciap / Iiap)2 • Cov (IJat, Ijat_m)
+ E
( " W 2 lOjht)2 - Var Ijh.I ' VarE Cjhp
hea jei
+ E ?. d^ihp)2 [«ih, >ih,-m-Cov (Ijh„ Ijht_m)]-VarECjhp
+ E E E OAIjhp Ij hp) (Ijh, ' Via - Cov «jhf Vh.)l
+?
hea jei
hea jei
hea jei j #j
• CovE (Cjhp, Cj-hp)
.?J '^ W V h p ) tW tu -m -C°VCjh,. V h .Jl
j
‘ CovE (Cjhp, Cj >hp)
where:
VarE Ciap is the variance of the cost-weight for item
aggregate i, area aggregate a at pivot month p,
CovE (Cjhp, Cjhp) is the covariance between the costweights for item strata j and j ' in item aggregate i and
index area h in pivot month p,
Iiat and Iiap are estimates of the index at times t and
the pivot month p, respectively,
Var Iiat is the conditional variance of the index at time
t, and
Cov (Ijat Ij,at) is the conditional covariance of the
index between item strata j and j ' in item aggregate
i, as previously defined above.
Summations indicated above are over all index areas in
area aggregate a, and over item strata in item aggregate i.
Estimates of the components of the conditional
variance of the index have been given above. The remain
ing quantities in the above expression are estimated as
follows:
An estimate of the variance of the pivot month costweight, Cjhp, is given by:
VarE Cihp = (Fjh)2
Var Cjh
Here, Fjh is the factor used to update the base-period costweight to the pivot month p, and
VarE Cjh 72_73 is the variance of the base-period costweight, which is estimated by:
VarE ^jh,72-73 = 1/2 f(Cjh,A,72-73 ~ ^jhf 72_73)2
+ (Cjh.B,72-73 “ Cjhf,72-73)2]
Here, Cjhf 72_73 denotes the full sample estimate of the
cost-weight for the base-period 1972-1973, and Cjh(A72_73
and Cjh B72_73 denote the corresponding replicate costweights.
An estimate of the pivot month cost-weight covariance
between item strata j and j ' is computed in a similar
manner:
CovE (Cjhp, Cj-hp) = Fjh • Fj-h [CovE (Cjh>72_73, Cj-h 72_73)]
where Fih and Fj-h are defined as above and the baseperiod covariance term is estimated by:
CovE (Cjh>72_73, Cj«h>72_73) =
1 / 2 [(C jh A>72_73 - C jhf 72_73) •
+ (C jh.B ,72-73 ~ C jhf 72_73) •
(C j'hjA 72_73 - C j-hf 72_73)
(C j«h)B>72_73 - C j-hf>72_73)]
Estimating Variances of the Index
and of Price Change, Starting in 1987
Variance estimates for the index and price change are
being computed by different methods starting in 1987.
The most important difference between these estimators
and those previously described is that they directly
incorporate the contribution to the variance from the
expenditure weights (the aggregation weights) in the com
putation of the unconditional variance of the index or
price change. Variance estimators previously described
measured the conditional variance of the index or price
change given the expenditure weights, and then measured
the unconditional variance using a two-step estimation
process involving first estimating the unconditional
variances of the expenditure weights, and then incor
porating the variance of the expenditure weights into a
final estimate of the unconditional variance of the index
or price change.
As in the past, all computations will be based on the
C P i - u population with sales and excise taxes included.
Expenditure weights used in computation of the index are
derived from the 1982-84 Consumer Expenditure Survey.
The variance estimators given here depend on the
aggregation structure of the index which supports the con
struction of indexes for higher levels of aggregation such
as item groups, regions, and the Nation from those for
basic item strata and index areas. The estimators also
depend on the replicate structure of the index sample. Use
of replicates provides a means of measuring the overall
variation of the index from those computed over subsets
of the sample. The full sample for each index area com
prises two or more replicate panels, of which half are
designated “ odd” and half, “ even.”
Variance estim ates for the index
To estimate the unconditional variance of an index,
consider the index for a particular item aggregate I,
geographic area aggregate M, for month t expressed as:
IX(I,M,t) = IX(I,M,p) • WI(I,M,t) / WI(I,M,p)
Here, p denotes the pivot month (December 1986). The
pivot index, IX(I,M,p), which in many cases will be 100,
acts as a normalization constant. The weighted index or
cost weight, WI(I,M,t) is the sum over all index areas in
179
the aggregate area M of the product of individual index
area aggregation weights multiplied by the corresponding
index. For example, for the full sample for month t and
index area m, the weighted index for item aggregate I is
given by:
W I(I,m,f,t) = AW(I,m,f) • IX(I,m,f,t)
where AW(I,m,f) and IX(I,m,f,t) are respectively the full
sample aggregation weight and month t index for the item
aggregate and index area. Similarly, for any sample
replicate r, for month t, and index area m, the weighted
index for item aggregate I is given by:
W I(I,m,r,t) = AW(I,m,r) • IX(I,m,r,t)
where AW(I,m,r) and IX(I,m,r,t) are respectively the cor
responding aggregation weight and index for replicate r.
Separate aggregation weights and indexes are computed
and maintained for each sample replicate.
If the variances of WI(I,M,t) and WI(I,M,p) and their
covariances are known, then the variance of IX(I,M,t)
can be estimated by a Taylor series linear approximation.
Components of these variances can be estimated in the
following way:
Consider the weighted index WI(I,M,t) = E WI(I,m,t)
meM
where WI(I,m,t) is the weighted index for index area m
in aggregate area M.
Then:
areas. Hence there are eight major areas in the Nation.
For aggregate areas larger than one index area, estimates
of between-index-area covariances for each ordered pair
m and m ' of different index areas in the same major area
are given by:
CovA(I,m ',m ,t) = 1/2 { [(W I(I,m ',odd,t) - W I(I,m ',f,t))
• (W I(I,m,odd,t) - W I(I,m,f,t))
+ (W I(I,m',even,t) - W I(I,m ',f,t))
• (WI(I,m,even,t) - WI(I,m,f,t))]}
Between-index-area covariances for the pivot month
p are computed similarly.
The estimator of the covariance between WI(I,M,t) and
WI(I,M,p) has two components. The first comprises the
sum across all index areas in the aggregate area of the
within-index-area time covariance between the weighted
indexes for month t and month p:
E
meM
• (W I(I,m,odd,t) - WI(I,m,f,t))l
+ [(WI(I,m,even,p) - W I(I,m,f,p))
• (WI(I,m,even,t) - WI(I,m,f,t))]}.
The second component comprises the sum across pairs
of index areas m and m ' in the aggregate area of the
mixed covariances between the weighted index for index
area m ', in pivot month p, and that for index area m,
in month t. It is given by:
Var(WI(I,M,t)) = 0 ^ ( 1 , M,t)
=
E o2WI(I,m,t) + E
meM
E
meM m ' #m
a2WI(I,m,t) = 1/2 [(WI(I,m,odd,t) - W I(I,m,f,t))2
+ (WI(I,m,even,t) - W I(I,m,f,t))2]
Here, W I(I,m,odd,t) and WI(I,m,even,t) are weighted
indexes for odd and even replicates, respectively, defined
by:
W I(I,m ,odd,t) = 2/NR(m)
E
• (W I(I,m,odd,t) - WI(I,m,f,t))]
-I- [(WI(I,m' ,even,p) - W I(I,m ' ,f,p))
• (WI(I,m,even,t) - WI(I,m,f,t))]}
Hence, the covariance between WI(I,M,t) and WI(I,M,p)
is given by:
CovWI (I,M ,p,t) = CovT(I,M ,p,t) + CovM(I,M ,p,t)
An estimate of the unconditional variance of IX(I,M,t)
is then:
°2IX(I,M,t) = IX(I,M,p)2 • [l/W I(I,M ,p)]2
• [ ^ ( I . M . t ) + (W I(I,M ,t)/W I(I,M ,p))2 • a2WI(I,M,p)
- 2 (WI(I,M,t)/W I(I,M,p)) • CovWI(I,M,p,t)]
E A W (I,m ,r,t)IX (I,m ,r,t),
and
E
The formula for the estimated bias of the index is:
AW (I,m ,r,t)-IX(I,m,r,t)
r even
and NR(m) denotes the total number of replicates for
index area m.
Each index area is in 1 of 4 Census regions. Each region
can further be divided into two major areas, one com
posed of the self-representing (A) index areas and one
composed of the non-self-representing (non-A) index
E
meM m '
r odd
WI(I,m,even,t) = 2/NR(m)
E
E
CovM(I,m ',m ,p,t) =
meM m ' #m
1/2 {[(W I(I,m ',odd,p) - W I(I,m ',f,p »
CovM(I,M,p,t) =
CovA(I,m ',m ,t)
where cr2WI(I,m t) is the variance of the index area level
weighted index and CovA(I,m ',m ,t) is the betweenindex-area covariance for the index month t.
The variance of the index area level weighted index is
estimated by:
E CovT(I,m,p,t) =
meM
1/2 {[(WI(I,m,odd,p) - W I(I,m,f,p))
CovT(I,M ,p,t) =
BiasIX(I,M,t) = IX(I,M,p) • {WI(I,M,t)
• o2WI(I,M,p) / (WI(I,M,p))3
- CovWI(I,M ,p,t) / (WI(I,M,P))2}
The mean squared error of the index is then estimated
by:
MSEIX(I,M,t) = o^xfl.M .t)
180
+ (BiasIX(I,M,t))2
Variance estimates for price change
The estimated price change, PC (I,M ,t't), from month
t ' to month t for item aggregate I and area aggregate M
is computed by:
PC(I,M,t',t) = (100 • WI(I,M,t) / WI(I,M,t')) - 100
Thus, price change is also a simple function of the ratio
of weighted indexes for two time periods. The formula
for the estimated variance of price change is given by:
(^ped.M.t'.t) = (100/WI(I,M,t'))2
• [a2WI (I,M,t) + (WI(I,M,t) / WI(I,M,t'))2 • o ^ d .M .t')
- 2 • (WI(I,M,t) / WI(I,M,t')) • Covwl(I,M,t',t)l
The formula for the estimated bias of price change is:
Biaspc(I,M,t',t) = 100 • (WI(I,M,t)
•
/ (WI(I,M,t'))3
- CovWI(I,M,t',t) / (WI(I,M,t'))2}
The mean squared error of price change is then
estimated by:
MSEpc(I,M,t',t) = (^p^I.M.t'.t) + (BiasPC(I,M,t',t))2
Nonsampling Error
c p i estimates are subject to nonsampling error as well
as sampling error. Surveys involve many operations that
must be performed in order to produce the final results.
All of these are potential sources of nonsampling error.
The errors arise from the survey process regardless of
whether the data are collected from the entire universe
or from a sample of the population. The most general
categories of nonsampling error are coverage error, non
response error, response error, processing error, and
estimation error.
Coverage error in an estimate results from the omis
sion of part of the target population (undercoverage) or
the inclusion of units from outside of the target popula
tion (overcoverage). Coverage errors would result from
the omission of cities, households, outlets, and items that
are part of the target populations from the relevant sam
pling frames or the double counting or inclusion in the
frames of such sample units when they should not be.
A potential source of coverage error is the timelag be
tween the Point of-Purchase Survey and the initiation of
price collection for commodities and services at resampled
outlets. Because of the timelag, the products offered by
the outlet at the time pricing is initiated may not exactly
coincide with the set from which the p o p s respondents
were purchasing.
Nonresponse error results when data are not collected
for some sampled units because of the failure to inter
view households or outlets. This can occur when selected
households and outlets cannot be contacted or refuse to
participate in the survey. Nonresponse rates at initiation
for the CPI commodities and services and housing surveys
are shown in tables 2 and 5. This nonresponse could bias
the c p i if the rate of price change at the lost survey units
differed from the rate of price change at the survey units
successfully initiated. Nonresponse rates during monthly
pricing for the c p i commodities and services and hous
ing surveys are shown in tables 3 and 6.
Response error results from the collection and use in
estimation of incorrect, inconsistent, or incomplete data.
Response error may arise because of the collection of data
from inappropriate respondents, respondent memory or
recall errors, deliberate distortion of responses, inter
viewer effects, misrecording of responses, pricing of
wrong items, misunderstanding or misapplication of data
collection procedures, or misunderstanding of the survey
needs and/or lack of cooperation from respondents, b l s
is currently developing a reinterview methodology for
investigation of various kinds of response variance in the
c p i . Response variance can be measured through a rein
terview conducted under conditions identical to those of
the original interview. Work in the area of response biases
has not yet been developed for the CPI. The pricing
methodology in the commodities and services component
of the c p i allows the previous period’s price to be
available at the time of collection. This dependent
pricing methodology is believed to reduce response
variance for measuring change, but may cause response
bias and lag.
Processing error arises from incorrect editing, cod
ing, and data transfer. Survey data are converted into
machine-readable form by two independent key entry
operators, and discrepancies are resolved by a third per
son. Processing errors can be introduced by an incorrect
resolution or by an identical miskeying of an element by
two operators. Errors can also result from software prob
lems in the computer processing which cause correctly
keyed data to be lost. Computer screening and profes
sional review of the data provide checks on processing
accuracy. Occasional studies of these processing errors
in the c p i have shown them to be extremely small.
Estimation error results when the survey process does
not accurately measure what is intended. Such errors may
be conceptual or procedural in nature, arising from a
misunderstanding of the underlying survey measurement
concepts or a misapplication of rules and procedures. A
source of estimation error due to conceptual problems
was the treatment of housing before 1983, which failed
to distinguish between the consumption and investment
aspects of homeownership. Prior to implementation of
the change to the owners’ equivalent rent, an experimental
181
m easure using rental equivalence diverged considerably
from the o fficia l c p i .24
Substitutions and adjustments for quality change in the
items priced for the c pi are possible sources of estima
tion error due to procedural difficulties. Ideally, c pi data
collection forms and procedures would yield all informa
tion necessary to determine or explain price and quality
differences for all items defined within an ELI. Since such
perfect information is not available, BLS economists sup
plement directly collected data with secondary data.
Estimation error will result if the bls adjustment process,
which necessarily has a significant judgment component
and may have key data unavailable, is misapplied, or if
it consistently overestimates or underestimates quality
change for particular kinds of items. While individual
problems arising from substitution in estimating quality
change have been identified, the evidence to date is that
on average there is no systematic bias from this process.
Cases where price change is overestimated are about as
frequent as those where it is underestimated.
An example of potential estimation error, which is
similar to the issue of quality change in commodities and
services, is the effect of the aging of housing units. Until
1988, bls did not adjust for the slow depreciation of
houses and apartments over time. Current BLS research
indicates that annual changes for the residential rent and
owners’ equivalent rent indexes would have been 0.2 to
0.3 percent larger if some type of aging adjustment had
been included.
The total effect of nonsampling error on the accuracy
of c pi estimates varies depending on the type of data col
lected, the methods of collection, the data processing
routines, and the nature and complexity of the estima
tion processes. The cumulative effect of nonsampling
error can be much greater than the effect of sampling
error. A long-term goal of bls is the publication of a full
error profile for the c p i .
Response rates: Commodities and services
Response rates at initiation of commodities and serv
ices components were calculated as follows: Approx
imately one-fifth of the p s u ’s are initiated each year.
The sample initiated in 1986 is presented here. Although
it is somewhat atypical in size-class distribution be
cause the areas are primarily non-self-representing, the
magnitude of response is similar to previous years’
results.
Approximately 70 percent of the data is collected for
pricing by personal visit and the remainder is collected
by telephone.
24 See Robert Gillingham and Walter Lane, “ Changing the Treat
ment of Shelter Costs for Homeowners in the Consumer Price Index,”
Statistical Reporter, December 1981.
T a b le 2 .
R e s p o n s e r a te s a t In it ia t io n f o r CPI c o m m o d it ie s a n d
s e r v ic e s In t h e o u t le t s u rv e y , 1 9 8 6
Type of Interview
Number of
outlets
Number of
quotes
Total designated outlets or q u o te s ........
Total outlets or quotes interviewed........
At least one quote obtained ................
Unable to price during pricing period .
Refusal .....................................................
Out of season .........................................
6,563
5,823
5,329
274
138
19
27,512
24,118
22,428
1,131
500
59
Total outlets or quotes not Interviewed .
No items available for ELI p ric in g ___
Out of business.......................................
Out of scope ...........................................
Unable to locate .................... ................
Outside pricing a r e a ...............................
Outlet moved ...........................................
740
352
124
36
136
83
9
1,352
770
542
115
342
248
25
Percent of interviewed outlets or quotes
responding ...............................................
Percent noninterview of total ..................
92.6
11.3
93.0
12.3
Response rates at pricing of commodities and services
were calculated from data shown in table 3.
T a b le 3 .
N u m b e r o f o u t le t s a n d q u o t e s f o r p r ic in g fo r c p i
c o m m o d it ie s a n d s e r v ic e s b y o u t le t r e s p o n s e c la s s if ic a t io n
Response classification
Outlets
Total ............................
Total outlets or quotes
sent to the f ie ld ........
Reporting one quote
available......................
Temporarily unavailable
Outlet refusal ................
Quotes not available . . .
Outlet out of season . . .
Outlet out of business .
Unable to locate outlets
Out of a r e a ....................
Moved ..........................
Out of s c o p e ..................
Total out of s c o p e ........
Quotes not sent to field
Quotes
April 1987 May 1987 April 1987 May 1987
98,126
100,240
23,104
23,159
96,111
97,652
21,998
841
15
69
24
49
13
8
5
5
226
21,970
822
23
57
36
56
11
16
2
6
308
92,810
2,889
23
140
47
151
18
44
7
7
202
2,014
93,859
3,298
48
107
65
194
23
19
3
11
275
2,588
For outlets reporting in April or May 1987, not all
quotes received are eligible for estimation. Of the 92,810
and 93,859 quotes returned with a price, 87,822 and
89,342 quotes are of the same version and are likely to
be used in index estimation. Quotes requiring initiation,
reinitiation, and deletion are not used in the index, and
quotes involving some form of substitution may be used
in estimation. Of the 92,810 and 93,859 quotes priced,
11,582 and 12,746 quotes are deemed not comparable and
not used in index estimation. Thus the number of quotes
used in index estimation, 81,227 and 81,113, respectively,
account for 84.5 percent and 83.1 percent of the quotes
sent to the field.
182
Table 4. Number of quotes for commodities and services
pricing by quote response classification
R e s p o n s e c la s s ific a tio n
A p r il 1 9 8 7
M ay 19 8 7
N u m b e r o f o u t le ts r e p o rtin g
o n e q u o t e a v a ila b le ............................................................
2 1 ,9 9 8
2 1 ,9 7 0
N u m b e r o f q u o t e s re tu rn e d w ith p r i c e ___
9 2 ,8 1 0
9 3 ,8 5 9
Q u o t e s r e p o r tin g :
S a m e v e r s i o n .................................................................................
R e in itia tio n .....................................................................................
D e le tio n o f q u o t e fro m o u t l e t ................................
C h a n g e In v e r s io n (s u b s tit u t io n ):
W i t h o u t o v e r l a p ...................................................................
W ith o v e r la p ..........................................................................
D a ta c o lle c t e d :
F o r p re v io u s p e rio d ..............................................
F o r p r e v io u s a n d c u r r e n t p e rio d . . .
In itia tio n ........................................................................................
8 7 ,8 2 2
28
173
8 9 ,3 4 2
25
145
2 ,1 2 9
472
1,9 6 8
358
20
47
2 ,1 1 9
40
11
1 ,9 7 0
telephone and personal visit; 53.2 percent were collected
by telephone, 44.7 percent were collected by personal visit,
and 2.1 percent did not report the collection mode.
Table 5. Initiation results after two screening attempts, JanuaryJune 1987
Ite m
To ta l
In itia tio n
r e s u lts
E x p e c te d
.............................................................................................................
A c t u a l .......................................................................................................................
R e m a in in g .........................................................................................................
P e r c e n t c o m p le te ....................................................................................
2 4 6 ,0 0 0
2 6 4 ,3 0 0
3 0 ,4 0 0
8 8 .5
In itia tio n r e s u lts fo r
o w n e r s a n d re n te rs
O w n e rs
E x p e c te d
.............................................................................................................
F o r e c a s t ................................................................................................................
C u r r e n t ....................................................................................................................
R e n te r s
2 0 ,3 6 6
2 0 ,8 4 2
18 ,0 4 0
3 9 ,2 0 0
3 5 ,4 4 2
2 9 ,9 3 3
Response rates: Housing
Table 6.
From the beginning of the survey through the collec
tion period ended June 30, 1987, b l s obtained the
following results in listing, screening, and initiation of
units. Approximately 485,000 units were listed in 10,000
segments, of which 264,300 units were chosen to be
screened. This was approximately 18,000 screenings above
the 246,000 projected from the 1980 census data. How
ever, the extra units were expected because of the inability
to distinguish post-1980 housing during listing. These
numbers can be seen in table 5.
After each panel had passed through two initiation
periods, 88.5 percent of the screening had been completed.
As of October 1987, the sample contained 18,041 owners
and 29,933 renters. The forecast final sample size of units
built before 1980 was about 20,800 owners and 35,400
renters. (See table 5.)
For the period January through June 1987, the 40,133
units which had passed screening were sent for repricing
collection. Of these, b l s contacted and obtained poten
tially usable responses from 34,005 (84.7 percent); 4,009
(10.0 percent) could not be contacted; and 446 (1.1 per
cent) were either temporarily or permanently out of the
scope of the survey.
Of the 34,005 units which provided information, 1,431
(4.2 percent) were not used in estimation because they
were deemed not comparable. The mode of collection used
in pricing the housing survey units is a combination of
Pricing results for January-June 1987
Ite m
Num ber
P e rc e n t
T o ta l c o n t a c te d ...........................................................................................
In fo r m a tio n o b ta in e d ...................................................................
C o u l d n o t p r o v id e In fo r m a tio n .......................................
R e fu s a l .............................................................................................................
3 5 ,6 7 8
3 4 ,0 0 5
833
840
8 8.9
8 4 .7
2.1
2.1
T o ta l n o t c o n t a c t e d .................................................................................
N o t c o n ta c te d ........................................................................................
O u t o f s c o p e ...........................................................................................
4 ,4 5 5
4009
449
11.1
10 .0
1.1
T o ta l n u m b e r o f u n i t s ..........................................................................
4 0 ,13 3
10 0
The following table provides a distribution of all con
tacted units by their usability in estimation by the mode
of collection. Some units were contacted at the place of
residence. For renter units, about one-third of the re
sponses were obtained from the managers of the
apartments.
Table 7. Usability of information for contacted units by type
of contact
T y p e o f c o n ta c t
T o ta l
T o ta l .................................................
O c c u p i e d u s a b le . . .
V a c a n t u s a b le ..............
O c c u p i e d u n u s a b le .
R e fu s a l ...................................
C o u l d n o t p r o v id e
in fo r m a tio n .........................
3 5 ,6 7 8
3 0 ,3 5 9
2 ,2 1 5
1 ,4 3 1
840
83 3
P e rso n a l
v is it
T e le p h o n e
N o t r e p o r te d
1 5 ,9 4 8
1 3 ,0 9 3
1 ,5 3 6
5 42
416
1 8 ,9 8 2
1 6 ,8 4 3
519
8 10
3 79
748
423
16 0
79
45
361
431
41
Technical References
Afriat, S.
The P rice In dex. Cambridge University Press, 1977.
Allen, R.G.D.
1975.
In dex N u m bers in T heory an d P ractice. Macmillan,
Armknecht, Paul. “ Quality Adjustment in the c p i and Methods to
Improve It,” Proceedings o f the Business a n d E conom ic S tatistics Sec
tion , American Statistical Association, 1984, pp. 57-63.
Armknecht, Paul A., and Weyback, Donald. “Adjustments for Qual
ity Change in the c p i : An Update,” P roceedin gs o f th e Business a n d
E con om ic Statistics Section, American Statistical Association, 1987,
forthcoming.
Banerjee, K.S. C o st o f L ivin g In dex N u m bers: Practice, P recision,
an d Theory. Marcel Dekker, Inc. 1975.
183
Technical References-Continued
Blackorby, C.; Primont, D.; and Russell, R.R.
D u ality, Separabil
ity, a n d F unctional Structure: Theory an d E con om ic A p p lication s.
American Elsevier, 1978.
Blanciforti, Laura A., and Galvin, John M. “ New Approaches for
Automobiles in the CPI,” P roceedin gs o f the B usiness an d E con om ic
Statistics Section, American Statistical Association, pp. 64-73.
Blanciforti, Laura. “Price Measurement for Interarea Comparisons,”
Paper presented at the meeting of the American Statistical Association,
New Orleans, December 1986.
Braithwait, S.D. “The Substitution Bias of the Laspeyeres Price
Index: An Analysis Using Estimated Cost-of-Living Indexes,”
A m erican E con om ic R eview , March 1980.
Christensen, L.R., and Mauser, M.E. “ Cost-of-Living Indexes and
Price Indexes for U.S. Meat and Produce, 1947-71,” in H ousehold P ro
duction a n d C on su m ption , Nestor Terleckjy, ed., Studies in Income
and Wealth, 40. National Bureau of Economic Research, 1975.
Cohn, Michael P., and Sommers, John P. “Evaluation of the Methods
of Composite Estimation of Cost Weights for the CPI,” Proceedings
o f the Business a n d E con om ic S tatistics Section, American Statistical
Association, 1984, pp. 466-71.
Diewert, W.E.
“ Exact and Superlative Index Numbers,” Journal o f
E con om etrics, Vol. 4, 1976, pp. 115-45.
Hageman, Robert P. “ The Variability of Inflation Rates Across
Household Types,” Journal o f M on ey, C redit a n d Banking, Vol. 14:4,
1982, pp, 494-510.
Helliwell, John. “ New Network Behind Improved CPI,” P C W eek,
February 3, 1987, pp. C /4-C /10.
Kokoski, Mary F. E xperim ental Indexes o f the C ost o f Living, Bureau
of Labor Statistics, forthcoming, working paper.
Kokoski, Mary F. ‘‘Consumer Price Indexes by Demographic Group, ’’
BLS Working Paper 167, April 1987.
Kosary, Carol L.; Sommers, John P,; and Branscome, James
M. “Evaluating Alternatives to the Rent Estimator,” P roceedin gs o f
th e Business an d E con om ic Statistics Section, American Statistical
Association, 1984, pp. 410-12.
Kosary, Carol L., and Sommers, John P. “ The Effect of Computer
Assisted Telephone Interviewing on Response Inconsistency,” P ro
ceedings o f the Survey Research M eth o d s Section, American Statistical
Association, forthcoming.
Lane, Walter F., and Sommers, John P. “ Improved Measures of
Shelter Costs,” P roceedin gs o f th e Business an d E con om ic Statistics
Section, American Statistical Association, 1984, pp. 49-56.
Essays in the Theory and M easurement o f Consum er Behavior in H on or
o f R ich ard Stone. Cambridge University Press, 1981, pp. 163-208.
Leaver, Sylvia G.; Weber, William L.; Cohen, Michael P.; and Archer,
Kenneth P. “ Determining an Optimal Item-Outlet Sample Design for
the 1987 U.S. Consumer Price Index Revision,” P roceedin g, In tern a
tion al Statistical In stitu te, forthcoming.
Dippo, Cathryn S., and Jacobs, Curtis A. “Area Sample Redesign
for the Consumer Price Index,” P roceedin gs o f th e S u rvey Research
M eth o d s Section, American Statistical Association, 1983, pp. 118-23
Manser, M.E. “A Note on Cost-of-living Indexes and Price Indexes
for U.S. Food Consumption, 1948-1973,” BLS Working Paper 57,
January 1976.
Early, John F., and Sinclair, James H. “ Quality Adjustment in the
Producer Price Indexes,” The U.S. N ation al In com e an d P ro d u ct
A cco u n ts S elected T opics, Murray F. Foss, ed. University of Chicago
Press, 1983.
Marcoot, John L. “Revision of the Consumer Price Index Now Under
way,” M o n th ly L a b o r R eview April, 1985, pp. 27-38.
Diewert, W.E.
“The Economic Theory of Index Numbers, a Survey,”
Early, John F., and Dmytrow, Eric D. “Managing Information Qualty in the CPI,” F ortieth A n n iversary Q u ality C ongress Transactions,
American Society for Quality Control, 1986, pp. 2-9.
Early, John F., and Galvin, John M. “ Comprehensive Quality
Measurement in the Consumer Price Index,” The Juran R eport, Number
10, Juran Institute, Inc., forthcoming.
Frisch, R. “ Annual Survey of Economic Theory. The Problem of
Index Numbers,” E con om etrica, Vol. 4, January 1936.
Gillingham, Robert F. “Estimating the User Cost of Owner-Occupied
Housing,” M o n th ly L a b o r R eview , February 1980, pp. 31-35.
Gillingham, Robert F., and Lane, Walter. “Changing the Treatment
of Shelter Costs for Homeowners in the CPI,” M on th ly L a b o r R eview ,
June 1982, pp. 9-14.
Gillingham, Robert F. “ Measuring the Cost of Shelter for Homeowners: Theoretical and Empirical Considerations,” The R eview o f
E con om ics a n d Statistics, 1983, pp. 254-65.
Gillingham, Robert F., and Greenlees, John S. “The Impact of Direct
Taxes on the Cost of Living,” Journal o f P olitical E con om y, 95(4),
August 1987.
Poliak, Robert A. “ The Theory of Cost-of-Living Indexes,” P rice
L evel M easurem ent, Proceedings of a conference sponsored by Statistics
Canada, 1983.
Randolph, William C., and Zieschang, Kimberly D. “ Aggregation
Consistent Restriction-Based Improvement of Local Area Estimation,”
P roceedings o f the Business an d E conom ic Statistics Section, American
Statistical Association, forthcoming.
Randolph, William C. “ Housing Depreciation and Aging Bias in the
Consumer Price Index,” BLS Working Paper 166, April 1987.
Randolph, William C. “ Short-Term Quality Change and Long-Term
Vintage Effects,” Journal o f Urban E con om ics, forthcoming.
Sommers, John P., and Rivers, Joseph D. “ Vacancy Imputation
Methodology for Rent, in the CPI,” P roceedin g o f th e Business, an d
E con om ic S tatistics S ection , American Statistical Association,
pp. 201-05.
Triplett, Jack E., and McDonald, Richard J. “Assessing the Quality
Error in Output Measures: The Case of Refrigerators,” R e view o f
In com e an d W ealth, 23, June 1977.
Triplett, Jack E. “ Reconciling the CPI and PCEi Deflator,” M o n th ly
L a b o r R eview , September 1981, pp. 3-15.
184
CPI Appendix 1.
Date
18901 ...
Chronology of changes in the Consumer Price Index, 1890 to date
Survey providing
expenditure weight
Base
period
Group
weights
Item
weights
None
1901
Varied
1917-19
1913
1919 . . . . 1917-19
Feb........
1921
Sept. ...
1935
Dec........
1923-25
19354
Aug.__ 1934-36 ®1934-36 1935-39
1940*
May . . . .
1941s
July . . . .
1943
Sept. ...
1945
Jan........ 1947-49 111934-36
195110
Census
Number
providing
of areas
population
included
weights
None
Family
composition
Varied Two or more
persons.
232
Earnings
of
chief earner
Source and
amount of
family income
Salaried worker
earning $1,200
or less during
year. No limit
ation on wage
earners.
No limitation.
No limitation.
Minimum of
Salaried worker At least 75 per
husband, wife, earning $2,000 cent from
and 1 child
or less. No
principal
who was not a limitation on
earner or
boarder or
wage earners.
others who
contributed all
lodger. No
boarders nor
earnings to
more than 3
family fund.
lodgers
present.
Economic level,
length of
residence,
nativity, and
race
Tltle(s)
No limitation.
Cost of living.
charity famll(os; white only;
in area entire
year and in
the United
States 5 years
or more; no
non-English
speaking
families.
3Average
1920-30
1930
733
Two or more
persons. Not
more than 2
boarders or
lodgers, or
guests for
more than 26
guest-weeks.
At least $300.
Salaried
worker earnIng less than
$2,000 during
year or less
than $200
during any
month. No
upper limltation on wage
earners.
At least 1,008
No relief fami- Indexes of the
At least $500.
lies, either on
cost of living
Less than one- hours spread
of wage earnover 36 weeks. direct or work
fourth from
ers and lowerrelief; white
interest,
only, except
salaried
dividends,
workers In
where black
royalties,
population was large cities.
speculative
significant
gains, rents,
part of total;
gifts, or Income in kind.
in area 9
No rent In
months or
payment of
more.
services. Less
than 3 months’
free rent. No
subsidiary
clerical work
er earning
$2,000 or over.
34
®1940
Consumer's
Price Index
for Moderate
Income Fami
lies in Large
Cities.
1950
Two or more
persons.
Family income Family head
No limitation.
(Family Income under $10,000
must have
after taxes in
been emnot in excess
of $10,000.)
the survey
ployed at
year. No minileast 26
mum income
weeks.
limit, except
that families
with no Income from
wages or
salaries were
excluded.
See footnotes at end of table.
Length
of
employment
185
No exclusion
for receipt of
relief as such,
but only families with wage
or salary
earnings ineluded. No
length of residence, nativity,
or racial
limitations.
CPI Appendix 1.
Date
Jan.
19531
Jan.
1962
Jan.
19641
Jan.
19661
Jan.
19711
Jan.
19782
Survey providing
expenditure weight
Group
weights
Item
weights
J1950
J1950
Chronology of changes in the Consumer Price Index, 1890 to date— Continued
Base
period
Census
Number
providing
of areas
population
included
weights
’1947-49
Family
composition
Earnings
of
chief earner
46
Source and
amount of
family income
Length
of
employment
Economic level,
length of
residence,
nativity, and
race
Title(s)
No specific re Short title.
quirement, but Consumer
major portion
Price Index
of income of
Complete
family head
name:
must be from
Index of
employment
Change in
as wage earn
Prices of
er or salaried
Goods and
Services Pur
clerical
worker.
chased by City
Wage-Earner
and ClericalWorker Fami
lies to
Maintain Their
Level of Living.
’1957-59
171960-61
1960-61
1960
50
Families of 2 or No limitation.
more persons
and single
workers; at
least 1 full
time wage
earner.
More than half A minimum of
37 weeks for
of combined
at least 1
family income
family
from wagemember.
earner or
clerical-worker
occupation.
Same as above
for earner and
clericalworker index.
No limitation
for urban
consumer
index.
Same as above Same as above Same as above
for wagefor wagefor wageearner and
earner and
earner and
clerical-worker clericalclericalworker index.
worker index.
index. No
No limitation
limitation for
No employ
for urbanurbanment required
consumer
for urbanconsumer
consumer
index.22
index.
index.
No restriction
Consumer Price
on other than
Index for
the wageUrban Wage
earner and
Earners and
clerical-worker Clerical
definition.
Workers.
56
1967
’1972-73 '1972-73
1970
85
1) Consumer
Price Index
for Urban
Wage Earners
and Clerical
Workers
(C P I- W ) .
2) Consumer
Price Index for
All Urban
Consumers
(C P I-U ).
Jan. ... 241982-84 ’ 1982-84
198723
Jan. ...
1988
251982-84
1980
Similar to above
except that
students re
siding in
coliegeregulated
housing are
treated as
separate
family units.
1 Food price index only
2 For 19 cities, data were available back to December 1914 and for 13 cities, back
to 1917. For the United States, data were available back to the 1913 annual average.
3 Indexes between 1918-29 were recomputed retroactively with population weights
based on the average of the 1920 and 1930 censuses.
4 Index published in December 1935 for July 15,1935; indexes were also calculated
on the 1913=100 base.
5 Indexes between 1925-29 were recomputed retroactively with group weights
based on the average of 1917-19 and 1934-36, indexes between March 15,1930, and
March 15, 1940, were recomputed retroactively using 1934-36 group weights.
6 During World War II, weights were adjusted to account for rationing and shortages.
7 51-56 cities included in the food index.
8 Index published in May 1941 for March 14,1941. Food indexes were based on 51
cities.
g
1940 census data were supplemented by ration book registration data.
10 Index published in March 1951 for January 1951.
11 Indexes between January 1950 and January 1951 were revised retroactively for
all items and group indexes. Indexes for rent and all items were corrected for the new
unit bias from 1940. Old series also published through 1952.
12 Item weights were revised for only the 7 cities for which 1947-49 expenditure
data were available. Index published in February for January 1953. Linked to old
series as of December 1952. Old series also published for a 6-month overlap period.
13 Data were adjusted to 1952 for weight derivation.
14 Indexes were also calculated on the base of 1935-39=100 through December 1957.
15 Index published in February for January 1962. Indexes were also calculated on bases of
1947-49=100 and 1939=100.
18
Index published March 3 for January 1964. Linked to old series as of December 1963.
Old series also published for a 6-month overlap period.
17 Data were adjusted to December 1963 for weight derivation.
18 Index published in February for January 1966. Linked to old series as of December 1965.
19 Index published in February for January 1971. Indexes were also calculated on the 195759=100 base.
20 Index published in February for January 1978. Linked to old series as of December 1977.
Old series also published for a 6-month overlap period.
21 Data were adjusted to December 1977 for weight deriviation.
22 Coverage was expanded to include wage earners and clerical workers in the entire non
farm parts of the metropolitan areas in addition to those living within the urbanized areas of
the metropolitan areas and urban places of 2,500 or more inhabitants.
23 Index published in February for January 1987, linked to old series as of December 1986.
Old series also published for a 6-month overlap period.
24 Data were adjusted to December 1986 for weight derivation.
23 Index published in February for January 1988. Indexes also calculated on the 1967=100 base
186
CPI Appendix 2.
December 1986
Relative importance of all components in the Consumer Price Indexes: U.S. city average,
(Percent of all items)
Item and group
All Urban
Consumers
(CPI-U)
All items .................................................
100.000
Food and beverages..........................
Urban
Wage Earners
and
Clerical Workers
(CPI-W)
100 000
17.758
19.652
Food ...............................................
Food at home ............................
Cereals and bakery products .
Cereal and cereal products .
Flour and prepared flour
m ixes............................
Cereal ..............................
Rice, pasta, and cornmeal
Bakery products .................
White bread ...........
Fresh other bread, biscuits,
rolls, and muffins........
Cookies, fresh cakes, and
cupcakes .....................
Other bakery products .. .
16.190
9.952
1.354
.428
17.972
11.263
1.539
.494
.076
.251
.101
.926
221
.090
.287
.117
1.045
.272
.206
.215
.235
.265
.278
.282
Meats, poultry, fish, and eggs . .
3.135
? 948
2.085
1.011
3.668
3 448
2.504
1.210
.373
085
.053
086
.078
.335
654
.116
.155
.158
.463
098
.059
112
.097
.380
796
.137
192
.192
226
420
.490
.164
274
498
.546
.194
235
091
373
.081
256
098
898
.089
.292
.187
.309
.220
1 261
.625
.365
1 394
.729
.452
260
.636
.348
277
.665
.359
.166
.179
.122
.127
1.654
1.017
.516
.102
.059
1.748
1.070
.538
.111
.063
.072
.283
.501
.087
.059
.087
.278
.532
.097
.063
Meats
Ground beef other than
canned .....................
Round roast .................
Sirloin steak .................
Other beef and veal .. .
Pork
Bacon ..........................
Chops ..........................
Ham ..
Other pork, including
sausage ...................
Other meats .. .
Poultry..................................
Fresh whole chicken . . . .
Fresh and frozen chicken
parts ............................
Fish and seafood .. .
Canned fish and seafood .
Fresh and frozen fish
and seafood .................
Dairy products.........
Fresh milk and cream.........
Fresh whole m ilk .............
Other fresh milk and
Processed dairy products ..
Cheese ............................
Ice cream and related
products .......................
Other dairy products,
including butter...........
Fruits and vegetables.............
Fresh fruits and vegetables .
Fresh fruits .....................
Apples ..........................
Bananas .......................
Oranges, including
tangerines.................
Other fresh fruits .......
Fresh vegetables.............
Potatoes ......................
Lettuce ........................
Item and group
Urban
Wage Earners
and
Clerical Workers
(CPI-W)
0 084
.272
0 097
.275
.637
365
.678
.367
.283
.082
.273
.093
.289
.078
.311
.102
.179
.209
Other food at home ...............
Sugar and sweets...............
Sugar and artificial
sweeteners ...................
Sweets, including candy .
Fats and oils ......................
Nonalcoholic beverages___
Carbonated drinks...........
Coffee ..............................
Other noncarbonated drinks
Other prepared food ...........
Canned and packaged soup
Frozen prepared food . . . .
Snacks ............................
Seasonings, condiments,
and spices ...................
Miscellaneous prepared
food, including baby
food ..............................
2.548
.358
2.913
.402
.094
.265
.265
.889
.445
.290
.153
1.035
.083
.186
.216
.116
.286
.303
1.026
.546
.306
.174
1.182
.091
.206
.245
.269
.304
.281
.336
Food away from hom e...............
Lunch ......................................
Dinner ......................................
Other meals and snacks.......
Unpriced items ......................
6.238
2.188
2.683
1.044
.323
6.709
2.478
2.763
1.236
.232
Alcoholic beverages ..........................
Alcoholic beverages at hom e.......
Beer and a le ................................
Distilled spirits ..........................
Wine at home ............................
Alcoholic beverages away from home
1.568
.870
.448
.228
.194
.698
1.680
.953
.560
.216
.176
.727
Housing .................................................
Shelter.................................................
Renters’ costs ................................
Rent, residential ........................
Other renters’ costs ...................
Lodging while out of town .. .
Lodging while at school .......
Tenants’ insurance .................
Homeowners’ costs ......................
Owners’ equivalent re n t.............
Household insurance .................
Maintenance and repairs ...............
Maintenance and repair services
Maintenance and repair
commodities ..........................
Materials, supplies, and
equipment for home repairs
Other maintenance and repair
commodities .......................
Fuels and other utilities.................
Fuels ...........................................
Fuel oil and other household
fuel commodities ...............
Fuel oil ................................
Other household fuel
commodities ...................
42.791
27.691
7.957
6.058
1.899
1.670
.193
.036
19.512
19.100
.412
.222
.133
40.318
25.433
8.174
6.922
1.253
1.101
.122
.030
17.049
16.700
.348
.210
.112
.089
.099
.040
.044
.049
7.908
4.456
.055
8.123
4.563
.394
.265
.360
.227
.128
.133
Other fresh vegetables .
Processed fruits and
vegetables ......................
Frozen juices and
frozen fruit ...............
Canned and dried fruits
Processed vegetables . . . .
Other processed
vegetables ...............
See fo o tn o te s at end o f table.
All Urban
Consumers
(CPI-U)
187
CPI Appendix 2. Relative importance of all components in the Consumer Price Indexes: U.S. city average,
December 1986— Continued
(Percent of all items)
Item and group
Gas (piped) and electricity___
Electricity ............................
Utility (piped) gas ...............
Other utilities and public services
Telephone services.................
Local charges .....................
Interstate toll calls .............
Intrastate toll calls .............
Water and sewerage maintenance
Cable television..........................
Refuse collection .......................
All Urban
Consumers
(CPI-U)
4.062
2.742
1.320
3.452
2.210
1.335
Urban
Wage Earners
and
Clerical Workers
(CPI-W)
Apparel commodities less
footwear ..................................
Men’s and boys’ .....................
Men’s ..................................
Suits, sport coats, coats,
and jackets...................
Furnishings and special
clothing.....................
Shirts................................
Dungarees, jeans, and
trousers.........................
Unpriced items ...............
Boys’ ....................................
Women’s and girls’ .................
Women’s ..............................
Coats and jackets...........
Dresses ............................
Separates and sportswear
Underwear, nightwear,
hosiery, and
accessories .................
Suits ................................
Unpriced items ...............
Girls’ ....................................
Infants’ and toddlers’ .............
Other apparel commodities . . .
Sewing materials, notions,
and luggage .....................
Watches and jewelry .........
Watches ..........................
Jewelry ............................
Footwear ......................................
Men’s ......................................
Boys’ and girls’ .......................
Women’s ..................................
Apparel services ............................
Laundry and dry cleaning other
than coin operated .................
Other apparel services...............
4.202
2.811
1.391
3.560
2.230
.500
.374
1.353
.497
.379
.661
.424
.157
.688
.487
.156
Household furnishings and operation
Housefurnishings ..........................
Textile housefurnishings ...........
Furniture and bedding ...............
Bedroom furniture...................
Sofas ........................................
Living room chairs and tables
Other furniture.........................
Appliances, including electronic
equipment ..............................
Television and sound
equipment ...........................
Television ............................
Other video equipment........
Sound equipment ...............
Unpriced items ...................
Major household appliances ..
Refrigerators and home
freezers ............................
Laundry equipment.............
Stoves, ovens, dishwashers,
and air-conditioners .......
Information processing
equipment ..........................
Other housefurnishings .............
Floor and window coverings,
infants’, laundry, cleaning,
and outdoor equipment___
Clocks, lamps, and decor items
Tableware, serving pieces, and
nonelectric kitchenware___
Lawn equipment, power tools,
and other hardware.............
Sewing, floor cleaning, small
kitchen, and portable heating
appliances ..........................
Indoor plants and fresh cut
flowers ................................
Unpriced items .......................
Housekeeping supplies .............
Laundry and cleaning products,
including soap.....................
Household paper products and
stationery supplies .............
Other household, lawn, and
garden supplies...................
Housekeeping services .............
Postage....................................
Appliance and furniture repair
Gardening and other household
services................................
Babysitting ..............................
Domestic services...................
Care of invalids, elderly, and
convalescents .....................
Unpriced items .......................
7.193
4.432
6.762
4.344
.397
Apparel and upkeep ..........................
Apparel commodities .....................
.425
1.301
.427
.258
Item and group
1.270
.472
.253
.210
.405
.201
.345
1.310
1.361
.720
.785
.254
.177
.290
.289
.182
.314
.000
.000
.393
.430
.114
.124
.123
.142
.157
.165
.196
.146
1.396
1.316
.193
.266
.161
.216
.240
.214
.238
.289
.186
.195
.174
.099
1.212
.152
.089
1.244
.422
.485
.395
.389
.396
1.549
.255
.370
1.174
.188
.147
.394
.270
.266
.225
.347
.076
.056
.121
.038
.094
6.309
5.743
6.333
5.805
Transportation ....................................
Private .............................................
Mew vehicles ..............................
New cars..................................
New trucks ..............................
New motorcycles ...................
Used cars ....................................
Motor fu e l....................................
Automobile maintenance and
repair........................................
Body work ..............................
Automobile drive train, brake,
miscellaneous mechanical
repair ....................................
Maintenance and servicing . . .
Power plant repair...................
Unpriced items .......................
Other private transportation . . . .
Other private transportation
commodities .......................
Motor oil, coolant, and other
products ..........................
Automobile parts and
equipment .......................
Tires ................................
Other parts and equipment
Other private transportation
services................................
Automobile insurance ........
Automobile finance charges
Automobile fees .................
.247
See fo o tn o te s at end o f table.
188
All Urban
Consumers
(CPI-U)
Urban
Wage Earners
and
Clerical Workers
(CPI-W)
4.898
1.543
1.243
4 875
1 540
1.170
.362
.278
.306
.310
.308
.249
.017
.271
.021
.300
2.600
2.219
.207
.370
2.550
.370
1.071
.374
.164
291
2.113
.170
.369
1.056
.033
.381
.376
.112
.030
.438
.236
.520
.303
.482
.098
.421
.093
.087
.395
.088
.328
.307
.845
.929
.270
.173
.402
.565
.315
.223
.391
.528
.294
.271
.225
.303
17.172
15.684
5.591
4.537
.971
.083
1.259
2.897
19.018
17.874
5.434
4.035
1.250
.149
2.253
3.552
1.543
.158
1.611
.163
.442
.480
.520
.399
.023
4.396
.449
.021
5.026
.769
.972
.066
.085
.703
.347
.356
.888
.412
.475
3.627
2.135
.775
.716
4.053
2.413
.969
.671
.497
CPI Appendix 2. Relative importance of all components in the Consumer Price Indexes: U.S. city average,
December 1986— Continued
(Percent of all Items)
Item and group
Automobile registration,
licensing, and
inspection fees ...........
Other automobile-related
fees ..............................
Unpriced items ...............
Public transportation ........................
Airline fares ....................................
Other intercity transportation . . .
Intracity public transportation ..
Unpriced items ..........................
Medical ca re .......................................
Medical care commodities ...........
Prescription drugs1 .....................
Nonprescription drugs and
medical supplies.....................
Internal and respiratory over-thecounter drugs .....................
Nonprescription medical
equipment and supplies . . .
Professional medical services ..
Physicians’ services2 .............
Services by other medical
professionals5 .....................
Hospital and related services
Hospital rooms6 .................
Other inpatient hospital
services7 ..........................
Outpatient services8 ...........
Unpriced items ...................
Health insurance9 ...................
Entertainment ....................................
Entertainment commodities .........
Reading materials .......................
Newspapers ............................
Magazines, periodicals, and
books
..............................
Unpriced items .......................
Sporting goods and equipment .
Sport vehicles, including
bicycles................................
Other sporting goods.............
Toys, hobbies, and other
Toys, hobbies, and music
equipment ..........................
Photographic supplies and
equipment ..........................
Pet supplies and expense___
Entertainment services.................
Club memberships .....................
Fees for participant sports,
excluding club memberships .
Admissions ................................
Fees for lessons or instructions
Other entertainment services .. .
Unpriced items ...................
All Urban
Consumers
(CPI-U)
Urban
Wage Earners
and
Clerical Workers
(CPI-W)
0.337
0.367
354
.025
1.488
.948
.163
.365
.012
.288
.017
1.144
.595
.112
.427
.010
5.749
1.086
.702
4.868
.893
.561
.384
.332
.248
.249
.136
4.663
2.926
1.554
.866
343
.083
3.975
2.476
1.319
.773
.286
.164
1.516
.608
.099
1.335
.580
.567
.337
.004
.221
.487
.265
.003
.163
4.385
2.116
.692
.330
4.067
2.210
.603
.304
.362
.000
.478
.299
.000
.587
.221
.258
.343
.244
945
1.020
.437
.479
.132
.365
011
2.270
.386
.117
.408
015
1.857
.209
.339
.302
.545
.152
.637
.013
.636
.214
.674
.021
Item and group
including hair and dental
products ..................................
Cosmetics, bath and nail
preparations, manicure and eye
makeup implements ...............
Personal care services ...................
Beauty parlor services for females
Haircuts and other barber shop
services for males...................
Unpriced items ..........................
Personal and educational expenses .
School books and supplies...........
School books and supplies for
colleges ..................................
Elementary and high school books
and supplies ..........................
Unpriced items ..........................
Personal and educational services
Tuition and other school fees . . .
College tuition ........................
Elementary and high school
tuition ..................................
Day care and nursery school .
Tuition for technical, business,
and other schools...........
Unpriced items ......................
Personal expenses .....................
Legal service fe e s ...................
Funeral expenses ...................
Personal financial services .. .
Unpriced items .......................
5.836
1.246
1.231
5.743
1.640
1 184
.661
.678
All item s.................................................
Commodities .....................................
Food and beverages......................
Commodities less food and
Nondurables less food and
beverages ............................
Apparel commodities .........
Nondurables less food,
beverages, and apparel ..
Durables ..................................
Services .............................................
Rent of shelter................................
Household services less rent of
shelter.........................................
Transportation services .................
Medical care services.....................
0.381
0.415
.280
.570
.453
.263
.506
.399
.116
.000
3.359
.217
.107
.000
2.919
.187
.150
.120
.055
.011
3.142
1.971
1.107
.056
.012
2.732
1.674
.820
.345
.319
.276
.388
.121
.080
1.171
.432
.305
.341
.093
.121
.069
1.058
.375
.251
.343
.089
100.000
45.468
17.758
100.000
49.323
19.652
27.710
29.671
15.584
5.743
16.464
5.805
9.840
12.126
54.532
27.020
6.058
10.660
13.207
50.677
24.845
6.922
9.644
6.658
4.663
6.547
9.426
6.808
3.975
5.624
83.810
72.309
80.488
94.251
29.278
17.152
11.408
33.342
27.512
82.028
74.567
82.951
95.132
31.351
18.144
12.340
36.116
25.832
Special indexes
All items less medical ca re ...................
Commodities less food ........................
Nondurables less foo d ..........................
Nondurables less food and apparel__
Nondurables...........................................
Services less rent of shelter.................
See fo o tn o te s at end o f table.
Urban
Wage Earners
and
Clerical Workers
(CPI-W)
Commodity and service group
All items less shelter............................
Other goods and services.....................
Tobacco and smoking products.......
Personal c a re ......................................
Toilet goods and personal care
appliances ..................................
Other toilet goods and small
personal care applianced,
All Urban
Consumers
(CPI-U)
189
CPI Appendix 2. Relative importance of all components in the Consumer Price Indexes: U.S. city average,
December 1986— Continued
(Percent of all Items)
Urban
Wage Earners
and
Clerical Workers
(CPI-W)
Item and group
All Urban
Consumers
(CPI-U)
Services less medical care ...................
49.869
46.703
Domestically produced farm fo o d .......
Selected beef cuts ................................
Motor fuel, motor oil, coolant, and other
products .............................................
Utilities and public transportation.......
8.632
.694
9.776
.851
2.963
9.002
3.636
8.906
1 Benefits provided by consumer-paid health insurance constitute 3.9 percent o f the
relative importance for the U-population and 6.3 percent for the W population.
2 Benefits provided by consumer-paid health insurance constitute 30.8 percent o f the
relative importance for the U-population and 30.8 percent for the W population.
3 Benefits provided by consumer-paid health insurance constitute 8.7 percent o f the
relative importance for the U-population and 8.7 percent for the W population.
4 Benefits provided by consumer-paid health insurance constitute 0.7 percent o f the
relative importance for the U-population and 0.7 percent for the W population.
3 Benefits provided by consumer-paid health insurance constitute 24.7 percent o f the
relative importance for the U-population and 29.3 percent for the W population.
6 Benefits provided by consumer-paid health insurance constitute 61.0 percent o f the
relative importance for the U-population and 54.4 percent for the W population.
Item and group
All Urban
Consumers
(CPI-U)
Urban
Wage Earners
and
Clerical Workers
(CPI-W)
Housekeeping and home maintenance
services...............................................
1.682
1.286
Energy.....................................................
All Items less energy............................
All Items less food and energy.........
Commodities less food and energy
Energy commodities...................
Services less energy.......................
7.352
92.648
76.458
25.988
3.290
50.471
8.114
91.886
73.914
27.439
3.912
46.475
7 Benefits provided by consumer-paid health insurance constitute 62.6 percent o f the
relative importance for the U-population and 62.8 percent for the W population.
8 Benefits provided by consumer-paid health insurance constitute 37.0 percent o f the
relative importance for the U-population and 53.4 percent for the W population.
9 Only health insurance premiums paid by the consumer are included in the CPI.
The health insurance relative importance includes only that portion o f the premium
that is retained by the insurance carrier for administrative cost and profit, 9.7 percent
o f the total premiums for the U population and 10.6 percent for the W population.
The portions o f the premium that are paid as benefits have been assigned to the rele
vant medical care categories.
NOTE: Dash indicates that data are not available.
190
CPI Appendix 3.
PSU
Sample areas, population weights, and pricing cycles
Sample areas and counties
Percent
ot Index
population
Pricing
cycle
Northeast Region
A109
A110
A111
A102
A103
A104
A105
L102
New York— Northern New
Jersey— Long Island,
NY-NJ-CT, CMSA:
New York City ..................
Bronx, Kings, New
York, Queens,
Richmond
New York—Connecticut
suburbs ..............................
New York portion:
Wassau, Orange,
Putnam, Rockland,
Suffolk, Westchester
Conneticut portion:
Fairfield, Litchfield
(part), New Haven
(part)
New Jersey suburbs ........
Bergen, Essex,
Hudson, Hunterdon,
Middlesex, Monmouth,
Morris, Ocean,
Passaic, Somerset,
Sussex, Union
Philadelphia—
Wilmington—T renton,
PA-DE-NJ-MD, CMSA . . . .
Pennsylvania portion:
Bucks, Chester,
Delaware,
Montgomery,
Philadelphia
New Jersey portion:
Burlington, Camden,
Cumberland,
Gloucester, Mercer,
Salem
Delaware portion:
New Castle
Maryland portion:
Cecil
Boston— Lawrence—Salem,
MA-NH, C M S A .....................
Massachusetts portion:
Bristol (part), Essex,
Middlesex (part),
Norfolk (part),
Plymouth (part),
Suffolk, Worcester
(part)
New Hampshire portion:
Hillsborough (part),
Rockingham
4.115
2.375
2.762
2.920
X
X
X
X
L104
Syracuse, NY, M SA ___
Madison, Onondaga
L106
Springfield, MA, M S A .........
Hampden (part),
Hampshire (part)
L108
Scranton—Wilkes-Barre,
PA, MSA .................................
Columbia, Lackawanna,
Luzerne, Wyoming
X
X
X
Sample areas and counties
0.767
Pricing
cycle
Odd
Even
months months
X
.847
X
.974
X
M102
Williamsport, PA, M S A ___
Lycoming
.824
X
M104
Lancaster, PA,
Lancaster
..............
.746
X
M106
Johnstown, PA, M S A ..........
Cambria, Somerset
.756
X
M108
Poughkeepsie, NY, MSA . . .
Dutchess
.771
X
R102
St. Lawrence Co, NY
Urban parts of:
St. Lawrence
.545
R104
Augusta, ME ..........
Urban parts of:
Kennebec, Lincoln
.535
m s a
X
X
X
Midwest Region
A207
A208
Chicago—Gary—Lake
County, IL-IN-WI, CMSA . .
Illinois portion:
Cook, Du Page,
Grundy, Kane, Kendall,
Lake, Mchenry, Will
Indiana portion:
Lake, Porter
Wisconsin portion:
Kenosha
2.141
X
2.363
X
X
A210
X
A211
X
191
X
X
Lapeer, Livingston,
Macomb, Oakland,
St. Clair, Washtenaw,
Wayne
A209
.991
4.039
Detroit—Ann Arbor, Ml,
CMSA ..........................................
Pittsburgh— Beaver Valley,
1.276
PA, C M S A ..........................
Allegheny, Beaver,
Fayette, Washington,
Westmoreland
Buffalo— Niagara Falls, NY,
CMSA ........................................... 653
Erie, Niagara,
New York
Hartford— New Britain—
Middletown, CT, CMSA . .
Hartford (part),
Litchfield (part),
Middlesex (part),
New London (part),
Tolland (part)
PSU
Odd
Even
months months
Percent
of index
population
St. Louis— East St. Louis,
MO-IL, CMSA ..........................
Missouri portion:
Franklin, Jefferson, St.
Charles, St. Louis,
St. Louis City
Illinois portion:
Clinton, Jersey,
Madison, Monroe,
St. Clair
Cleveland—Akron—Lorain,
OH, C M S A .................................
Cuyahoga, Geauga,
Lake, Lorain,
Medina, Portage,
Summit
Minneapolis—St. Paul,
MN-WI, MSA ............................
Minnesota portion:
Anoka, Carver, Chisago,
Dakota, Hennepin,
Isanti, Ramsey, Scott,
Washington, Wright
Wisconsin portion:
St. Croix
1.201
1.478
X
1.155
X
CPI Appendix 3.
PSU
Sample areas, population weights, and pricing cycles— Continued
Sample areas and counties
A212
Milwaukee, W l, p m s a ............
Milwaukee, Ozaukee,
Washington, Waukesha
A213
Cincinnati— Hamilton,
OH-KY-IN, C M S A ...................
Ohio portion:
Butler, Clermont,
Hamilton, Warren
Kentucky portion:
Boone, Campbell,
Kenton
Indiana portion:
Dearborn
Percent
of index
population
Pricing
cycle
0 .740
.855
PSU
Odd
Even
months months
X
Sample areas and counties
Percent
of index
population
R208
Kennett, MO ..............
Urban parts of:
Dunklin, Pemiscot
R210
Mexico, M O ...................................801
Urban parts of:
Audrain, Lincoln,
Pike, Ralls
R212
Ft. Dodge, I A ................................ 761
Urban parts of:
Calhoun, Hamilton,
Webster
A315
Washington, DC-MD-VA, MSA
District of Columbia
portion:
Washington
Maryland portion:
Calvert, Charles,
Frederick,
Montgomery,
Prince Georges
Virginia portion:
Arlington, Fairfax,
Loudoun, Prince
William, Stafford,
Alexandria City,
Fairfax City, Falls
Church City, Manassas
City, Manassas Park
City
A316
Dallas— Fort Worth, TX,
X
Pricing
cycle
Odd
Even
months months
0.739
X
X
X
South Region
A214
Kansas City, MO— Kansas
City, KS, C M S A .....................
Missouri portion:
Cass, Clay, Jackson,
Lafayette, Platte, Ray
Kansas portion:
Johnson, Leavenworth
Miami, Wyandotte
.754
X
A215
Columbus, OH, MSA .........
Delaware, Fairfield,
Franklin, Licking,
Madison, Pickaway,
Union
.677
X
L210
Flint, Ml, MSA .
Genesee
.709
X
L212
Dayton—Springfield, OH,
.871
X
MSA
........................................
Clark, Greene, Miami,
Montgomery
L214
CMSA .
........................................
.769
X
.840
X
Mahoning, Trumbull
L216
Indianapolis, IN, MSA ___
Boone, Hamilton,
Hancock, Hendricks,
Johnson, Marion,
Morgan, Shelby
M210
Steubenville—Weirton,
OH-WV, M S A .....................
Ohio portion:
Jefferson
West Virginia portion:
Brooke, Hancock
.794
M212
Racine, Wl,
Racine
..............
.792
M214
Waterloo—Cedar Falls,
IA, MSA .................................
Black Hawk, Bremer
.839
pm sa
X
X
A317
Baltimore, MD, M S A ..........
Anne Arundel,
Baltimore, Carroll,
Harford, Howard,
Queen Annes,
Baltimore City
A318
Houston—Galveston—
Brazoria, TX, C M S A ..........
Brazoria, Fort Bend,
Galveston, Harris,
Liberty, Montgomery,
Waller
A319
Atlanta, GA, M S A ................
Barrow, Butts,
Cherokee, Clayton,
Cobb, Coweta,
De Kalb, Douglas,
Fayette, Forsyth,
Fulton, Gwinnett,
Henry, Newton,
Paulding, Rockdale,
Spalding, Walton
A320
Miami— Fort Lauderdale, FL,
CMSA ...................................
Broward, Dade
X
M216
Lawrence, KS, M S A ..........
Douglas
1.004
M218
Terre Haute, IN, M S A ___
Clay, Vigo
.838
M220
Elkhart—Goshen, IN, MSA
Elkhart
.809
X
R206
Grand Island, NE . . .
Urban parts of:
Hall, Hamilton,
Howard, Merrick
779
X
X
X
1.556
1.124
X
1.621
X
1.118
X
X
A321
X
Collin, Dallas, Denton,
Ellis, Johnson,
Kaufman, Parker,
Rockwall, Tarrant
Youngstown—Warren, OH,
MSA
.................................
1.766
192
Tampa—St. Petersburg—
Clearwater, FL, MSA . . .
Hernando,
Willsborough, Pasco,
Pinellas
1.526
.953
X
X
CPI Appendix 3.
PSU
Sample areas, population weights, and pricing cycles—Continued
Sample areas and counties
Percent
of Index
popul atlon
New Orleans, LA, m s a . .
Jefferson, Orleans,
St. Bernard, St.
Charles, St. John The
Baptist, St. Tammany
0.639
Richmond, VA, M S A .........
Charles City,
Chesterfield,
Goochland, Hanover,
Henrico, New Kent,
Powhatan, Richmond
City
.792
L320
Jacksonville, FL, MSA . . .
Clay, Duval, Nassau,
St. Johns
.812
L322
Charlotte—Gastonia— Rock
Hill, WC-SC, M S A ..............
North Carolina portion:
Cabarrus, Gaston,
Lincoln, Mecklenburg,
Rowan, Union
South Carolina portion:
York
A322
L318
.800
Pricing
cycle
PSU
Even
Odd
months months
Sample areas and counties
Percent
of Index
population
M330
Albany, GA, m s a ............
Dougherty, Lee
M332
Florence, SC,
Florence
___
.826
M334
Gainesville, FL, m s a . . .
Alachua, Bradford
.785
M336
Huntsville, AL, MSA . . .
Madison
.788
M338
Beaumont—Port Arthur,
TX, MSA ..........................
Hardin, Jefferson,
Orange
m s a
0.793
.778
M340
Ocala, FL, m s a ..............
Marion
.897
R314
Cleveland, TN ..............
Urban parts of:
Bradley, Polk
.654
R316
Hammond, L A ..............
Urban parts of:
East Feliciana, St.
Helena, Tangipahoa
.693
L324
Tulsa, OK, M S A .....................
Creek, Osage, Rogers,
Tulsa, Wagoner
.836
L326
Raleigh—Durham, NC,
Durham, Franklin,
Orange, Wake
.897
R318
Raeford, N C ............
Urban parts of:
Woke, Scotland
.680
L328
Norfolk—Virginia Beach—
Newport News, VA, m s a
Gloucester, James
City, York, Chesapeake
City, Hampton City,
Newport News City,
Norfolk City, Poquoson
City, Portsmouth City,
Suffolk City, Virginia
Beach City,
Williamsburg City
.761
R320
Pontotoc, M S ............ .
Urban parts of:
Benton, Pontotoc,
Tippah, Union
.688
R322
Halifax, NC ..........
Urban parts of:
Halifax
.627
R324
Central KY ..........................
Urban parts of:
Breathitt, Estill,
Garrard, Jackson, Lee,
Madison, Montgomery,
Owsley, Powell,
Rockcastle
.631
msa
L330
Washville, TN, m s a ...........
Cheatham, Davidson,
Dickson, Robertson,
Rutherford, Sumner,
Williamson, Wilson
.909
L332
El Paso, TX,
El Paso
___
.664
L334
Birmingham, AL, m s a
Blount, Jefferson,
St. Clair, Shelby,
Walker
.747
L336
Orlando, FL, m s a ___
Orange, Osceola,
Seminole
.720
M322
Corpus Christi, TX, m s a .
Nueces, San Patricio
.817
M324
Pine Bluff, AR, M S A ___
Jefferson
.774
M326
Fort Smith, AR-OK, MSA
Arkansas portion:
Crawford, Sebastian
.811
M328
m s a
Brownsville—Harlingen,
TX, MSA ............................
Cameron
West Region
A419
A420
.612
A422
Los Angeles—Anaheim—
Riverside, CA, CMSA:
Los Angeles City ........
Los Angeles
Greater Los Angeles ..
Orange, Riverside,
San Bernardino,
Ventura
San Francisco—Oakland—
San Jose, CA, C M S A ___
Alameda, Contra Costa,
Marin, Napa, San
Francisco, San Mateo,
Santa Clara, Santa
Cruz, Solano, Sonoma
A423
Seattle—Tacoma, WA, cmsa
King, Pierce, Snohomish
A424
San Diego, CA,
San Diego
193
m sa
4.189
2.102
3.156
1.193
.987
Pricing
cycle
Even
Odd
months months
CPI Appendix 3.
PS U
A425
Sample areas, population weights, and pricing cycles— Continued
Sample areas and counties
Portland—Vancouver,
OR-WA, c m s a .....................
Oregon portion:
Clackamas, Multnomah,
Washington, Yamhill
Washington portion:
Clark
Percent
of index
population
.320
Anchorage, AK, MSA .........
Anchorage Borough
.086
Phoenix, AZ, M S A ................
Maricopa
.816
A433
Denver— Boulder, CO, c m s a
Adams, Arapahoe,
Boulder, Denver,
Douglas, Jefferson
.929
L438
Sacramento, CA, M S A ___
El Dorado, Placer,
Sacramento, Yolo
.814
A427
A429
PS U
Odd
Even
months months
L440
0.744
Honolulu, HI, MSA ..............
Honolulu
A426
Pricing
cycle
X
Sample areas and counties
Salt Lake City—Ogden,
UT MSA ...................................
Davis, Salt Lake, Weber
Percent
of index
population
Pricing
cycle
Even
Odd
months months
X
0.619
L442
Tucson, AZ, M S A ...................
Pima
.521
L444
Fresno, CA,
Fresno
...................
.513
M442
Redding, CA, M S A ................
Shasta
.642
X
M444
Colorado Springs, CO,
El Paso
msa
.581
X
M446
Yakima, WA,
Yakima
................
.654
X
M448
Provo—Orem, UT,
...
.647
X
R426
Alamogordo, N M ................
Urban parts of: Otero
.892
X
R428
Yuma, AZ ............................
893
X
m s a
X
X
X
X
X
X
194
msa
msa
X
CPI Appendix 4.
EC
Expenditure classes, item strata, and entry level items
01 Cereals and cereal products
0101 Flour and prepared flour m ixes
01011 Flour
05011 Frankfurters
05012 B ologn a, liverwurst, salami
14 Processed vegetables
1401 Frozen vegetables
14011 Frozen vegetables
01012 Prepared flour m ixes
05013 Other lunchm eats (excluding bologn a,
liverwurst, salam i)
1402 C anned and other processed vegetables
14021 C anned beans other than lim a beans
0102 Cereal
01021 Cereal
05014 Lam b, organ m eats, and gam e
14022 C anned cut c o m
06 Poultry
0601 Fresh w hole chicken
06011 Fresh w hole chicken
14023 Other processed vegetables
0103 R ice, pasta, and cornm eal
01031 Rice
EC
01032 M acaroni, similar products, and
cornm eal
EC
EC
0602 Fresh or frozen chicken parts
06021 Fresh or frozen chicken parts
02 Bakery products
0201 W hite bread
02011 W hite bread
0202 Other breads, rolls, biscuits, and m uffins
02021 Bread other than w hite
EC
15012 Other sweets (excluding candy and
gum )
0603 Other poultry
06031 Other poultry
EC
02022 R olls, biscuits, m u ffin s (excluding
frozen)
07 Fish and seafood
0701 C anned fish and seafood
07011 Canned fish or seafood
1502 Sugar and artificial sweeteners
15021 Sugar and artificial sweeteners
EC
0702 Fresh or frozen fish and seafood
07021 S hellfish (excluding canned)
0204 C akes, cupcakes, and cook ies
02041 C akes and cupcakes (excluding frozen)
15 Sugar and sweets
1501 C andy and other sweets
15011 C andy and chew ing gum
16 Fats and oils
1601 Fats and oils
16011 Margarine
16012 Other fats and oils
07022 Fish (excluding canned)
16013 N ondairy cream substitutes
02042 C ookies
EC
0206 Other bakery products
02061 Crackers
02062 Bread and cracker products
02063 Sw eetrolls, co ffee cake, and doughnuts
(excluding frozen)
EC
03 B eef and veal
0301 G round b eef
03011 Ground b eef
0302 Chuck roast
03021 Chuck roast
EC
17032 Instant and freeze dried co ffee
10012 Other dairy products
17052 Tea
1002 Cheese
10021 Cheese
17053 Other noncarbonated drinks
18022 Frozen prepared fo o d s other than
m eals
1803 Snacks
18031 P o ta to chips and other snacks
0305 R ound steak
03051 R ound steak
1104 Other fresh fruits
11041 Other fresh fruits
18032 N uts
EC
18042 O lives, pickles, relishes
1202 Lettuce
12021 L ettuce
0402 Pork chops
04021 Pork chops
1203 T om atoes
12031 T om atoes
0403 H am
04031 H am (excluding canned)
1204 Other fresh vegetables
12041 Other fresh vegetables
05 Other meats
0501 Lunchm eat, lam b, organ m eats, and
gam e
EC
1804 Spices, seasonings, con d im en ts, sauces
18041 Salt and other seasonings and spices
12 Fresh vegetables
1201 P otatoes
12011 P otatoes
04 Pork
0401 Bacon
04011 Bacon
04042 Pork sausage
18 Other prepared fo o d s
1801 Canned and packaged soup
18011 Canned and packaged soup
1802 Frozen prepared fo o d s
18021 Frozen prepared m eals
1103 Oranges
11031 Oranges
0404 Other pork, including sausage
04041 Pork roasts, picnics, other pork
1705 Other noncarbonated drinks
17051 N oncarbonated fruit-flavored drinks
11 Fresh fruits
1101 A pples
11011 A pples
03043 Other b eef
04032 Canned ham
EC
10 Processed dairy products
1001 Butter and other dairy products
(excluding cheese, ice cream)
10011 Butter
1102 Bananas
11021 Bananas
03042 Other steak (excluding round and
sirloin)
0306 Sirloin steak
03061 Sirloin steak
1703 C o ffee
17031 R oasted co ffee
EC
0304 Other steak, roast, and other b eef
03041 Other roasts (excluding chuck and
round)
17 N o n a lco h o lic beverages
1701 Carbonated drinks
17011 C ola drinks
17012 C arbonated drinks other than cola
1004 Ice cream and related products
10041 Ice cream and related products
0303 R ound roast
03031 R ound roast
EC
09 Fresh m ilk and cream
0901 Fresh w hole milk
09011 Fresh w hole milk
0902 Other fresh m ilk and cream
09021 Other fresh milk and cream
02065 Pies, tarts, turnovers (excluding frozen)
EC
16014 Peanut butter
EC
EC
02064 Frozen bakery products and
frozen/refrigerated doughs and batters
08 Eggs
0801 Eggs
08011 Eggs
18043 Sauces and gravies
18044 Other condim ents (excluding olives,
pickles, relishes)
1806 Other prepared fo o d
18061 Canned or packaged salads and
desserts
18062 Baby fo o d
13 Processed fruits
1301 Fruit juices and frozen fruits
13011 Frozen orange juice
18063 Other canned or packaged prepared
fo o d s
13013 Fresh, ca n n ed /o r bottled fruit juices
19 F o o d aw ay from hom e
1901 Lunch
19011 Lunch
1303 C anned and dried fruits
13031 C anned and dried fruits
1902 Dinner
19021 Dinner
13012 Other frozen fruits and fruit juices
195
EC
CPI Appendix 4.
Expenditure classes, item strata, and entry level items— Continued
1903 Other m eals and snacks
19031 Snacks and n onalcoh olic beverages
19032 Breakfast or brunch
EC
25 Fuel oil and other fuels
2501 Fuel oil
25011 Fuel oil
30031 Stoves and ovens (excluding m icrow ave
ovens)
30032 M icrow ave ovens
EC
1909 U npriced board and catered affairs
19090 U npriced item s
2502 Other fuels
25021 B ottled or tank gas
30033 P ortable dishwashers
20 A lcoh olic beverages
2001 Beer, ale, and other alcoholic m alt
beverages at hom e
20011 Beer, ale, and other alcoholic m alt
beverages at hom e
25022 C oal
30034 W indow air-conditioners
25023 Other fuels
EC
2002 D istilled spirits at hom e
20021 W hiskey at hom e
20022 D istilled spirits at h om e (excluding
whiskey)
EC
2003 W ine at hom e
20031 W ine at hom e
2005 A lco h o lic beverages aw ay from hom e
20051 Beer, ale, and other alcoholic m alt
beverages aw ay from hom e
20052 W ine aw ay from hom e
EC
3102 V id eo cassette recorders, disc players,
and tapes
31021 V id eo cassette recorders, disc players,
cam eras, and accessories
27 Other utilities and public services
2701 T elephone services, local charges
27011 T elephone services, local charges
31022 V id eo cassettes and discs, blank and
prerecorded
2702 W ater and sewerage m aintenance
27021 R esidential water and sewer service
31023 V ideo gam e hardware, softw are and
accessories
2703 C om m unity antenna and cable television
27031 C om m unity antenna or cable TV
3103 A u d io com p on en ts, rad ios, recordings,
and other audio equipm ent
31031 R adios, phon ograp h s, and tape
recorders/players
2704 Garbage and trash collection
27041 Garbage and trash collection
21 Pure rent-renter occupied
2101 Rent o f dw elling
21011 Rent o f dwelling
2705 Interstate telephone services
27051 Interstate telephone services
2103 L odging w hile at sch ool
21031 H ou sin g at sch ool, excluding board
31032 C om pon en ts and other sound
equipm ent
31033 R ecords and tap es, prerecorded and
blank
2706 Intrastate telephone services
27061 Intrastate telephone services
EC
31 T elevision and soun d equipm ent
3101 T elevision sets
31011 T elevision sets
2602 U tility natural gas service
26021 U tility natural gas service
20053 D istilled spirits aw ay from h om e
2102 L odging w hile out o f tow n
21021 L odging w hile ou t o f tow n
EC
EC
26 Gas (piped) and electricity
2601 E lectricity
26011 Electricity
28 T extile h ousefu m ish in gs
2801 L inens, curtains, drapes, sewing
m aterials
28011 B athroom linens
22 R ental equivalence and h ouseh old
insurance
2201 O w ners’ equivalent rent
22011 O w ners’ equivalent rent
28012 B edroom linens
2202 H ou seh old insurance
22021 H ou seh old insurance
28014 Curtains and drapes
28013 Kitchen and dining room linens
3109 U npriced accessories for electronic
equipm ent
31090 U npriced item s
EC
32 Other h ou seh old equipm ent and
furnishings
3201 F lo o r/w in d o w coverings, outd o o r/in fa n t/la u n d r y /d e a n in g equipm ent
32011 Floor coverings
32012 W indow coverings
28015 Slipcovers and decorative pillow s
EC
23 M aintenance and repair services
2301 Property m aintenance and repair
services
23011 Inside h om e m aintenance and repair
services
32013 In fa n ts’ equipm ent
28016 Sew ing m aterials for household
item s
EC
23012 R epair/replacem ent o f hard surface
flooring
29012 B edroom furniture other than
m attress/an d springs
23013 R eplacem ent o f installed w all-to-w all
carpet
24 M aintenance and repair com m odities
2401 M aterials, supplies, equipm ent for hom e
repairs
24011 P aint, w allpaper and supplies
24012 T o o ls and equipm ent for painting
24015 Plum bing supplies and equipm ent
24016 Electrical supplies, heating and coolin g
equipm ent
2404 Other property m aintenance
com m od ities
24041 M iscellaneous supplies and equipm ent
29032 Living room tables
32032 C hina and other dinnerware
2904 Other furniture
29041 Kitchen and dining room furniture
32033 Flatware
29044 O ccasional furniture
EC
32022 Lam ps and lighting fixtures
3203 Tablew are, serving pieces, n onelectric
kitchenw are
32031 Plastic dinnerware
29043 O utdoor furniture
24014 Blacktop and m asonry materials
3202 C lock s, lam p s, and decorator item s
32021 C locks
2903 L iving room chairs and tables
29031 Living room chairs
29042 Infan ts’ furniture
24013 Lum ber, paneling, wall and ceiling tile;
aw nings, glass
32015 O utdoor equipm ent
32023 H o u seh o ld decorative item s
2902 S ofas
29021 S ofas
23014 R epair o f disp osal, built-in dishwasher,
range hood
EC
29 Furniture and bedding
2901 B edroom furniture
29011 M attress and springs
32014 Laundry and cleaning equipm ent
30 H ou seh old appliances
3001 R efrigerators and h om e freezers
30011 R efrigerators and h om e freezers
3002 Laundry equipm ent
30021 W ashers
32034 Glassw are
32035 Silver serving pieces
32036 Serving pieces other than silver or glass
32037 N onelectric cook in gw are
32038 T ablew are and nonelectric kitchenw are
3204 Lawn and garden eq uipm ent, to o ls,
hardware
32041 Law n and garden equipm ent
32042 Pow er to o ls
30022 Dryers
32043 Other hardware
24042 Hard surface floor covering
24043 Landscaping item s
3003 S toves, oven s, portable dishwashers,
w indow air-conditioners
196
32044 N onp ow ered hand tools
CPI Appendix 4.
Expenditure classes, item strata, and entry level items— Continued
3205 Small kitchen appliances, sew ing m achines,
portable h ea tin g /co o lin g equipm ent
32051 F loor cleaning equipm ent and sew ing
m achines
36013
32052 Portable h eatin g/coolin g equipm ent, small
electric kitchen appliances
36032 M en ’s nightwear
3206 Indoor plants and fresh cut flow ers
32061 Indoor plants and fresh cut flow ers
36034 M en ’s sweaters
39017 G irls’ hosiery and accessories
M en ’s coats and jackets
3909 U npriced girls’ uniform s and other
clothing
39090 U npriced item s
3603 M en ’s furnishings
36031 M en’s underwear and hosiery
36033 M en ’s accessories
EC
4002 B o y s’ and girls’ footw ear
40021 B o y s’ footw ear
36035.M en ’s active sportswear
3209 U npriced h ousehold equipm ent parts,
sm all furnishings
32090 U npriced item s
EC
33 H ousekeeping supplies
3301 Laundry and cleaning products
33011 Soaps and detergents
3303 H ou seh old paper products, including
stationery
33031 C leansing and toilet tissue, paper
tow els, napkins
EC
40022 G irls’ footw ear
3605 M en’s pants and shorts
36051 M en’s pants and shorts
4003 W om en ’s footw ear
40031 W om en ’s footw ear
41 Infan ts’ and toddlers’ apparel
4101 Infan ts’ and tod dlers’ apparel
41011 Infan ts’ and tod dlers’ outerwear
41012 Infan ts’ and tod dlers’ play and
dresswear
41013 Infan ts’ and tod dlers’ underwear
37012 B o y s’ sweaters
3305 Other h ouseh old products, law n and
garden supplies
33051 M iscellaneous h ouseh old products
37013 B o y s’ shirts
41014 Infan ts’ and tod dlers’ sleepwear
4109 U npriced in fan ts’ accessories and other
clothing
41090 U npriced item s
37014 B o y s’ underwear, nightwear, and
hosiery
37016 B o y s’ suits, sport coats, and pants
42 Sew ing m aterials and luggage
4201 Sewing m aterials, n o tio n s, luggage
42011 Fabric for m aking clothes
37017 B o y s’ active sportswear
42012 Sew ing n otion s and patterns
37015 B o y s’ accessories
34 H ousekeeping services
3401 P ostage
34011 P ostage
3402 Babysitting
34021 Babysitting services
3403 D om estic service
34031 D om estic services
EC
37 B o y s’ apparel
3701 B o y s’ apparel
37011 B o y s’ coats and jackets
33032 Stationery, stationery supplies, gift wrap
33052 Lawn and garden supplies
EC
3604 M en’s shirts
36041 M en’s shirts
3609 U npriced m en ’s uniform s and other
clothing
36090 U npriced item s
33012 Other laundry and cleaning products
40 Footw ear
4001 M en’s footw ear
40011 M en ’s footw ear
3709 U npriced b o y s’ uniform s and other
clothing
37090 U npriced item s
EC
3404 Other h ouseh old services
34041 G ardening and law ncare services
34043 M oving, storage, freight expense
42013 Luggage
EC
38 W o m en ’s apparel
3801 W om en ’s coats and jackets
38011 W om en ’s coats and jackets
3802 W o m en ’s dresses
38021 W om en ’s dresses
34042 W ater soften in g service
EC
4302 Jewelry
43021 Jewelry
EC
3803 W om en ’s separates, sportswear
38031 W om en ’s tops
34044 H ou seh old laundry and drycleaning,
excluding coin operated
34045 C oin-operated household laundry and
drycleaning
3406 A ppliance and furniture repair
34061 Repair o f television, radio, and sound
equipm ent
43 Jewelry
4301 W atches
43011 W atches
44 A pparel services
4401 Other apparel services
44011 S hoe repair and other shoe services
38032 W om en ’s skirts
44012 C oin-operated apparel laundry and
drycleaning
38033 W om en ’s pants and shorts
44013 A lterations and repairs
38034 W om en ’s active sportswear
44014 C lothing rental
3804 Women’s underwear, nightwear,
accessories
38041 W o m en ’s nightwear
44015 Watch and jewelry repair
4402 A pparel laundry and drycleaning,
excluding coin operated
44021 A pparel laundry and drycleaning,
excluding coin operated
34062 Repair o f h ouseh old appliances
38042 W o m en ’s underwear
34063 R eupholstery o f furniture
38043 W om en ’s hosiery
3407 Care o f invalids, elderly, and con
valescents in the hom e
34071 C are o f invalids, elderly, and con
valescents in the h om e
EC
38044 W om en ’s accessories
3805 W om en ’s suits
38051 W o m en ’s suits
3409 U npriced rent/repair o f h ouseh old equip
m ent, sound equipm ent
34090 U npriced item s
34091 U npriced items
EC
35 T en an ts’ insurance
3501 T en an ts’ insurance
35011 T en an ts’ insurance
EC
36 M en’s apparel
3601 M en ’s suits, coats, sportcoats, jackets
36011 M en ’s suits
36012 M en ’s sport coats and tailored jackets
4502 N ew trucks
45021 N ew trucks
3809 U npriced w om en ’s uniform s and other
clothing
38090 U npriced item s
EC
39 G irls’ apparel
3901 G irls’ apparel
39011 G irls’ coats and jackets
45 N ew vehicles
4501 N ew cars
45011 N ew cars
4503 N ew m otorcycles
45031 N ew m otorcycles
EC
39012 G irls’ dresses and suits
46 U sed vehicles
4601 U sed cars
46011 U sed cars
4609 U npriced other used m otor vehicles
46090 U npriced item s
39013 G irls’ tops
39015 G irls’ active sportswear
47 M otor fu el, m otor o il, co o la n t, and fluids
4701 M otor fuel
47011 M otor fuels
39016 G irls’ underwear and nightwear
4702 M otor o il, co o la n t, and other fluids
39014 G irls’ skirts and pants
197
EC
CPI Appendix 4.
EC
Expenditure classes, item strata, and entry level items— Continued
47021 M otor oil
53032 Taxi fare
47022 C oolan t, brake fluid, transm ission
flu id , additives
53033 Car and van p ools
48 A u tom ob ile parts and equipm ent
4801 Tires
48011 Tires
4802 V ehicle parts and equipm ent other than
tires
48021 V ehicle parts and equipm ent other
than tires
EC
5309 U npriced sch ool bus
53090 U npriced item s
EC
54 Prescription drugs and m edical supplies
5401 Prescription drugs and m edical supplies
54011 Prescription drugs and m edical supplies
EC
55 N onprescription drugs and m edical
supplies
5502 Internal and respiratory over-thecounter drugs
55021 Internal and respiratory over-thecounter drugs
49 A utom ob ile m aintenance and repair
4901 A utom otive body work
49011 A utom otive body work
4902 A utom otive drive-train, front end repair
49021 A utom otive drive-train repair
5503 N onprescription m edical equipm ent and
supplies
55031 T opicals and dressings
49022 A utom otive brake work
55032 M edical equipm ent for general use
49023 R epair to steering, front end, coolin g
system , and air-conditioning
4903 A utom otive m aintenance and servicing
49031 A utom otive m aintenance and servicing
55034 Hearing aids
4904 A utom otive pow er plant repair
49041 A utom otive pow er plant repair
5509 U npriced drugs
55090 U npriced item s
4909 U npriced au tom otive repair service
policy
49090 U npriced items
EC
50 A u tom ob ile insurance
5001 A u tom ob ile insurance
50011 A u tom ob ile insurance
EC
51 V ehicle finance charges
5101 A u tom ob ile finance charges
51011 A u tom ob ile finance charges
5109 U npriced other vehicle finance charges
51090 U npriced items
EC
55033 Supportive and convalescent m edical
equipm ent
EC
5602 D ental services
56021 D ental services
5603 Eyeglasses and eye care
56031 Eyeglasses and eye care
5604 Services by other m edical professionals
56041 Services by other m edical professionals
EC
52 V ehicle rental, registration, and inspection
5201 State and local autom obile registration,
license, inspection
52011 State au tom obile registration
57 H osp ital and other m edical care services
5701 H ospital room , in-patient
57011 H ospital room , in-patient
5702 Other in-patient services
57021 H osp ital in-patient services other than
room
52012 Local au tom obile registration
57022 N ursing and convalescent h om e care
52013 D river’s license
5703 H ospital out-patient services
57031 H ospital out-patient services
52014 V ehicle inspection
5709 U npriced rent or repair o f m edical
equipm ent
57090 U npriced item s
5205 Other autom obile-related fees
52051 A u tom ob ile rental
52052 Truck rental
52053 Parking fees
52054 V ehicle tolls
52055 A u tom ob ile tow in g charges
52056 Other vehicle rentals
5209 U npriced docking and landing fees
52090 U npriced items
EC
56 P rofessional services
5601 P h ysicians’ services
56011 P h ysicians’ services
53 Public transportation
5301 A irline fare
53011 A irline fare
5302 Other intercity transportation
53021 Intercity bus fare
53022 Intercity train fare
53023 Ship fares
5303 Intracity transportation
53031 Intracity mass transit
EC
58 H ealth insurance
5811 C om m ercial health insurance retained
earnings-prescription drugs
58111 C om m ercial health insurance retained
earnings-prescription drugs
5812 C om m ercial health insurance retained
earnings-p hysician s’ services
58121 C om m ercial health insurance retained
earnings-p hysician s’ services
5813 Com m ercial health insurance retained
earnings-dental services
58131 Com m ercial health insurance retained
earnings-dental services
5816 Commercial health insurance retained
earnings-hospital room
58161 Commercial health insurance retained
earnings-hospital room
5817 Commercial health insurance retained
earnings-other in-patient hospital services
58171 Commercial health insurance retained
earnings-other in-patient hospital services
5818 Commercial health insurance retained
earnings-out-patient hosp services
58181 Commercial health insurance retained
earnings-out-patient hosp services
58211 Blue Cross/Blue Shield retained
earnings-prescription drugs
5822Blue Cross/Blue Shield retained
earnings-physicians’ services
58221 Blue Cross/Blue Shield retained
earnings-physicians’ services
5823 Blue Cross/Blue Shield retained
earnings-dental services
58231 Blue Cross/Blue Shield retained
earnings-dental services
5824 Blue Cross/Blue Shield retained
earnings-eye care services
58241 Blue Cross/Blue Shield retained
earnings-eye care services
5825 Blue Cross/Blue Shield retained
earnings-other professionals’ services
58251 Blue Cross/Blue Shield retained
earnings-other professionals’ services
5826 Blue Cross/Blue Shield retained
earnings-hospital room
58261 Blue Cross/Blue Shield retained
earnings-hospital room
5827 Blue Cross/Blue Shield retained
earnings-other in-patient hospital services
58271 Blue Cross/Blue Shield retained
earnings-other in-patient hospital services
5828 Blue Cross/Blue Shield retained
earnings-out-patient hospital services
58281 Blue Cross/Blue Shield retained
earnings-out-patient hospital services
5831HMO retained earnings-prescription
drugs
583117 HMO retained earnings-prescription
drugs
5832 HMO retained earnings-physicians’
services
58321 HMO retained earnings-physicians’
services
5833 HMO retained earnings-dental services
58331 HMO retained earnings-dental services
5834 HMO retained earnings-eyecare services
58341 HMO retained earnings-eyecare services
5835 HMO retained earnings-other profes
sionals’ services
58351 HMO retained earnings-other profes
sionals’ services
5836 HMO retained earnings-hospital room
58361 HMO retained earnings-hospital room
5814 C om m ercial health insurance retained
earnings-eye care services
58141 C om m ercial health insurance retained
earnings-eye care services
5837 HMO retained earnings-other in-patient
hospital services
58371 HMO retained earnings-other in-patient
hospital services
5815 Com m ercial health insurance retained
earnings-other professional services
58151 Com m ercial health insurance retained
earnings-other professional services
5838 HMO retained earnings-out-patient
hospital services
58381 HMO retained earnings-out-patient
hospital services
198
CPI Appendix 4.
Expenditure classes, item strata, and entry level items—Continued
5841 Other health insurance retained
earnings-prescription drugs
58411 Other health insurance retained
earnings-prescription drugs
5842 Other health insurance retained
earnings-p hysician s’ services
58421 Other health insurance retained
earnings-p hysician s’ services
61032 Purchase o f pets, pet supplies,
accessories
6109 U npriced souvenirs, fireworks, optic
good s
61090 U npriced item s
EC
5843 Other health insurance retained
earnings-dental services
58431 Other health insurance retained
earnings-dental services
62054 Veterinarian services
6209 U npriced rental o f recreational
vehicles
62090 U npriced item s
59 Reading materials
5901 N ew spapers
59011 N ew spapers
EC
5909 U npriced newsletters
59090 U npriced items
60 Sporting g ood s and equipm ent
6001 Sports vehicles, including bicycles
60011 O utboard m otors and powered sports
vehicles
64 T oilet g o o d s and personal care
appliances
6401 H air, dental, shaving, m iscellaneous
personal care products
64011 Products for the hair
68023 Tax return preparation and
other accounting fees
6803 Cem etery lots and funeral
expenses
68031 Funeral expenses
6809 U npriced m iscellaneous personal
services
68090 U npriced item s
EC
64016 D eod o ra n t/su n ta n preparations,
sa n itary/footcare products
EC
68022 C hecking accounts and special
check services
64015 Shaving products, nonelectric
shaving articles
61 T oys, hobbies, and other entertainment
com m odities
6101 T oys, h obb les, and m usic equipm ent
61011 T oys, gam es, and hobbies
6103 Pets and pet products
61031 Pet food
6309 U npriced sm oking products
63090 U npriced item s
64014 D ental products, nonelectric dental
articles
60022 H unting, fishing, and cam ping
equipm ent
61023 P hotographic equipm ent
6802 Banking and accounting expenses
68021 S afe deposit box rental
68032 Cem etery lots and crypts
6002 Sports equipm ent
60021 Ind oor, warm w eather, and winter
sports equipm ent
61022 P hotographic and darkroom supplies
68 Legal, financial, and funeral
services
6801 Legal fees
68011 Legal fees
63013 Sm oking accessories
64013 W om an ’s hair pieces and wigs
60013 Bicycles
6102 P hotographic supplies and equipm ent
61021 Film
EC
64012 N onelectric articles for the hair
60012 U npow ered boats and trailers
61013 M usic instrum ents and accessories
63 T ob acco products
6301 T ob acco and sm oking supplies
63011 Cigarettes
6709 U npriced m iscellaneous school
item s, rentals and other services
67090 U npriced item s
63012 T ob acco products other than
cigarettes
59022 B ook s purchased through book clubs
61012 Playground equipm ent
6704 Other tuition and fees
67041 Technical and business school
tuition and fixed fees
62055 Other entertainm ent services
5902 M agazines, periodicals, and books
59021 M agazines
EC
6703 Child daycare, nursery school
67031 D aycare and nursery school
62053 Pet services
EC
67 D aycare, tu ition , and other sch ool
fees
6701 C ollege tuition and fees
67011 C ollege tuition and fixed fees
6702 E lem entary and high sch ool
tuition and fees
67021 Elem entary and high school
tu ition and fixed fees
62052 Film processing
5847 Other health insurance retained
earnings-other in-patient hospital services
58471 Other health insurance retained
earnings-other in-patient hospital services
EC
EC
6205 P h otographers, film processing, pet
services
62051 Photographer fees
5846 Other health insurance retained
earnings-h osp ital room
58461 Other health insurance retained
earnings-h osp ital room
59023 B ook s not purchased through b ook
clubs
6609 U npriced m iscellaneous sch ool
purchases
66090 U npriced item s
6204 Fees for lessons or instructions
62041 Fees for lessons or instructions
5845 Other health insurance retained
earnings-other p rofession als’ services
58451 Other health insurance retained
earnings-other p rofession als’ services
EC
66022 E ncyclopedias and other sets o f
reference books
6203 A dm issions
62031 A dm ission to m ovies, theaters, and
concerts
62032 A dm ission to sporting events
66 S ch ool b ooks and supplies
6601 Sch ool b ook s and supplies for
college
66011 C ollege textbooks
6602 R eference b ook s and elem entary
and high school books
66021 E lem entary and high school
b ook s and supplies
62 Entertainm ent services
6201 Club m em bership dues and fees
62011 C lub m em bership dues and fees
6202 Fees for participant sports
62021 Fees for participant sports
5844 Other health insurance retained
earnings-eyecare services
58441 Other health insurance retained
earnings-eyecare services
5848 Other health insurance retained
earnings-ou t-patien t hospital services
58481 Other health insurance retained
earnings-ou t-patien t hospital services
EC
69 Inform ation processing equipm ent
6901 Inform ation processing
equipm ent
69011 Personal com puters and
peripheral equipm ent
64017 Electric personal care appliances
69012 Com puter softw are and
accessories
6403 C o sm e tics/b a th /n a il preparations and
im plem ents
64031 C osm etics, b a th /n a il/m a k e-u p
preparations and im plem ents
69013 T elephone, peripheral
equipm ent, and accessories
69014 C alculators, adding m achines,
and typewriters
65 Personal care services
6501 Beauty parlor services for fem ales
65011 Beauty parlor services for fem ales
6502 H aircuts and other barber shop
services for m ales
65021 Haircuts and other barber shop
services for m ales
6509 U npriced repair o f personal care
appliances
65090 U npriced item s
199
69015 Other in form ation processing
equipm ent
EC
72 U tility average prices
7260 U tility natural gas, 40 therm s
72601 U tility natural gas, 40 therms
7261 U tility natural gas, 100 therm s
72611 U tility natural gas, 100 therm s
7262 Electricity, 500 kilow att hours
72621 Electricity, 500 kilow att hours
CPI Appendix 5.
Sample Allocation Methodology for Commodities and Services
a2 . . = a2
Introduction
unit, j
. . + a 2 . . + a2
. . + a2
p,unit,j
e,umt,j
o,unit,j
r,umt,j
where:
The objective of the item-outlet sample design for the
commodities and services component of the CPI revision
was to determine the sample resource allocation by p s u ,
replicate, item stratum, and p o p s category which
minimizes the sampling variance of the CPI at the U.S.
level, while at the same time meeting certain budgetary
constraints on total expenditures and total travel
expenditures.
Certain simplifying assumptions were made to render
this problem manageable. First, all item strata were di
vided into eight major groups:
Food and beverages
Fuel and utilities
Household furnishings
Apparel and upkeep
Transportation
Medical care
Entertainment
Other commodities and services
Second, all sample p s u ’s were divided into 10 groups.
Further, it was decided that the number of item selec
tions in each major group would remain the same across
all p s u ’s , and that the number of outlet selections per
p o p s category would remain the same within a PSUmajor group. This reduced the design problem to one in
88 variables, {M-} and {Kj}, i= 1,...,10, j = 1,...,8,
where M;j represents the designated outlet sample size
per POPS category for PSU group i and major group j,
and Kj represents the designated item sample size for
major group j.
Sampling variance and cost functions were developed
in terms of these design variables.
is the component of unit variance due to the
sampling of psu ’s,
a 2
...
is the component
of unit variance due to the same.u m t.j
r
pling of eli’s within item strata,
a2
... is the component of unit variance due to the
sampling of outlets, and
° r unit j
*s
residual component of unit variance.
a 2
. .
Given these assumptions, it follows that each compo
nent of o2k can be expressed in terms of its correspon
ding unit variance components:
a2 . =
p ,j,k
a2
. ,/N '
P.unit.j
k
where:
N 'k is the number of non-self-representing psu ’s in the
index area (Note:
. k is 0 for self-representing psu ’s);
a2
. = (a2
,/N k, • H.k • K.)
e,j,k
v e,u m t,j
y • fpC:
r J • NC:J
where:
Nk is the number of psu ’s,
Hk is the number of replicates per psu in the index area,
fpcj = (1-Kj/TIj) is a finite population correction factor,
TIj is the number of eli’s in the major group,
NCj is the percent of the strata in the major group which
are noncertainty strata, i.e., containing more than one
eli;
a2o ,j,k
. = a2o ,u n it,j
,f ./(N.
v k
The sampling variance function
H,
M.'J . k
pi>
where:
The variance function for the CPI revision was modeled
for index areas, i.e., geographic areas defined by p s u ’s
or groups of p s u ’s . Each self-representing PSU constitutes
a single index area. Non-self-representing p s u ’s were
grouped into 12 index areas, each composed of 2 to 10
p s u ’s . It was assumed that the total variance of price
change for major commodity group j within index area
k can be expressed as the sum of four components:
.,
AJ.k = a2P.J.k
a2
e,j,k
.k
P,J’
< 2 . k
j
.
k
<t^ . k
is the component of variance due to the sampling
of psu’s,
is the component of variance due to the sampling
of eli’s within item strata,
is the component of variance due to the sampling
of outlets, and
is the residual component of variance.
Similarly, it was assumed that the variance of price
change of a sample unit (i.e., a single quote) within a
major group j can be given by:
P=
is the number of pops categories in the major group,
and
a2r,j,k = a2
.f ./(N.
r,unit,j v k
H,k • M.j,k
K.
-j
•V
This gives the sampling variance of the national com
modities and services index as:
+ a2o ,j,k
. , + o2
■.
r.j.k
where:
a 2
Mj k is the number of unique inscope outlets selected per
psu-replicate per pops category,
4 otal =
f
(relimPj)2 E (wk)2 «f>k
where:
wk is the population weight of index area k,
relimpj is the relative importance of major group j.
The relative importance of an item stratum or major
group is obtained from the Consumer Expenditure Survey
and is the percentage of total expenditures on all items
which are expenditures on items in the stratum or major
group. In this application, the relative importance data
used were index area averages.
200
The cost function
Cpv T j is the travel cost for a personal visit for pricing
per outlet for major group j,
The total annual cost of the commodities and services
components of the CPI revision includes costs of initia
tion data collection, processing and review, personal visit
and telephone pricing, and pricing processing and review.
The costs of initiation of data collection, processing, and
review were developed as either outlet- or quote-related
costs. For PSU group i and major group j, outlet-related
costs for initiation are:
RT 0 j
is the proportion of outlets priced by telephone
for major group j,
CT o j
is the cost for telephone collection per outlet for
major group j;
MBy
is a factor to adjust for the monthly/bimonthly
mix of outlets and quotes by psu and major pro
duct group,
Cpv
CI0 (Mij.Kj) = .2 Nj • Hj • (COJ +CT>j)
q
RT
q
j
CT
q
j
Cp
q
j
• <aij Mij2 + bij Mij + cij) • p j
and quote-related costs for initiation are:
CIQ (My, Kp = .2 Nj • Hj • (CQJ + C 'QJ) • My • K- • NRj
where:
Nj
is the number of psu ’s in psu group i,
Hj
is the number of replicates per psu in psu group i,
is the per quote cost for a personal visit for
pricing,
is the proportion of telephone collected quotes
for major group j,
is the per quote cost for telephone collection for
major group j, and
is the per quote cost for processing repricing data.
The total cost fun ction associated w ith data collection
and processing for the com m odities and services index,
sum m ed over all m ajor groups and p s u groups, is then
given by:
CQ j is the initiation cost per outlet for major group j,
TCOST = E CIG (My, Kj) + CIQ (My, Kj)
CTj is the travel cost at initiation per outlet for major
+ CP0 (My, Kj) + CPQ (My, Kj)
group j,
(•a H i2 + byMy + Cjj) is a quadratic overlap function
used to predict the number of unique sample outlets, account
ing for the overlap of elements in the outlet sample within and
between major groups for a psu-replicate,
Pj
is the number of pops categories in major group j,
The total travel cost function depends only on the
expected number of unique outlets in the sample and is
a subtotal of the above, namely:
TRCOST = E Nj ■Hj • Pj {.2Cx j (ay My2 + by My + Cy)
Cq j is the initiation cost per quote for major group j,
Cqj is the initiation processing cost per quote for major
group j,
NRj is the outlet in-scope rate for major group j.
The 0.2 factor in the above cost formulas accounts for
the rotation or reinitiation of the outlet sample in onefifth of the sample p s u ’s each year. Note that the
expected number of quotes per PSU-replicate-major
group is estimated by the product of the number of
designated outlets and the number of item strata hits,
+ MBy NRj (ay Mjj2 + by My + Cy) • Cpy Tj (l-Ry^.j)}
Note that the variance and total cost functions are
nonlinear in the sample design variables {My} and {Kj}.
The total travel cost function, however, is linear in the
variables {My} and does not depend on the {Kj}.
Thus the sample design problem can be expressed as:
minimize
My • Kj.
• (aij Mij2 + bij M'ij + cij) • p j
[(Cpv o.j + Cpv.T.j) ' ( I - P t .Oj ) + *-T,0,j ' ^T.O jJ
and quote-related costs for ongoing pricing are:
TCLIM,
TRCOST < TRAVLIM,
i= 1 ,. .. ,
M y> 1,
j = l.. • • ,
Kj > STRATAj, j = l.- • • ,
1 j = l , . .. ,
Kj < TI
The costs of ongoing price data collection, processing,
and review were also developed as either outlet- or quoterelated costs. For PSU group i and major group j, outletrelated costs for ongoing pricing are:
CPQ (Mj j, Kj) = MBjj • Nj ■ Hj ■NRj
^total
{My},{Kj} integer
subject to
TCOST <
10,
8,
8,
8.
Here, tc lim and t r a v l im are the design parameters
representing total expenditure and total travel expenditure
ceilings, respectively, and STRATAj and Tij are the design
parameters denoting the number of item strata and total
number of e l i ’s , respectively, in the jth major group.
CPQ (My, Kj) = MBy • N; • H; • My
Kj • [Cpv qj (1-RT qj) + CT q j Rt qj + Cp q]
where:
Cpv 0)j is the cost for a personal visit for pricing per outlet
for major group j,
Model coefficients
Estimates of components of the cost function were
developed from agency administrative records and
directly collected studies of travel time and within-outlet
201
time. Response rates for each major group were
developed from past initiation and pricing experience.
Overlap functions, used to project the number of uni
que outlets, were developed by modeling the number of
unique outlets obtained in simulations of the sampling
procedures for each PSU type. Since outlet samples are
selected independently for each point-of-purchase
category, an individual outlet may be selected for more
than one category. For example, a grocery store could
be selected for both bakery products and dairy products.
The number of unique outlets yielded by the sampling
process is needed to project outlet-related costs.
Estimates of the components of the variance function
were developed as follows. The total variance of price
change for 2- and 6-month changes was computed by
major group and for all items less housing using CPI data
for January 1980 through December 1982. Components
of the variance of price change for each major group were
estimated using a three-way analysis of variance technique.
The total unit variance and unit components of variance
were computed by dividing the total variance and com
ponents of variance by the sample sizes employed for the
respective indexes for each time period. The component
unit variances and the total unit variance computed as
the sum of the components were ratio-adjusted to the total
unit variance computed directly from the c pi data to
assure data consistency. Finally, generalized unit variance
functions were obtained by modeling observations of the
unit variance of price change as a function of price change
by major group. The total unit variance of price change
used in the solution of the sample design problem was
generated from the generalized unit variance functions.
Intraclass correlations for each major group were
obtained from models of the component unit relative
variances. Relative variances were computed for the PSU,
outlet, and ELI components by dividing the correspond
ing unit variances by the squared price change. Compo
nents of relative variance were then modeled as functions
of price change with the functional form:
Y = bl X b2
where:
Y = the PSU, ELI, or outlet relative variance and
X = price change.
Intraclass correlations were then calculated by dividing
the components of relative variance by their sum for each
price change of interest. Final estimates of components
of relative variance for a given time change and infla
tion rate were then obtained by multiplying modeled unit
relvariances and their corresponding modeled estimates
of intraclass correlations.
The solution
A sequential unconstrained minimization technique,1
implemented in the nonlinear programming code Sym
bolic Factorable su m t 2 was used to solve the design pro
blem. Solution values of the sets {M^} and {Kj} were
computed for various values of tclim and t r a v l im and
for modeled estimates of components of variance com
puted for various annual inflation rates and time periods.
For each major group j, Kj was bounded below by the
number of item strata in the major group and above by
the number of e l i ’s in the major group. For the food
and beverages group, a lower constraint of 73 item strata
selections was imposed in order to support average food
prices.
Unit variance and intraclass correlation estimates used
were for a 6-month price change at a 10-percent annual
rate. Design solutions were also found using model
estimates for 2-month price changes at both 8-percent and
10-percent annual rates. Only minor differences were
observed between the problem solutions found with vari
ance estimates for the 6-month price change at 8-percent
and 10-percent annual rates. The problem solution found
for a 6-month price change at a 10-percent annual rate
was selected as the final sample design because the esti
mates of unit variances and intraclass correlations for
some major groups were slightly less stable at the 8-percent annual rate. Solutions using estimates for 2-month
price changes were not used since some major groups have
little or no price change in a short period.
1 A . V . F ia c c o a n d G . P . M c C o r m ic k , N on lin ear P ro g ra m m
ing: Sequential U nconstrained M in im ization Techniques ( N e w Y o r k ,
W ile y , 1 9 6 8 ).
2 A . G h a e m i a n d G . P . M c C o r m ic k , “ F a c t o r a b le SU M T : W h a t Is
It? H o w is It U s e d ? ” T e c h n ic a l P a p e r S e r ia l T - 4 0 2 ( W a s h i n g t o n , D C .
T h e G e o r g e W a s h in g to n U n iv e r s it y , I n s t it u t e f o r M a n a g e m e n t S c ie n c e
a n d E n g in e e r in g , 1 9 7 9 ).
202
CPI Appendix
001
6.
pops
Categories
Prescription drugs
54011 Prescription drugs and medical
supplies
002
Over-the-counter drugs, medicines, and
medical supplies
55021 Internal and respiratory over-thecounter drugs
55031 Topicals and dressings
55032 Medical equipment for general use
55033 Supportive and convalescent medical
equipment
55034 Hearing aids
003
005
006
Laundry and drycleaning, not coin
operated
34044 Household laundry and drycleaning,
excluding coin operated
44021 Apparel laundry and drycleaning,
excluding coin operated
007
36033 M en ’s accessories
022 R ecords, tapes, needles
31033 R ecords and tapes, prerecorded and
blank
023
Fees for participant sports
28011 B athroom linens
28012 B edroom linens
28013 Kitchen and dining room linens
026
Wine for home use
027
Whiskey and other liquors for home use
20021 Whiskey at home
20022 Distilled spirits at home (excluding
whiskey)
013
In d o o r/o u td o o r plants and garden supplies
M en ’s suits
028 M en ’s sportcoats and tailored jackets
36012 M en’s sport coats and tailored jackets
029 M en ’s overcoats, top coats, raincoats, jackets
36013 M en ’s coats and jackets
030 W om en ’s suits, including pantsuits
Beer and ale for home use
20011 Beer, ale, and other alcoholic malt
beverages at home
Plastic dinnerware
C hina and other dinnerware
Flatware
G lassw are
Silver serving pieces
Serving pieces other than silver or glass
36011 M en’s suits
20031 Wine at home
012
D innerw are, glassware, flatw are, and serving
pieces
24043 Landscaping item s
32061 Indoor plants and fresh cut flow ers
33052 Lawn and garden supplies
62021 Fees for participant sports
011
38051 W om en ’s suits
031
W om en ’s dresses
38021 W o m en ’s dresses
014
015
Alcoholic beverages purchased in
restaurants and bars
38011 W om en ’s coats and jackets
034 W om en ’s active sportswear and playwear
Toys, games, hobbies, tricycles, and
battery-powered riders
035
and tod dlers’ outerw ear
and tod dlers’ play and
and tod dlers’ underwear
and tod dlers’ sleepwear
S hoe repair and other shoe services
44011 S hoe repair and other shoe services
042 Repair o f h ousehold appliances, except radio,
T V , sound equipm ent
23014 R epair o f disp osal, built-in dishwasher,
range hood
34062 Repair o f h ouseh old appliances
043 M ajor h ouseh old appliances
30011 R efrigerators and h om e freezers
30021 W ashers
30022 Dryers
30031 Stoves and oven s excluding m icrow ave
ovens
30032 M icrow ave ovens
30033 Portable dishwashers
30034 W indow air-conditioners
32051 Floor cleaning equipm ent and sew ing
m achines
69015 Other in form ation processing
equipm ent
044 Sm all electric appliances for kitchen, personal
care, etc.
32052 Portable c o o l/h e a t equipm ent, sm all
electric kitchen appliances
64017 Electric personal care appliances
045 S o ft surface flo o r covering
23013 R eplacem ent o f installed w all-to-w all
carpet
32011 F loor coverings
046 W indow and furniture coverings, upholstery,
decorative pillow s
032 W om en ’s co a ts, jack ets, raincoats
20051 Beer, ale, and other alcoholic malt
beverages away from home
20052 Wine away from home
20053 Distilled spirits away from home
61011 Toys, games, and hobbies
41011 Infan ts’
41012 Infan ts’
dresswear
41013 Infan ts’
41014 Infan ts’
041
024 H ou seh old linens
62031 Admission to movies, theaters, and
concerts
010
039 Infan ts’ and tod dlers’ clothing and
accessories
Repair o f T V , rajiio, other soun d equipm ent
32031
32032
32033
32034
32035
32036
coats and jackets
dresses and suits
tops
skirts and pants
active sportsw ear
underwear and nightwear
hosiery and accessories
40022 G irls’ footw ear
34061 Repair o f television , radio, and sound
equipm ent
025
G irls’
G irls’
G irls’
G irls’
G irls’
G irls’
G irls’
038 G irls’ footw ear
36051 M en ’s pants and shorts
Stationery, greeting cards, gift wrap, wrap
accessories
Admissions to movies, theaters, concerts:
combined season and single (367)
39011
39012
39013
39014
39015
39016
39017
020 M en ’s accessories
33032 Stationery, stationery supplies, gift
wrap
009
40021 B o y s’ footw ear
38041 W om en ’s nightwear
M en’s trousers
accessories
suits, sport co a ts, and pants
active sportsw ear
037 G irls’ clothing and accessories
019 W om en ’s sleepwear
021
shirts
underwear, nightw ear, and
036 B o y s’ footw ear
W om en ’s accessories
38044 W om en ’s accessories
Laundry and drycleaning, self-service
34045 Coin-operated household laundry
and drycleaning
44012 Coin-operated apparel laundry and
drycleaning
008
018
Personal care services for males
65021 Haircuts and other barber shop
services for males
M edical/su rgical care by general practitioners
and specialists
56011 P h ysicians’ services
Women’s hosiery
38043 Women’s hosiery
37013 B o y s’
37014 B o y s’
hosiery
37015 B o y s’
37016 B o y s’
37017 B o y s’
56031 Eyeglasses and eye care
017
Personal care services for female
65011 Beauty parlor services for females
004
016 Eye exam ination, eye care, glasses, contact
lenses
38034 W o m en ’s active sportswear
B oys’ clothing and accessories
37011 B o y s’ coats and jackets
37012 B o y s’ sweaters
203
28014
28015
32012
34063
C urtains and drapes
Slipcovers and decorative pillow s
W indow coverings
R eupholstery o f furniture
047 A uto m o tiv e b ody repair
49011 A uto m o tiv e b ody work
048 Inside repair, replacem ent, in stallation, and
m aintenance o f property
CPI Appendix
6.
pops
Categories— Continued
23011 Inside h om e m aintenance and repair
services
049
R adios, tape recorders/players, and
phonographs
31021 V ideo cassette recorders, disc
players, cam eras, and accessories
31031 R adios phonographs, and tape
recorders/players
31032 C om pon en ts and other sound
equipm ent
067 Sports equipment, including unpowered
sports vehicles
60012 Unpowered boats and trailers
60021 Indoor, warm weather, and winter
sports equipment
60022 Hunting, fishing, and camping
equipment
P h otograp h ic equipm ent
61023 P hotographic equipm ent
052
Lam ps and lighting fixtures
32022 Lam ps and lighting fixtures
053
Pictures, m irrors, clock s, and other hom e
decorations
32021 C locks
32023 H ou seh old decorative item s
054
H ou seh old furniture
29011 M attress and springs
29012 B edroom furniture other than
m attress and springs
29021 S ofas
29031 Living room chairs
29032 Living room tables
29041 Kitchen and dining room furniture
29044 O ccasional furniture
055
TV and TV com b in ation s
31011 T elevision sets
056
M en’s sweaters and vests
M oving expenses, including freight and
storage
34043 M oving, storage, freight expense
059
H ospital care
57011 H ospital room , in-patient
57021 H osp ital, in-patient services other
than room
57031 H osp ital out-patient services
060
N ew cars
062
O ffice equipm ent for h om e u se such as
typew riters, etc.
45011 N ew cars
69014 C alculators, adding m achines, and
typewriters
070 Patio, porch, other outdoor furniture and
equipment
44013 A lteration s and repairs
44014 C loth in g rental
46011 Used cars
076 Nonelectric cookware, kitchen utensils,
laundry and cleaning equipment, closetstorage items
32014 Laundry and cleaning equipment
32037 Nonelectric cookingware
32038 Tableware and nonelectric kitchenware
077 Automotive repair to engine and related
equipment
49021 Automotive drive train repair
49041 Automotive power plant repair
49022 Automotive brake work
49023 Repair to steering, front end, cooling
system, and air-conditioning
49031 Automotive maintenance and servicing
52055 Automobile towing charges
079 Automotive parts, accessories, and products,
excluding tires
47021 Motor oil
47022 Coolant, brake fluid, transmission
fluid, additives
48021 Vehicle parts and equipment other than
tires
080 Luggage
42013 Luggage
082 Fish and seafood
32042 Pow er tools
B icycles, bicycle parts and accessories, and
bicycle repair
60013 Bicycles
066
088
Playground equipm ent
61012 Playground equipm ent
62032 A dm ission s to sporting events
101
G asoline and other vehicle fuels
47011 M otor fuels
102 T ob acco products
63011 C igarettes
63012 T ob acco products other than cigarettes
103
Personal care item s
64011 Products for the hair
64012 N onelectric articles for the hair
64014 D en tal products, n onelectric dental
articles
64015 Shaving products, nonelectric shaving
articles
64016 D eo d o ra n t/su n ta n preparations,
sa n ita ry /fo o tca re products
64031 C osm etics, b a th /n a il/m a k eu p
preparations and im plem ents
104 C leaning and laundry p roducts, paper
supplies, other h ouseh old supplies
33011 Soaps and detergents
33012 Other laundry and cleaning products
33031 C leansing and toilet tissue, paper
tow els, napkins
33051 M iscellaneous h ouseh old products
105 Bakery products
07011 Canned fish or seafood
07021 Shellfish (excluding canned)
07022 Fish (excluding canned)
083 Processed fruits and vegetables
13011
13012
13013
13031
14011
W atches, jew elry, and repair
43011 W atches
43021 Jewelry
44015 W atch and jew elry repair
090 Tickets to sporting events: com b in ed season
and single tickets (366)
078 Miscellaneous automotive repair,
maintenance, servicing
M en ’s active sportsw ear and playwear
Pow er tools
R oasted co ffee
Instant and freeze dried co ffee
N oncarb onated fruit flavored drinks
T ea
Other n oncarbonated drinks
087 A pparel and accessory alteration, repair, and
rental
073 Used cars
64013 Women’s hair pieces and wigs
065
17031
17032
17051
17052
17053
072 Hard surface flooring and floor covering
36035 M en ’s active sportsw ear
064
Sugar and other sw eets, for h o m e use
086 C o ffee, tea, fruit flavored drinks, other
noncarbonated beverages
29043 Outdoor furniture
32015 Outdoor equipment
23012 Repair/replacement of hard surface
flooring
24042 Hard surface floor covering
M argarine
Other fats and oils
N ond airy cream substitutes
P eanut butter
15011 C andy and chew ing gun
15012 Other sweets (excluding can d y and
gum )
15021 Sugar and artificial sweeteners
29042 Infants’ furniture
32013 Infants’ equipment
081 Wigs and hairpieces for females
063
085
069 Infants’ furniture and equipment
36034 M en ’s sweaters
058
16011
16012
16013
16014
Lawn m ow ing and other yard equipm ent
32041 Lawn and garden equipm ent
057
084 Fats, o ils, peanut butter , salad dressings,
dairy product substitutes
068 Musical instruments and accessories,
including sheet music
61013 Music instruments and accessories
051
14021 C anned beans other than lim a beans
14022 C anned cut corn
14023 Other processed vegetables
Frozen orange juice
Other frozen fruits and fruit juices
Fresh, canned, or bottled fruit juices
Canned and dried fruits
Frozen vegetables
204
02011 W hite bread
02021 Bread other than w hite
02022 R o lls, biscuits, m u ffin s (excluding
frozen)
02041 C akes and cupcakes (excluding frozen)
02042 C ookies
02061 Crackers
02062 Bread and cracker products
02063 Sw eetrolls, co ffe e cake and d oughnuts
(excluding frozen)
CPI Appendix
6.
pops
Categories— Continued
02064 Frozen bakery products and
frozen/refrigerated d oughs and batters
02065 Pies, tarts, turnovers (excluding
frozen)
106
108
19031 Snacks and nonalcoh olic beverages
Eggs
Fresh w hole m ilk
Other fresh m ilk and cream
Butter
Other dairy products
C heese
Ice cream and related products
A pples
Bananas
Oranges
Other fresh fruits
P otatoes
Lettuce
T om atoes
Other fresh vegetables
68023 Tax return preparation and other
accounting fees
134 Business and technical schools
67041 Technical and business school tuition
and fixed fees
27011 Telephone services, local charges
61031 Pet food
140 Other vehicle rentals
114 Footw ear for men
40011 M en ’s footw ear
52056 Other vehicle rentals
141 Passenger ship carriers
115 Shirts for men
53023 Ship fares
36041 M en’s shirts
117 S ocks, underwear, sleepw ear, and bathrobes
for m en
36031 M en ’s underwear and hosiery
36032 M en’s nightwear
118 Footw ear for w om en
40031 W om en ’s footw ear
120 Separates and coordinates for w om en
38031 W om en ’s tops
38032 W o m en ’s skirts
38033 W o m en ’s pants and shorts
122 Underwear for w om en
38042 W om en ’s underwear
123 Sew ing m aterials and notions
28016 Sew ing m aterials for household item s
42011 Fabric for m aking clothes
42012 Sew ing n otion s and patterns
142 Van and carpools used for commuting
53033 Car and van pools
144 Water softening service
34042 Water softening service
146 Elementary and high school books and
supplies
66021 Elementary and high school books and
supplies
301 Automobile insurance
50011 Automobile insurance
302 Pipes, lighters, lighter fuel, and other
smoking accessories
63013 Smoking accessories
303 College tuition and fixed fees
67011 College tuition and fixed fees
304 Housing at school (excluding board)
124 D ental care
56021 D ental services
126 Film and film processing
61021 Film
62052 Film processing
21031 Housing at school (excluding board)
305 Electricity
26011 Electricity
307 Homeowners’ and tenants’ insurance
Carbonated beverages for h om e use
17011 C ola drinks
127 M aterials and supplies for m ajor hom e
repairs
17012 Carbonated drinks other than cola
110
133 Personal income tax preparation fees and
other accounting fees
137 Local telephone service
113 Pet fo o d .
Fresh fruits and vegetables
11011
11021
11031
11041
12011
12021
12031
12041
109
112 Snacks and beverages aw ay from hom e
Dairy products, including eggs
08011
09011
09021
10011
10012
10021
10041
M eals in restaurants, cafeterias, carryouts,
drive-ins
19011 Lunch
19021 Dinner
19032 Breakfast or brunch
M eats and poultry
03011 Ground b eef
03021 Chuck roast
03031 R ound roast
03041 Other roasts (excluding chuck and
round)
03042 Other steak (excluding round and
sirloin)
03043 Other b eef
03051 R ound steak
03061 Sirloin steak
04011 Bacon
04021 Pork chops
04031 H am (excluding canned)
04032 Canned ham
04041 Pork roasts, picnics, other pork
04042 Pork sausage
05011 Frankfurters
05012 B ologn a, liverwurst, salami
05013 Other lunchm eats (excluding
b ologn a, liverwurst, salam i)
05014 Lam b, organ m eats, and gam e
06011 Fresh w hole chicken
06021 Fresh or frozen chicken parts
06031 Other poultry
107
111
24013 Lum ber, paneling, wall and ceiling tile,
aw nings, glass
24014 B lacktop and m asonry m aterials
M iscellaneous prepared fo o d s, cereals,
con d im en ts, and seasonings
01011 Flour
01012 Prepared flour mixes
01021 Cereal
01031 Rice
01032 M acaroni, similar products, and
cornm eal
18011 Canned and packaged soup
18021 Frozen prepared m eals
18022 Frozen prepared food s other than
m eals
18031 P otato chips and other snacks
18032 N uts
18041 Salt and other seasonings and spices
18042 O lives, pickles, relishes
18043 Other condim ents (excluding olives,
pickles, relishes)
18044 Sauces and gravies
18061 Canned or packaged salads and
desserts
18062 Baby food
18063 Other canned or packaged prepared
food s
128 A utom ob ile tires
26021 Utility natural gas service
309 Rental of miscellaneous equipment
62055 Other entertainment services
48011 Tires
129 Hardware item s, h and tools, and other
materials for m inor h om e repairs
24011 Paint, wallpaper and supplies
24012 T o o ls and equipm ent for painting
24015 Plum bing supplies and equipm ent
24016 Electrical supplies, heating and cooling
equipm ent
24041 M iscellaneous supplies and equipm ent
32043 Other hardware
32044 N onpow ered h andtools
131
35011 Tenants’ insurance
308 Utility natural gas service
Pet services
310 Fuel oil, kerosene, bottled or tank gas, coal,
and wood
25011
25021
25022
25023
Fuel oil
Bottled or tank gas
Coal
Other fuels
311 Electronic equipment for nonbusiness use in
the home
31023 Video game hardware, software, and
accessories
69011 Personal computers and peripheral
equipment
69012 Computer software and accessories
62053 Pet services
312 Telephones and accessories
132 Veterinarian services
62054 Veterinarian services
205
69013 Telephone peripheral equipment, and
accessories
CPI Appendix 6.
313
po ps Categories— Continued
Long-distance telephone service
27051 Interstate telephone services
27061 Intrastate telephone services
314
Finance charges for automobilesand other
vehicles
68021 Safe deposit box rental
68022 C hecking accounts and special check
services
State vehicle registration
68031 Funeral expenses
Local automobile registration (notState)
Driver’s license
Vehicle inspection
Automobile rental
27031 C om m unity antenna or cable TV
341
Truck and van rental
61032 Purchase o f pets, pet supplies,
accessories
360 N ew trucks and vans
45021 N ew trucks
Garbage and trash collection
27041 Garbage and trash collection
361
342 L odging away from hom e
52051 Automobile rental
322
357 Purchase o f pets, pet accessories, and pet
supplies (excluding fo o d )
339 C om m unity antenna and cable TV
52014 Vehicle inspection
321
60011 O utboard m otors and pow ered
sports vehicles
53032 T axi fares
52013 Driver’s license
319
356 Pow ered sports veh icles, such as b oats,
dunebuggies, g o lf carts, snow m ob iles
27021 R esidential water and sewer service
338 Taxicabs
52012 Local automobile registration
318
59022 B ook s purchased through b ook clubs
66022 E ncyclopedias and other sets o f
reference b ooks
337 W ater and sewer m aintenance
52011 State automobile registration
316
21021 L odging w hile out o f tow n
344 Cem etery lots and crypts
52052 Truck rental
68032 C em etery lots and crypts
323
Automobile parking
346 Services by practitioners other than
physicians
324
Vehicle tolls
326
59011 New spapers
59021 M agazines
363
327
328
331
34021 Babysitting services
67031 D aycare and nursery school
Legal services, excludingclosing costs for
purchase of real estate
68011 Legal fees
M em bership dues and fees
62011 C lub m em bership dues and fees
368
L essons or instructions in g o lf, sw im m ing,
p ia n o , dancing, crafts, hobbies
350 V ideo cassettes, tapes, and discs
62041 Fees for lessons or instructions
31022 V ideo cassettes and discs, blank and
prerecorded
351
fixed fees
333
365
349 Child daycare services and nursery school
Elementary and high schooltuition and
67021 Elementary and high school tuition
and fixed fees
34041 Gardening and law ncare services
348 Babysitting services
College textbooks
66011 C ollege textb ook s
364 Garden or law n services
34011 P ostage
Intracity mass transit
53031 Intracity mass transit
330
347 P ostage
Intercity bus fares
53021 Intercity bus fares
D om estic h ouseh old services
34031 D om estic services
34071 C are o f invalids, elderly, and
convalescents in the h om e
56041 Services by other m edical professionals
Airline fares
53011 Airline fares
N ew spapers: com bined single cop ies and
subscriptions (377)
362 M agazines: com bined single cop ies and
subscriptions (378)
52053 Parking fees
52054 Vehicle tolls
B ook s purchased through b ook clubs and
sets o f reference b o o k s
336 Funeral services
51011 Automobile finance charges
315
355
334 Bank services
61022 P hotographic and darkroom supplies
352 N ew m otorcycles
45031 N ew m otorcycles
354
371
F lash b u lb s/cu b es, darkroom supplies, and
other p hotographic supplies
N ursing and convalescent h om e care
57022 N ursing and convalescent hom e care
206
Intercity train fares
53022 Intercity train fares
375
Individual b ook s not purchased through
clubs
59023 B o o k s not purchased through b ook
clubs
376 Photographers
62051 Photographer fees
CPI Appendix 7.
Non-POPS Sample Designs
For each non-POPS entry level item (electricity, for
example), the following information is given below:
1.
Source of the universe data
2.
Sampling unit for outlets
3.
Measure of size
4.
Desired final pricing unit
5.
Number of designated outlets and designated
quotes.
1.
Schools reported for college tuition in the Pointof-Purchase Survey.
2.
Schools reported for college tuition in each
sample area.
3.
Expenditures reported for college tuition.
4.
Specific housing fee for the college.
5.
Outlets, 136; quotes, 136.
c pi
b.
c.
d.
2.
3.
5.
Outlets, 260; quotes, 1,040.
Department of Energy publication: Typical
Electric Bills-January 1, 1984.
Department of Energy publication: Financial
Statistics o f Selected Electric Utilities, 1982.
American Public Power Association 1986
Directory.
Consumer Expenditure Survey (CE).
Telephony’s Directory and Buyers Guide.
Companies providing local telephone service in
each c p i sample area.
3.
Number of residential customers.
4.
Specific service such as main station costs, addi
tional message units, extension costs, etc.
5.
Outlets, 660; quotes, 660.
1.
The distribution of household mail by type o f
postal service, by postal zone as determined by the
Household Mailstream Study, Final Report, pre
pared for the U.S. Postal Service.
2.
U.S. Postal Service.
3.
Postal revenue for each type o f service and postal
zone.
4.
Specific postal service and postal zones traveled.
5.
Outlets, 140; quotes 140.
34021 Babysitting
Electric utility companies serving each of the 91
c p i sample areas or electric utility companies
reported in the CE Survey.
Annual revenue from sales of electricity to resi
dents of the respective sample areas or expend
itures reported for electricity in the CE Survey.
1.
2.
34011 Postage
26011 Electricity
a.
Specific type of service and specific number o f
cubic feet or therms o f gas.
27011 Local telephone charges
21031 Housing while at school
1.
4.
1.
Federal mimimum wage.
2.
State mimimum wage.
3.
Outlets, 7; quotes 250.
35011/50011 Insurance—-auto and tenants
1.
Data file o f insurance companies obtained from
A.M . Best Data Center.
4.
Specific type of service for a specific number of
kilowatt hours.
2.
Insurance companies serving the States in which
the c pi sample areas are located.
5.
Outlets, 604; quotes, 1,208.
3.
Total revenue for noncommercial policies by type
o f insurance.
4.
Specific policy within the sample area.
5.
Auto: Outlets, 342; quotes, 486.
Tenants: Outlets, 388; quotes, 388.
26021 Utility natural gas
1.
a.
b.
2.
3.
Brown’s Directory o f North American and
International Gas Companies.
Consumer Expenditure (CE) Survey.
Gas (or electricity and gas) companies serving
each of the 91 C P I sample areas or gas utility com
panies reported in the CE Survey.
46011 Used cars
1.
International.
Annual revenue from sales of natural gas to
residents of the respective sample areas or ex
penditures reported for natural gas in the CE
Survey.
1985/86 Survey and analysis o f Business Car
Policies and Costs, published by Rungheimer
2.
207
Selection of used car prices in The Official Used
Car Trade-In Guide, Published by National Auto
mobile Dealers Association.
CPI Appendix 7.
Non-pops Sample Designs— Continued
3.
Total sales of used cars from the business and
government sectors to the consumer sector.
4.
Specific used cars equiped with specific options.
5.
Outlets, 348; quotes 348.
53011 Airline fares
52011/52013/52014 State vehicle registration,
driver’s license, and State vehicle inspection
1.
Digest o f Motor Laws.
2.
State motor vehicle departments in each
ple area.
3.
Revenue generated by the fees.
4.
Specific class/vehicle registration, type of license,
or inspection service.
5.
State vehicle registration: outlets, 88; quotes,
CPI
b.
3.
Number of nonbusiness passengers per airline, per
trip itinerary, per fare class.
A specific trip itinerary and fare class for the
selected airline.
Outlets, 348; quotes, 348.
RusselVs Official National Motor Coach Guide.
2. Bus companies serving each of the 91 CPI sample
areas.
3. Number of trips per week.
4. Specific trip (origin and destination) and class of
service.
5. Outlets, 140; quotes, 140.
1.
53022 Intercity train
1.
Data file of intercity train trips provided by
Amtrak and the Alaskan Railroad.
Total toll revenue for each toll facility.
2.
Amtrak and the Alaskan railroad.
Specific toll fee for a specified use of the
facility.
3.
Number of tickets sold.
4.
Specific trip and class.
Outlets, 53; quotes, 53.
5.
Outlets, 2; quotes 64.
2.
All toll facilities in the United States.
3.
4.
5.
All airlines providing service from the 91 CPI sam
ple areas.
53021 Intercity bus
52054 Tolls
Toll Facilities in the U.S., the Federal
Highway Administration,
Highway Statistics 1982, the Federal Highway
Administration.
2.
5.
88.
a.
Civil Aeronautics Board data file consisting of a
10-percent sample of all passenger itineraries
originating in the United States.
4.
sam
Driver’s license: Outlets, 8; quotes, 8.
State vehicle inspection: Outlets, 8; quotes, 8.
1.
1.
208
Appendix A. Seasonal
Adjustment Methodology at BLS
An economic time series may be affected by regular
intrayearly (seasonal) movements which result from cli
matic conditions, model changeovers, vacation practices,
holidays, and similar factors. Often such effects are large
enough to mask the short-term, underlying movements
of the series. If the effect of such intrayearly repetitive
movements can be isolated and removed, the evaluation
of a series may be made more perceptive.
Seasonal movements are found in almost all economic
time series. They may be regular, yet they do show varia
tion from year to year and are subject to changes in pat
tern over time. Because these intrayearly patterns are
combined with the underlying growth or decline and
cyclical movements of the series (trend-cycle) and also
random irregularities, it is difficult to estimate the pat
tern with exactness.
More than a half-century ago, attempts were made to
isolate seasonal factors from time series. Some early
methods depended upon smoothing curves by using
personal judgment. Other formal approaches were
periodogram analysis, regression analysis, and correla
tion analysis. Because these methods involved a large
amount of work, relatively little application of seasonal
factor adjustment procedures was carried out.
In the mid-1950’s, new electronic equipment made
more elaborate approaches feasible in seasonal factor
methods as well as in other areas. Using a computer, the
Bureau of the Census developed seasonal factors based
on a ratio-to-moving-average approach. This was a major
forward step, as it made possible the uniform applica
tion of a method to a large number of series at a relatively
low cost.1 Subsequent improvements in methods and in
computer technology have led to more refined procedures
which are both faster and cheaper than the original
technique.
The Bureau of Labor Statistics began work on seasonal
factor methods in 1959. Prior to that time, when addi
tional data became available and seasonal factors were
generated from the lengthened series, the new factors
sometimes differed markedly from the corresponding fac
tors based on the shorter series. This difference could
affect any portion of the series. It was difficult to accept
1 Julius Shiskin, E lectronic C o m pu ters a n d Business Indicators,
Occasional Paper No. 57 (New York, National Bureau of Economic
Research, 1957).
a process by which the addition of recent information
could affect significantly the seasonal factors for periods
as much as 15 years earlier, especially since this meant
that factors could never become final. The first b l s
method, introduced in 1960, had two goals: First, to
stabilize the seasonal factors for the earlier part of the
series; second, to minimize the revisions in the factors
for the recent period.
Since 1960, the Bureau has made numerous changes
and improvements in its techniques and in methods of
applying them. Thus far, all the changes have been made
within the scope of the ratio-to-moving-average or difference-from-moving-average types of approaches. The b l s
1960 method, entitled “ The b l s Seasonal Factor
M ethod,” was further refined, with the final version
being introduced in 1966. It was in continuous use for
many Bureau series (especially employment series based
on the establishment data) until 1980. In 1967, the Bureau
of the Census introduced “ The X - ll Variant of the
Census Method II Seasonal Adjustment Program ,” bet
ter known as simply X - l l. The X - ll provided some
useful analytical measures along with many more options
than the b l s method. Taking advantage of the X - l l ’s
additional flexibility, BLS began making increasing use
of the X - ll method in the early 1970’s, especially for
seasonal adjustment of the labor force data based on the
household survey. Later in the 1970’s, Statistics Canada,
the Canadian national statistical agency, developed an
extension of the X -l 1 called “ The X - ll a r i m a Seasonal
Adjustment Method.” The X - ll a r i m a provided the
option of using a r i m a (Autoregressive Integrated Mov
ing Average) modeling and forecasting techniques to
extrapolate some extra data at the end of a time series
to be seasonally adjusted. The extrapolated data help to
alleviate the effects of the inherent limitations of the mov
ing average techniques at the ends of series. After exten
sive testing and research showed that use of X -l 1 a r im a
would help to further minimize revisions in factors for
recent periods, b l s began using the X - ll a r i m a pro
cedure in 1980 for most of its official seasonal
adjustment.
The standard practice at BLS for current seasonal
adjustment of data as it is initially released is to use pro
jected seasonal factors which are published ahead of time.
The time series are generally run through the seasonal
adjustment program once a year to provide the projected
209
factors for the ensuing months and the revised seasonally
adjusted data for the recent history of the series, usually
the last 5 or 6 years. It has generally been unnecessary
to revise any further back in time because the programs
which have been used have all accomplished the objec
tive of stabilizing the factors for the earlier part of the
series, and any further revisions would produce only
trivial changes. For the projected factors, the factors for
the last complete year of actual data were selected when
the X - ll or the bls method programs were used. With
the X -l 1 a r im a procedure, the projected year-ahead fac
tors produced by the program are normally used. For the
labor force data since 1980, only the first 6 months of
factors projected from the annual run are used—a special
midyear run of the program is done, with up-to-date data
included—to project the factors for the remaining 6
months of the year.
The alternative to the use of projected factors is con
current adjustment, where all data are run through the
seasonal adjustment program each month, and the cur
rent observation participates in the calculation of the
current factor. Of course, the concurrent approach
precludes the prior publication of factors and requires
substantially more staff and computer resources to run,
monitor, and evaluate the seasonal adjustment process.
However, recent research has shown potentially signifi
cant technical advantages in the area of minimization of
factor revisions that are possible with concurrent adjust
ment. If future findings suggest the desirability of a
change to a concurrent procedure or to some other type
of methodology, such a change will be seriously con
sidered in consultation with the Government’s working
group on statistics.
In applying any method of seasonal adjustment, the
user should be aware that the result of combining series
which have been adjusted separately will usually be a lit
tle different from the direct adjustment of the combined
series. For example, the quotient of seasonally adjusted
unemployment divided by seasonally adjusted labor force
will not be quite the same as when the unemployment rate
is adjusted directly. Similarly, the sum of seasonally ad
justed unemployment and seasonally adjusted employ
ment will not quite match the directly adjusted labor force.
Separate adjustment of components and summing of
them to the total usually provides series that are easier
to analyze; it is also generally preferable in cases where
the relative weights among components with greatly dif
ferent seasonal factors may shift radically. For other series,
however, it may be better to adjust the total directly if high
irregularity among some of the components makes a good
adjustment of all components difficult.
Finally, it is worth noting that the availability of a fast,
efficient procedure for making seasonal adjustment com
putations can easily lead to the processing of large
numbers of series without allotting enough time to review
the results. No standard procedure can take the place of
careful review and evaluation by a skilled analyst. A review
of all results is strongly recommended. And it should also
be remembered that, whenever one applies seasonal fac
tors and analyzes seasonally adjusted data, seasonal
adjustment is a process which estimates a set of not
directly observable components (seasonal, trend-cycle,
irregular) from the observed series and is, therefore, sub
ject to error. Because of the complex nature of methods
such as X -ll a r im a , the precise statistical properties of
these errors are not yet known.
Technical References
Barton, H. C., Jr. “ Adjustment for Seasonal Variation,”
Federal Reserve Bulletin, June 1941.
The classic account of the frb ratio-to-moving-average
method, in which the analyst uses skilled judgment to draw
freehand curves at key stages of the procedure.
Dagum, Estela Bee. The X - ll a r im a Seasonal Adjustment
Method. Ottawa, Statistics Canada, January 1983
(Statistics Canada Catalogue No. 12-564E).
Macaulay, Frederick R. The Smoothing o f Time Series,
nber No. 19. New York, National Bureau of Economic
Research, 1931.
An early discussion of moving averages and of the
criteria for choosing one average rather than another.
Organization for Economic Cooperation and Development.
Seasonal Adjustm ent on Electronic Computers. Paris,
1961.
The report and proceedings of an international confer
ence held in November 1960. Describes experience in the
United States, Canada, and several European countries.
Includes theoretical sections relating to calendar (trading
day) variation and general properties o f moving averages.
Shiskin, Julius. Electronic Computers and Business Indicators,
Occasional Paper No. 57. New York, National Bureau of
Economic Research, 1957. Also published in Journal o f
Business, Vol. 30, October 1957.
Describes applications of the first widely used computer
program for making seasonal adjustments.
U.S. Department of Commerce, Bureau of the Census. The
X - ll Variant o f the Census M ethod II Seasonal A djust
ment Program. Technical Paper No. 15 (1967 revision).
Bureau of the Census. Seasonal Analysis o f Economic Time
Series, Economic Research Report, ER-1, issued December
1978.
Proceedings of a 1976 conference jointly sponsored by
the National Bureau of Economic Research and the Bureau
of the Census.
U.S. Department of Labor, Bureau of Labor Statistics. Employ
ment and Earnings, January 1988.
Bureau of Labor Statistics.
1966.
210
The
bls
Seasonal Factor Method,
Appendix B. Industrial
Classification
bls and other Federal and State agencies follow as
closely as possible a single system to define and classify
industries in the U.S. economy. The Office of Manage
ment and Budget, in the Executive Office of the Presi
dent, publishes the Standard Industrial Classification
Manual (sic) based on principles set forth by a technical
group made up of Government and industry experts. The
Bureau of Labor Statistics participated in the initial
development of the classification and continues to work
with the Office of Management and Budget and other
agencies in seeking to improve it. The manual is revised
periodically to reflect the economy’s changing industrial
composition and organization.
Three basic principles were followed in developing the
sic: (1) the classification should conform to the existing
structure of American industry; (2) each establishment
is to be classified according to its primary activity; and
(3) to be recognized as an industry, the group of estab
lishments constituting the proposed classification must
be statistically significant in the number of persons em
ployed, the volume of business done, and other measures
of economic activity.
As there are thousands of products and activities, the
SIC provides for grouping these into categories, both
narrow and broad, to enhance the value of industrial
statistics for users interested in different levels of detail.
In the 1972 edition of the sic manual, the broadest
grouping divides the economy into 11 divisions:
Agriculture, forestry, and fishing; mining; construction;
manufacturing; transportation, communications, electric,
gas, and sanitary services; wholesale trade; retail trade;
finance, insurance, and real estate; services; public
administration; and nonclassifiable establishments. At the
2-digit level, all products and services are combined into
84 “ major groups.” Thus, in the manufacturing division,
establishments engaged in manufacturing machinery,
apparatus, and supplies for the generation, storage,
transmission, transformation, and use of electrical energy
are combined into Major Group 36—Electrical and elec
tronic machinery, equipment, and supplies.
The 3-digit level provides several hundred categories.
In the electrical machinery major group, the Sic provides
eight groups of industries: Electric transmission and
distribution equipment; Electrical industrial apparatus;
Household appliances; Electric lighting and wiring equip
ment; Radio and television receiving equipment, except
communication types; Communication equipment; Elec
tronic components and accessories; and Miscellaneous
electrical machinery, equipment, and supplies.
Thousands of products and activities are distinguished
at the 4-digit level. For example, in Group 367, nine
industries are defined: Radio and television receiving type
electron tubes, except cathode ray; Cathode ray televi
sion picture tubes; Transmitting, industrial, and special
purpose electron tubes; Semiconductors and related
devices; Electronic capacitors; Resistors, for electronic
applications; Electronic coils, transformers, and other
inductors; Connectors, for electronic applications; and
Electronic components, not elsewhere classified.
The Bureau classifies reports from survey respondents,
usually based on an establishment concept, according to
their primary product or activity. The sic is used in the
same way by the agencies supplying the Bureau with
universe lists and benchmark data. Hence, a high degree
of orderliness and consistency is attained, which benefits
not only the users of all bls establishment statistics, but
also the users of all Government figures.
An establishment is defined as an economic unit,
generally at a single physical location, where business is
conducted or where services or industrial operations are
performed (for example: A factory, mill, store, hotel,
movie theater, mine, etc.).
Where separate economic activities are performed at
a single physical location (such as construction activities
operated out of the same location as a lumber yard), each
activity should be treated as a separate establishment
wherever (1) no one industry description in the classifica
tion includes such combined activities; (2) the employ
ment in each such economic activity is significant; and
(3) reports can be prepared on the number of employees,
their wages and salaries, sales or receipts, and other
establishment type data.
For activities such as construction and similar physi
cally dispersed operations, establishments are represented
by those relatively permanent main or branch offices,
terminals, stations, etc., which are either (1) directly
responsible for supervising such activities, or (2) the base
from which personnel operate to carry out these activities.
Hence, the individual sites, projects, fields, networks,
lines, or systems of such dispersed activities are not
ordinarily considered to be establishments.
An establishment is not necessarily identical with an
211
enterprise or company, which may consist of one or more
establishments. Also, it is to be distinguished from sub
units, departments, or divisions. Supplemental interpreta
tions of the definition of an establishment are included
in the industry descriptions of the Standard Industrial
Classification where appropriate.
Beginning with the 1972 sic manual, the classification
system was changed so that all establishments primarily
engaged in the same kind of economic activity are now
classified in the same 4-digit industry, regardless of the
type of ownership. Hence, their owners may include
such diverse organizations as corporations, partnerships,
individual proprietors, government agencies, joint ven
tures, etc.
After a lengthy review process, the manual was revised
again in 1987 to reflect the technological changes that had
occurred in the economy as well as some institutional
changes such as the deregulation of the transportation
(airlines and trucking) and banking industries. Also
included in the manual revision were additional industry
breakouts for the services sector of the economy, which
has experienced significant expansion during the past 1015 years.
Although the 1987 manual revision was effective
January 1,1987, implementation will begin in 1988. The
Bureau’s current plans are for the Employment and
Wages program (the ES-202 program) to be converted
to the 1987 sic manual with the submittal of the first
quarter 1988 ES-202 report. That quarter’s report will
be submitted on both classification systems to measure
the employment shifts caused by the change in the
classification system. The ES-202 program was selected
to be the first to be revised since data from this program
are used as the sampling frame for many of the Bureau’s
directly collected surveys as well as many of the
Federal/State cooperative surveys. In addition, employ
ment data from this program are used to establish
industry employment benchmarks for the Current
Employment Statistics program and as a key element in
the estimation component of other Bureau programs.
Other bls data series will be converted to the 1987
classification system at a later date.
Technical References
Hostetter, S. “ The Verification Method as a Solution to
the Industry Coding Problem,” Proceedings o f the Sec
tion on Survey Research Methods. American Statistical
Association, Washington, DC, 1983.
ards . 1977 Supplement-Standard Industrial Classifica
tion Manual.
Office of Management and Budget, Statistical Policy Division.
Standard Industrial Classification Manual, 1972.
Statistical Policy Office, Office of Information and Regulatory
Affairs, and Office of Management and Budget. A
Review o f Industry Coding Systems. Statistical Policy
Working Paper II, 1984.
Office of Management and Budget and U. S. Department of
Commerce, Office of Federal Statistical Policy and Stand
U.S. Department of Labor, Bureau of Labor Statistics,
Coding Interpretations Manual, April 1984.
212
sic
Appendix C. Geographic
Classification
The geographic detail for which bls publishes data
varies with the scope and size of the surveys it undertakes.
In addition to national summaries, the Bureau publishes
data for four regions; individual States, the District of
Columbia, and outlying areas (Puerto Rico, Guam, and
the Virgin Islands); Metropolitan Statistical Areas
(MSA’s); Labor Market Areas ( l m a ’s); individual cities;
and other area designations developed to meet specific
survey objectives. (See table C -l.)
bls
regions
For survey estimates and indexes (including estimates
of the civilian labor force and unemployment, Area Wage
Surveys,1Employment Cost Index, productivity surveys,
and the Consumer Price Index), bls generally uses a
four-region classification system2 as follows:
Northeast: Connecticut, Maine, Massachusetts, New
Hampshire, New Jersey, New York, Pennsylvania,
Rhode Island, Vermont;
Midwest: Illinois, Indiana, Iowa, Kansas, Michigan,
Minnesota, Missouri, Nebraska, North Dakota, Ohio,
South Dakota, Wisconsin;
South: Alabama, Arkansas, Delaware, District of Co
lumbia, Florida, Georgia, Kentucky, Louisiana, Mary
land, Mississippi, North Carolina, Oklahoma, South
Carolina, Tennessee, Texas, Virginia, West Virginia;
West: Alaska, Arizona, California, Colorado, Hawaii,
Idaho, Montana, Nevada, New Mexico, Oregon, Utah,
Washington, Wyoming.
Data for the Industry Wage Surveys are published for
nine regions.3
Alaska and Hawaii are not covered in the Area Wage Surveys.
This classification is the same as the four regions used by the
Bureau of the Census.
3 New England: Connecticut, Maine, Massachusetts, New Hamp
shire, Rhode Island, Vermont; Middle Atlantic: New Jersey, New
York, Pennsylvania; Border States: Delaware, District of Columbia,
Kentucky, Maryland, Virginia, West Virginia; Southeast: Alabama,
Florida, Georgia, Mississippi, North Carolina, South Carolina, Ten
nessee; Southwest: Arkansas, Louisiana, Oklahoma, Texas; Great
Lakes: Illinois, Indiana, Michigan, Minnesota, Ohio, Wisconsin; Mid
dle West: Iowa, Kansas, Missouri, Nebraska, North Dakota, South
Dakota; Mountain: Arizona, Colorado, Idaho, Montana, New Mex
ico, Utah, Wyoming; Pacific: Alaska, California, Hawaii, Nevada,
Oregon, Washington.
1
2
Metropolitan Statistical Areas
Metropolitan Statistical Areas are designated by the
Office of Management and Budget through the Federal
Committee on Standard Metropolitan Statistical Areas.
bls is represented on this committee along with other
organizations.4
The Office of Management and Budget has changed
the official title, “ Standard Metropolitan Statistical
Area,” to “ Metropolitan Statistical Area.”
The general concept of a metropolitan statistical area
is one of a large population nucleus together with adja
cent communities which have a high degree of economic
and social integration with that nucleus.
Metropolitan statistical areas are relatively “ free-stand
ing” and not closely associated with other metropolitan
statistical areas. These areas are typically surrounded by
nonmetropolitan counties. Areas qualifying for recogni
tion as metropolitan statistical areas have either a city
with a population of at least 50,000 or a Bureau of the
Census urbanized area of at least 50,000 and a total
metropolitan statistical area population of at least
100,000 .
Each metropolitan statistical area has one or more cen
tral counties, containing the area’s main population con
centration. A metropolitan statistical area may also
include outlying counties which have close economic and
social relationships with the central counties. Such coun
ties must have a specified level of commuting to the cen
tral counties and must meet certain standards regarding
metropolitan character, such as population density. In
New England, metropolitan statistical areas are composed
of cities and towns, rather than whole counties. Under
specified conditions, two adjacent areas may be con
solidated or combined into a single metropolitan
statistical area.
Each metropolitan statistical area has at least one cen
tral city. The titles of metropolitan statistical areas include
up to three central city names, as well as the name of each
State into which the metropolitan statistical area extends.
4 The other organizations include the Employment and Training
Administration of the Department of Labor, the Department of Housing
and Urban Development, the Bureau of the Census, the Federal Reserve
Board, the Department of Agriculture, and the Department of Transpor
tation. The committee is chaired by a representative of the Office of
Management and Budget.
213
Each metropolitan statistical area is categorized in one
of the following levels based on total population:
Level A - Metropolitan Statistical Areas of 1 million
or more.
Level B - Metropolitan Statistical Areas of 250,000
to 1 million.
Level C - Metropolitan Statistical Areas of 100,000
to 250,000.
Level D - Metropolitan Statistical Areas of less than
100, 000 .
Areas assigned to Levels B, C, or D are designated as
metropolitan statistical areas. In areas with over 1 million
population (Level A), primary metropolitan statistical
areas may be identified. These areas consist of a large
urbanized county, or cluster of counties, that demon
strates very strong internal economic and social links, in
addition to close ties to neighboring areas. When primary
metropolitan statistical areas are defined, the large area
of which they are components is designated a con
solidated metropolitan statistical area.5
5 Federal Register, Vol. 45, No. 2, Jan. 3, 1980, pp. 956-63.
Labor market areas
A labor market area ( l m a ) is defined by the Bureau
of Labor Statistics as a geographic area consisting of a
central community and contiguous areas which are
economically integrated into that community. Within a
labor market area, workers generally can change jobs
without relocating. BLS defines l m a ’s in terms of entire
counties, except in New England where cities and towns
are used, l m a ’s are categorized as either “ major,” which
is usually coterminous with a metropolitan statistical area,
or as “ small.” A “ small” labor market area is defined
as a county or group of counties with a central community
of at least 5,000 population and which meets