The full text on this page is automatically extracted from the file linked above and may contain errors and inconsistencies.
Volume I
Handbook
of Methods
U.S. Department of Labor
Bureau of Labor Statistics
December 1982
Bulletin 2134-1
Volume I
BUS
Handbook
o f M (§ ffii]® ^ g
U.S. Department of Labor
Raymond J. Donovan, Secretary
Bureau of Labor Statistics
Janet L. Norwood, Commissioner
December 1982
Bulletin 2134-1
For sale by the Superintendent of Documents, U.S. Government Printing Office
U.S. GOVERNMENT PRINTING OFFICE
Washington, D.C. 20402 - Price $6.50
: 1982
0 -3 8 1 -6 0 8
(3 8 7 6 )
The BLS Handbook o f Methods presents in two vol
umes detailed explanations of how the Bureau of Labor
Statistics obtains and prepares the economic data it
publishes. Volume I contains this information for all
bls programs except the Consumer Price Index.
Volume II contains information for the cpi .
B ls statistics are used for many purposes, and
sometimes data well suited to one purpose may have
limitations for another. This edition o f the Handbook,
like its predecessors, aims to provide users o f bls data
with the information necessary to evaluate the suitabili
ty o f 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 ad
ditional technical information are given at the end of
most chapters.
The Handbook was written by members of the staffs
of the various BLS 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 may, with appropriate credit, be reproduced
without permission.
Contents
Volume I:
Introduction.........................................................................................................................
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 survey...................
Occupational employment statistics...................................................................
Measurement of unemployment in States and local areas .................................
Employment and wages covered by unemployment insurance............................
3
13
25
28
32
Prices and Living Conditions
6.
7.
8.
Consumer expenditures and income...................................................................
Producer prices....................................................................................................
International price indexes...................................................................................
38
43
62
Wages and industrial Relations
9.
10.
11.
12.
Occupational pay and supplementary benefits..................................................
Negotiated wage and benefit changes.................................................................
Employment cost in d e x .......................................................................................
Employee benefit plans........................................................................................
67
74
78
88
Productivity and Technology
13.
14.
15.
16.
Productivity measures: Business economy and major sectors...........................
Productivity measures: Industries and the Federal Government.....................
Technological change..........................................................................................
Foreign labor statistics and trade m onitoring....................................................
93
99
109
114
Occupational Safety and Health
17.
Occupational safety and health statistics............................................................
122
Economic Growth and Employment Projections
18.
19.
20.
21.
Labor force projections......................................................................................
Economic growth studies.....................................................................................
National industry-occupational matrix...............................................................
Occupational outlook..........................................................................................
135
137
142
145
Appendix:
A.
B.
C.
Seasonal adjustment methodology at bls ........................................................
Industrial classification ....................................................................................
Geographic classification..................................................................................
147
150
152
Volume II. Consumer Price Index
Introduction
When U.S. Commissioner of Labor Carroll Wright
issued his first annual report in March 1886, he
established the policy of explaining his statistical
methods to his readers and of seeking to avoid misinter
pretation of the figures presented. During the 96 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 in
struct them in the proper 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 ap
proach to its data collection activities also are discussed
briefly.
nish technical advice and assistance to State agencies
and other cooperating organizations. An important
aspect of the work of the regional staffs has been ex
plaining the concepts 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 in
fluence, and sometimes are crucial to, the determining
and shaping of public policy. Over the years, it has
developed a staff of professional analysts, trained to
search out the implications of survey findings for the
welfare of workers and to present them as cogently and
as promptly as possible in written and oral form. How
successfully this can be accomplished depends greatly
upon the competence of the analysts and their sup
porting personnel.
In bls , analytical and statistical work is performed by
economists, statisticians, and mathematical statisticians
with the aid of an experienced corps of programmers,
systems analysts, and other professionals. For analytical
work, economists at even the lowest grade level must
meet requirements roughly equivalent to a college major
in economics. There are comparable requirements for
other professionals. Great efforts are made to hire the
best qualified persons available.
The Bureau provides training needed for on-the-job
skills, as background to special assignments, to keep
professionals abreast of changes in their fields, and to
aid higher level and executive professionals in obtaining
the best results from their staffs. In training staff, a
special effort is made to impart detailed knowledge of
the techniques used in collecting and compiling the
statistics, so that maximum application of the data to
current problems can be made without a risk of ex
ceeding the limits of their significance.
SIS r® §
l<
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, bls statistics are often
quite specialized, yet they meet general economic and
social data requirements. As the needs of users are likely
to differ, no statistic is ideal for all. This makes it im
portant that the characteristics of the measures and their
limitations be well understood.
Organization
The statistical programs of the Bureau were
developed, for the most part, independently of each
other, taking on characteristics suited to the re
quirements of the subject under observation. As a
result, the Bureau was organized according to subjectmatter areas, an arrangement which has proved efficient
and has been continued over the years. Expertise in
techniques, economic analysis, and other staff ac
tivities 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 fur
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, and
academic organizations and from members of the
public. Over the years, the advice the Commissioner of
Labor Statistics has received on policy and technical
1
matters from responsible parties, relating to the collec
tion and analysis of Bureau statistics, has been very
helpful. Of course, decisions on statistical policy have
always been the final responsibility of the Commis
sioner.
In order to keep in touch with the current and an
ticipated needs of business and labor groups and to seek
advice on technical problems, the Commissioner
established 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,
although not council members, have special com
petence. The councils may take formal action through
resolutions or recommendations on appropriate mat
ters, but such resolutions are merely advisory. Members
of the councils and the subcommittees serve in their in
dividual capacities, not as representatives of their
organizations.
The members of the Business Research Advisory
Council are designated by the Commissioner under
authorization of the Secretary of Labor, after nomina
tion 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 of Labor
Statistics under authorization of the Secretary of Labor,
from nominations by the Director of Research, afl -cio .
All research directors of international unions
represented in the afl -cio are invited to attend the
general meetings of the council. The council provides
general direction to the advisory activities of trade
union research directors in relation to the Bureau.
The Bureau often seeks the advice of professional
economists, statisticians, social scientists, educators,
and others, either in their individual capacities or as
members of professional organizations. This is most
likely to occur when a conceptual or theoretical question
arises which is considered fundamental to the work of
the Bureau in a specialized field, and where professional
acceptance of the Bureau’s work in that field may be
reinforced by the findings of an independent analyst.
thousands of firms and individuals to provide informa
tion closely related to their daily affairs and their per
sonal lives. To some who have supplied the desired in
formation, the Bureau has gone back often for later in
formation on the same subject or for new types of infor
mation. The response has been remarkable in its
generosity. In no small measure, the cooperation
received is due to the great care taken to avoid identify
ing the firm or the person supplying the information.
The fact that Bureau employees pledge themselves to
protect these data is less important than that they have a
deep understanding of the adverse long-run conse
quences of even a single lapse. They are aware of the
greater worth, in terms of pure statistical validity, of the
information provided voluntarily compared with that
supplied under legal sanctions. The only inducement
employed 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 im
plemented by combining the data reported by the dif
ferent sources and issuing the findings in summary
form.
Another assurance given respondents is that their
reports will be used for statistical purposes only. At
tempts to “ break” this policy, by organizations or in
dividuals who wanted access to the data and were will
ing to go to the courts to secure it, have been success
fully resisted.1 A similar problem occurs when an ad
ministrative agency of government seeks court action to
compel a company to release its file copy of information
provided in confidence to a statistical agency.2
While it cannot be proved that these policies result in
more reliable statistics, Bureau Commissioners and
their staffs over the years have been convinced from ex
perience that it is so. It is notable that some other
Federal agencies (especially the Bureau of the Census),
well equipped with authority to compel the submittal of
certain reports, rarely if ever invoke this power. Rather,
they also choose to rely on persuasion. The Bureau of
Labor Statistics, while its functions as a statistical agen
cy are prescribed by law,3has always relied upon volun
tary cooperation of respondents in collecting informa
tion.
1 For example, see Norwegian Nitrogen Company v. United States, 288 U .S.
294; United States v. Kohler, 13 Fed. Rules Serv. 33.333 (E .D ., Pa., 1949);
Hawes v. Walsh, 111 Fed. 569, the Court o f Appeals for the District o f Colum
bia. In all o f these cases, the courts sustained the policy o f protecting the con
fidentiality o f information given voluntarily and in confidence to an agency of
the Federal Government.
2 See Supreme Court o f the United States, St. Regis Paper Company, Peti
tioner, v. United States, N o. 47, October term, 1961.
3 Excerpts from 29 U .S.C . 1, acts o f June 27, 1884, ch. 127, 23 Stat. 60; June
13, 1888, ch. 3 8 9 ,1 ,2 5 Stat. 182; Feb. 14,1903, ch. 552,4, 32 Stat. 826; Mar. 18,
1904, ch. 716, 33 Stat. 136; Mar. 4, 1913, ch. 141, 3, 37 Stat. 737.
Voluntary reporting and confidentiality
Voluntary reporting and the preserving of the con
fidential nature of reported data are important
characteristics of bls programs. Over the course of
almost a century, the Bureau has asked hundreds of
2
Chapter 1. Labor Force, Employment,
and Unemployment from the
Omrrainit Population Sunrwty
sample survey of households, called the Monthly Report
of Unemployment, initiated by the Works Progress Ad
ministration 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 ti
tle was changed once more in 1948 to the present Cur
rent 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
cps labor force data was transferred to bls , although
the Bureau of the Census continued to collect and
tabulate the statistics.
Each month, the Bureau analyzes and publishes
statistics on the labor force, employment, unemploy
ment, 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 (CPS), which is conducted by
the Bureau of the Census for the bls . This monthly
survey of the population is conducted using a scien
tifically 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 persons. 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 govern
ments, began experimenting 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 in
dividual. In most of the surveys, the unemployed were
defined as those who were not working but were “ will
ing 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 inter
pretation 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 classifica
tion of an individual depended principally on his actual
activity within a designated time period; i.e., was he
working, looking for work, or engaged in other ac
tivities? These concepts were adopted for the national
Description ®f Surwey
The CPS provides statistics on the civilian noninstitu
tional population 16 years of age and over. Figures on
the Armed Forces (obtained monthly from the Depart
ment of Defense) are added to the cps estimates to
derive estimates of the “ total labor force” and the
“ total 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 work
ing. The institutional population, which is also excluded
from coverage, consists of inmates of penal and mental
institutions, sanitariums, and homes for the aged, in
firm, and needy.
The CPS is collected each month from a probability
sample of approximately 60,000 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 so short that the occurrence of holidays or other
3
since 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
unemployment are the mean and the median. Mean
duration is the arithmetic average computed from single
weeks of unemployment. Median duration is the mid
point 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) lob
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 per
sons who never worked at a full-time job lasting 2 weeks
or longer.
accidental events might cause erratic fluctuations in the
information 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 follow
ing week, which is the week containing the 19th day of
the month.
C @ G i< s @ p fts
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 un
paid 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 were seeking other jobs. 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. In
cluded 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 per
sons whose only activity consisted of work around their
own home (such as housework, 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 total labor force, in addition, includes members of
the Armed Forces stationed either in the United States
or abroad.
Unemployment rate. The unemployment rate 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 characteristics. Because their
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 overall
unemployment rate.
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 register
ing at a public or private employment agency, writing
letters of application, canvassing for work, etc.—are
considered to be looking for work. Also included as
unemployed are those who did not work at all during
the survey week, were available for work, and (a) were
waiting to be called back to a job from which they had
been laid off, or (b) were waiting to report to a new
wage or salary job scheduled to start within 30 days.
Duration of unemployment represents the length of
time (through the current survey week) during which
persons 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
N ot in labor force. 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 school” includes persons attending school
4
during the survey week who had new jobs to which they
were scheduled 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 attending school—who had new jobs not scheduled
to begin until after 30 days (and who were not working
or looking 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”
rotation groups, those which had been in the sample for
3 previous months and would not be in for the subse
quent 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.)
in existence at the time of the 1970 census constituted a
separate p s u . Outside s m s a ’s , counties normally are
combined, except where the geographic area of the
single county is excessive. By combining counties to
form p s u ’s , greater heterogeneity is accomplished.
Moreover, another important consideration is to have
the p s u sufficiently compact in area so that, with a small
sample spread throughout, it can be efficiently canvass
ed without undue travel cost. A typical primary sam
pling unit, for example, includes urban and rural residents
of both high and low economic levels and encompasses,
to the extent feasible, diverse occupations and in
dustries.
The p s u ’s are grouped into strata. Among these
p s u ’ s , the largest s m s a ’s and a small number of the
other areas which are not s m s a ’s are separate strata
representing themselves. In general, however, a stratum
consists of a set of p s u ’s as much alike as possible in
various characteristics such as geography, population
density, rate of growth in the 1960-70 decade, propor
tion of blacks and other minorities, principal industry,
number of farms, and so on. Except for the selfrepresenting p s u ’s , each of which is a complete stratum,
the strata are established so that their sizes in terms of
1970 population are approximately equal. Where a p s u
is a stratum by itself, it automatically falls in the sam
ple. The rest of the 629 sample p s u ’s are selected from
the remaining strata in a random manner in such a way
that their probability of selection is proportionate to
their 1970 population. For example, within a stratum,
the chance that a p s u with a population of 50,000 would
be selected for the sample is twice that for a unit having
a population of 25,000.
Sampling
The CPS national sample is located in 629 areas com
prising 1,148 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 60,000 are occupied and thus eligible for inter
view. The remainder are units found to be vacant, con
verted to nonresidential use, containing persons who
reside elsewhere, 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
temporarily absent, refuse to cooperate, or are
unavailable for other reasons. Information is obtained
each month for approximately 120,000 individuals 16
years and over.
The description of the sampling design and selection
that follows is based on the procedures used in the
redesign following the 1970 Census of Population. The
present sampled households used each month, updated
by information on new construction, are derived from
this design. The survey is expected to be redesigned,
based on the 1980 census, by January 1985, and these
plans are described briefly in the last section of this
chapter.
Selection o f sample households. The sample design calls
for a sampling ratio which depends on the predeter
mined total sample size. At present, the sampling ratio
is roughly 1 household for every 1,500 households in
each stratum. The sampling ratio is modified slightly
each month, as the size of the sample is held relatively
constant despite the overall growth of the population.
The sampling ratio used within each sample PSU
depends on the proportion that the population of the
sample area was of the stratum population at the time of
the 1970 census. In a sample area which was one-tenth
of the stratum, the within-PSU sampling ratio that
results is 1 in 150, thereby achieving the desired ratio of
1 in 1,500 for the stratum.
With each of the 629 sample p s u ’s , the number of
households to be enumerated each month is deter
mined by the application of the within-PSU sampling
ratio rather than through the assignment of a fixed
quota. This procedure makes it possible to reflect, on a
current basis, population changes within the sample
area. Consequently, the sample as a whole properly
reflects the changing distribution of the population and
Selection o f sample areas. The entire area of the United
States consisting of 3,146 counties and independent
cities is divided into 1,931 primary sampling units
( p s u ’s). With some minor exceptions, a p s u consists of a
county or a number of contiguous counties. Each of the
238 Standard Metropolitan Statistical Areas ( s m s a ’s) 1
1 See appendix C.
5
avoids the distortion which would result from the ap
plication of fixed quotas of households or persons based
on the population at an earlier date.
Within each designated psu , several stages of sam
pling may be used in selecting the units to be
enumerated. The first step is the selection of a sample of
census enumeration districts (ed ’s), which are ad
ministrative units used in the 1970 census and contain,
on the average, about 300 households. These are
selected systematically from a geographically arranged
listing so that the sample ed ’s are spread over the entire
PSU. The probability of selection of any one ED is pro
portionate to its 1970 population.
The next step is to select a cluster of approximately
four households to be enumerated within each
designated ED. This selection is made wherever possible
from the list of addresses for the ED compiled during the
1970 census or, if the addresses are incomplete or inade
quate, 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 consist of as geographically con
tiguous addresses as possible.
This list sample is supplemented by a selection of the
appropriate proportion of units newly constructed in
the psu since the census date. The addresses of these
units are obtained mainly from records of building per
mits in that area. A special procedure is also followed to
include units in the sample that had been missed in the
census. In those enumeration districts where area sam
pling methods are used, mainly rural areas, the ed ’s are
subdivided into segments; that is, small land areas hav
ing 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
designated for the sample; the probability of selection is
proportionate to the estimated size of the segment.
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 interviews at
all housing units in the segment but uses a systematic
sampling pattern to achieve the equivalent of a fourhousehold cluster which is canvassed completely. The
remaining housing units in the segment are then
available for further samples.
consecutive months 1 year, leaves the sample during the
following 8 months, and then returns for the same 4
calendar months of the next year. In any 1 month, oneeighth of the sample segments are in their first month of
enumeration, another eighth are in their second month,
and so 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 month-to-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
household members, including their personal
characteristics (date of birth, sex, race, ethnic origin,
marital status, educational attainment, veteran status,
etc.) and their relationship to the household head. 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. Approximately 60 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
economic 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
during 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 work
Rotation o f sample. Part of the sample is changed each
month. For each sample, eight systematic subsamples
(rotation groups) of segments are identified. A given
rotation group is interviewed for a total of 8 months,
divided into two equal periods. It is in the sample for 4
6
no information was obtained because of absence, im
passable roads, refusals, or unavailability of the
respondent for other reasons. This adjustment is made
separately by combinations of sample areas within each
State and the District of Columbia, and within these,
for six groups—two race categories (white, and black
and other) within three residence categories. For sample
areas which are smsa ’s, these residence categories are
the central cities, and the urban and the rural balance of
the smsa ’s. For other sample areas, the residence
categories are urban, rural nonfarm, and rural farm.
The proportion of sample households not interviewed
varies from 4 to 5 percent depending on weather, vaca
tions, etc.
ing during 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 had started to look for work, (3)
the length of time they had been looking for work, (4)
whether they were seeking 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 nonparticipation, and
jobseeking intentions of individuals not in the labor
force. In 1979, questions were added to collect data on
hourly and weekly earnings from a quarter of the sam
ple 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
omissions, 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
transmitted to the computers in the Bureau of the Cen
sus’ Washington office where they are checked for com
pleteness and consistency.
Although the interviewers on the CPS are chiefly parttime 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 ques
tionnaires, repeated observation during enumeration,
and a systematic reinterview of part of their assignments
by the field supervisory staff, the work of the inter
viewers is kept under control and errors or deficiencies
are brought directly to their attention.
2.
Ratio estimates. The distribution of the popula
tion selected for the sample may differ somewhat, by
chance, from that of the population as a whole, in such
characteristics as age, race, sex, and residence. Since
these characteristics are closely correlated with labor
force participation and other principal measurements
made from the sample, the survey estimates can be
substantially improved when weighted appropriately by
the known d istrib u tio n of these p o p ulation
characteristics. This is accomplished through two stages
of ratio estimates as follows:
a. First-stage ratio estimate. In the CPS, a portion of
the 629 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 procedure is performed at two
geographic levels: First, by the four Census regions
(Northeast, North Central, South, and West), and
second, for each of the 46 States which contain nonsam
ple areas. The procedure corrects for the differences
that existed at the time of the 1970 census between the
distribution by race and residence of the population in
the sample areas and the known race-residence distribu
tion in the portions of the census region or State
represented by these areas. The regional adjustment is
performed by metropolitan-nonmetropolitan residence
and race, while the State adjustment is done by urbanrural status and race.
Estimating Methods
The cps estimation procedure involves weighting the
data from each sample person. The basic weight, which
is the inverse of the probability of the person being in
the sample, is a rough measure of the number of actual
persons that the sample person represents. The basic
weights are then adjusted for noninterview, and the
ratio estimation procedure is applied.1
b. Second-stage ratio estimate. The sample propor
tions in the categories described below are adjusted to
the distribution of independent current estimates of the
population in the same categories. The second-stage
1.
Noninterview adjustment. The weights for all in ratio estimate is performed in order to increase the
reliability of the estimates and is carried out in two
terviewed households are adjusted to the extent needed
steps. In the first step, the sample estimates are adjusted
to account for occupied sample households for which
7
within each State and the District of Columbia to an in
dependent control for the population 16 years and over
for the State. The second step of the adjustment is ap
plied to all sample persons and is a weighting to nation
wide independent population estimates within 68 agesex-race groups. The entire second-stage ratio estima
tion procedure is iterated six times, each time beginning
at the weights developed the previous time. This itera
tion ensures that the sample estimates both of State
population and of the national age-sex-race categories
will be virtually equal to the independent population
estimates.
The controls by State for the civilian noninstitutional
population 16 years and over are an arithmetic ex
trapolation 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.
Prior to January 1982, the independent national con
trols used for the age-sex-race groups in the final step of
the second stage ratio adjustment were obtained from
the “ inflation-deflation” method. This procedure in
flated the most recent census counts to include the
estimated net census undercount 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 were then “ deflated” to census level to reflect
the pattern of net undercount in the most recent census
by age, sex, and race. The actual percent change over
time in the population in any age group was preserved.
Introduced into the cps estimation procedure in
January 1982, the independent population controls are
computed by carrying forward the April 1, 1980, total
population (including Armed Forces overseas) by age,
race, and sex, taking account of the subsequent aging of
the population, fertility, mortality, and net immigra
tion.
The CPS sample returns (taking into account the
weights determined after the first stage of ratio
estimates) are, in effect, used to determine only the per
cent distribution within a given age-race-sex group by
employment status and other characteristics. In
developing absolute numbers, these percentage distribu
tions are multiplied by the independent population
estimate for the appropriate age-race-sex group.
Composite estimate. The last step in the preparation of
estimates makes use of a composite estimate. In this
procedure, a weighted average of two estimates is ob
tained for the current month for any particular item.
The first estimate is the result of the two stages of ratio
estimates described above. The second estimate consists
of the composite estimate for the preceding month to
which has been added an estimate of the change in each
item between the preceding month and the present
month, based upon that part of the sample which is
common to both months (75 percent). Although the
weights for the two components of such a composite
estimate do not necessarily have to be equal, in this in
stance the weights used for combining these two
estimates are each one-half. Equal weights in this case
satisfy the condition that for virtually all items there will
be some gain in reliability over the estimation procedure
after the first two stages of ratio estimates.
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
unemployment insurance. It is a comprehensive source
of information on the personal characteristics such as
age, sex, race, Hispanic origin, educational attainment,
and the marital and family status of the total civilian
population (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 ag
gregate or average hours for an industry, permitting
separate analyses of part-time workers, workers on
overtime, etc. It is a comprehensive current source of in
formation on the occupation of workers, whether
teachers, stenographers, 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 considerd to be a “ labor reserve.” The
characteristics of such persons—whether married
women with or without young children, disabled per
sons, students, retired workers, etc.—can be determin
ed. Also, through special inquiries, it is possible to ob
tain information on their skills and past work ex
perience.
Each month, a significant amount of basic informa
tion about the labor force is analyzed and published in
Employment and Earnings. The detailed tables in this
report provide information on the labor force, em
ployment, and unemployment by a number of
characteristics, such as age, sex, race, marital status, in
dustry, 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 N a tio n ’s m etropolitan and
nonmetropolitan areas, special data for Vietnam-era
veterans, etc., are published every quarter. Addition
ally, data are published quarterly on employment and
unemployment by family relationship and on median
weekly earnings broken down by a variety of
characteristics. Approximately 250 of the most impor
tant estimates from the CPS are presented each month
on a seasonally adjusted basis.2
The CPS is used also for a program of special inquiries
to obtain detailed information from particular
segments, or for particular characteristics of the popula
tion 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 earn
ings 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 attainment of workers. Suveys have
been made periodically on subjects such as job mobility,
job tenure, job-search activities of the unemployed, and
the intensity of the job search.
Generally, the persons who provide information for
the monthly c p s questions also answer the supplemental
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
published as news releases and subsequently in the
Monthly Labor Review. Reprints of the articles,
together with technical notes and additional tables have
been published as Special Labor Force Reports.
In addition to the regularly tabulated statistics
described above, special data can be generated through
the use of the c p s individual record (micro) tapes. These
tape files contain records of the responses to the survey
questionnaire for all individuals in the survey. While the
tapes can be used simply to create additional crosssectional 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.
Limitations
Geographic. The c p s is designed to produce reliable
monthly national estimates. A sample which could pro
duce monthly estimates for all States as reliable as those
now published for the Nation would have to be about as
large as the present national sample in each State. Sub
national data derived from the c p s are published
monthly for 10 large States and annually for all States,
30 large s m s a ’s , and selected central cities. The produc
tion 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
particularly large in the case of persons with marginal
attachments 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 con
tinuous 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 uni
formity in the statistics may result. Similarly, the data
are limited by the adequacy of the information possess
ed by the respondent and the willingness to report ac
curately.
The estimates from the survey also are subject to
various other types of errors. Some of these are:
2 Since 1980, the X -ll a rim a seasonal adjustment method has
been used to seasonally adjust labor force data. For a detailed descrip
tion of the X -ll a rim a method, see Estela Bee Dagum, The X - ll
arima Seasonal Adjustment Method, Statistics Canada Catalogue
No. 12-564E, February 1980.
Nonresponse—about 4 to 5 percent of occupied units
are not interviewed in a typical month because of tem
9
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 processing the completed questionnaires, en
tries usually are supplied for omitted items on the basis
of the distributions in 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
underenumeration of certain population groups or er
rors 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, immigra
tion, 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
processing errors are almost inevitable in a large
statistical operation 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 er
rors 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
differences 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 aris
ing 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 prac
tice, 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 c p s data, thus permitting the
10
user to take this factor into account in interpreting the
data. In general, the smaller figures and small dif
ferences between figures are subject to relatively large
variation and should be interpreted with caution.
Estimation of response bias is one of the most dif
ficult aspects of survey and census work. In many in
stances, available techniques are not sufficiently precise
to provide satisfactory estimates. Continuing ex
perimentation is carried out with the aim of developing
more precise measurements and improving the overall
accuracy of the series.
Planned CHuaim
gss f©r tins CPS
Several changes are scheduled to be made in the
design and content of the c p s in the near future. These
will result primarily from a planned redesign of the c p s
sample based on information collected in the 1980
decennial census.
Redesign of the CPS sample
Since the inception of the c p s in 1940, it has been the
practice to update and revise the sample after each
decennial census. At the same time, improvements in
sampling methodology or survey procedures are in
troduced.
As 1980 census data become available, several revi
sions in the sample design and survey procedures will be
introduced into the CPS. Consistent with procedures
following past censuses, new sample areas for the CPS
will be introduced which, for the most part, will remain
in the c p s sample until it is redesigned after the 1990
census. In the redesign, currently underway, the sam
pling frame is also being changed. Heretofore, the c p s
has been designed as a national sample, with its goal to
provide the best estimates of employment and
unemployment for the United States as a whole. During
the 1970’s, however, growing demands were placed on
the c p s for the development of State and local labor
force estimates used in the allocation of Federal
revenues to States and areas and for other purposes.
The c p s was selectively expanded on several occasions
in recent years to improve its ability to provide State and
local labor force estimates. Even with these efforts, it is
still difficult for the CPS to provide very reliable subna
tional data except in large States and metropolitan
areas. Therefore, in an effort to provide more accurate
and reliable sub national estimates, the survey is ex
pected to be designed as 51 separate samples for each of
the States and the District of Columbia. At the same
time, the redesign is expected to improve the statistical
reliability of the national estimates. As of mid-1982, the
plan envisioned that the new samples will be introduced
into the CPS on a flow basis during July-December 1984
and will be fully in place in January 1985. A detailed
description of the procedures used to select the new
samples and any other changes in sampling procedures
will be available at that time.
arrive at the measure known as the total labor force.
This change was recommended by the National Com
mission on Employment and Unemployment Statistics
in their final report, Counting the Labor Force.
The other scheduled change will be the introduction
of a revised occupational and industry classification
system into the labor force data. The new occupational
classification system, which was used in the 1980 census
and is based on the Standard Occupational Classifica
tion system issued in 1977 and revised in 1980,
represents the most substantial revision in occupational
classification since 1940. Occupational data for 1982
based on both the old and new systems are to be made
available to help users make the transition to the new
system. Only minor changes are planned in the industry
classification system.
Content changes
In addition to the sample redesign, two changes are
scheduled to be introduced into the CPS in early 1983.
The first is that U.S.-based Armed Forces will be includ
ed in the official national labor force estimates. As in
dicated earlier in the section on concepts, present labor
force statistics encompass only the civilian noninstitutional population. Data on members of the Armed
Forces stationed either in the United States or abroad
are reported separately and are not included in the
employment totals or in the civilian labor force base
used to calculate the unemployment rate. The Armed
Forces, however, are added to the civilian labor force to
Technical References
Report 463 and Current Population Reports, Series
P-23, No. 62, 1976.
A concise description of the methodology used in
obtaining labor force information from sample house
holds. Labor force concepts and definitions are set
forth. The adequacy of labor force data and quality
controls are discussed, and major improvements in the
cps are listed chronologically.
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 o f all Federal statistical series on employ
ment and unemployment and a comparison of the
sources and uses o f each series; includes a brief history
of the development of labor force statistics, an evalua
tion o f concepts and techniques, and recommendations
for further research and improvements.
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 pre
sent the sampling errors for labor force series.
Bureau o f 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.
U .S. Department of Commerce, Bureau o f the Census. “ The
Current Population Survey Reinterview Program, Jan
uary 1961 through December 1966,” Technical Paper
No. Ip, 1968.
A summary of procedures and results of the cps
reinterview program.
Bureau o f Labor Statistics. How the Government Measures
Unemployment, Report 505, 1977.
A short, nontechnical discussion o f the concepts and
methods used in obtaining labor force statistics from
the Current Population Survey.
Bureau o f the Census. “ The Current Population Survey:
Design and M ethodology,” Technical Paper No. 40,
1978.
A comprehensive description of the cps , based on the
design following the 1970 census.
Bureau of Labor Statistics. Technical Description o f the Quar
terly Data on Weekly Earnings from the Current Pop
ulation Survey, Bulletin 2113, 1982.
U.S. Department of Commerce, Office of Federal Statistical
Policy and Standards. “ An Error Profile: Employment
as Measured by the Current Population Survey,” Sta
tistical Policy Working Paper 3, 1978.
A description of the potential sources of error in the
cps as the affect the national employment statistics.
A description of the collection, processing, and reli
ability o f the weekly and hourly earnings data obtained
from the cps .
Bureau o f Labor Statistics. Using the Current Population
Survey as a Longitudinal Data Base, Report 608, 1980.
A discussion of the potential and problems in using
cps labor force data for purposes of longitudinal
analysis.
U.S. Department of Labor, Bureau of Labor Statistics, and
U.S. Department o f Commerce, Bureau o f the Census.
Concepts and Methods Used in Labor Force Statistics
Derived from the Current Population Survey, bls
11
18. LINE NUMBER
20. Did . . . do any work at all
21. ( I f J in 19, skip to 21A.)
22. ( I f LK In 19, Skip to 22A.)
24. INTERVIEWER CHECK ITEM
LAST WEEK, not counting
19. What was. . . doing most of
LAST WEEK -
D id . . . have a job or
Has. . . been looking for work
work around the house?
(Note: I f farm or business
operator in hh., ask about
unpaid work)
^
business from which he/she
during the past 4 weeks?
Yes O
Working
Keeping house
No O
(Go to 21)
20A. How many hours
did . . . work
Going to school
or something else?
LAST WEEK
at all jobs?
Working (Skip to 2 0 A )___ WK
O
With a job but not at work .. J
O
Looking for work ................ LK
O
Keeping house......................... H
O
Going to school.......................S
20B. INTERVIEWER
CHECK ITEM
O
. OT
\
3 3
4- 45 5
G G
? ?
item 23)
O
O
Other (Specify).............
IX
88
88
(Go to 20C)
9 9
O
Unable to work (Skip to 2 4 ) .. U
Retired.......................................R
0 0
3 5 -4 8
O
2)
fjD
(Go to 2j
20D. Did . . . loss any time or
take any time o ff LAST
WEEK for any reason
was temporarily absent or
on layoff LAST WEEK?
‘
Yes
O
No
O
^Yes
Yes
O
What is the reason . . .
USUALLY works less
than 35 hours a week?
(Mark the appropriate reason)
Slack w o r k .................................
||
On vacation . . . .
O
Checked
_
O
Bad weather. . . .
pvt. employ, agency
O
New job to begin
within 30 days O
Material shortage........................
O
Plant or machine repair.............
O
work? Was it because . . . lost
New job started during week . . .
O
Holiday (Legal or religious).........
I
g
^
Labor dispute...............................
(Correct 20A and 20B as
necessary i f extra hours
not already included and
skip to 23.)
ago was. . . laid
S
O
On vacation.................................
Too busy with housework.
school, personal bus., etc. . .
at this job?
off?
9 9
part-timework?
Full O
(Skip to 23 and enter job
held last week)
Yes O
O
Did not want full-time work . . . O
Full-time
work week under 35 hours.. O
Other reason (Specify)................
O
\
(Skip to 23 and enter job worked
at last week)
Part O
0
( Already has a jo b ____ O
O
O
(. Other (Specify in notes)
O
C
D
E
F
G
H
J
K
L
M
Ref.
0 N o
I P o
8 Q o
3 R o
4- S o
5 T o
G U o
V o
w o
X o
Y o
z o
O
0 0
O
© Other pers. handicap in finding job
O
O
8 8
9 9
2 5 D . H ow much does . . .
U S U A L L Y earn per week
© Can't arrange child c a re ............
at this jo b B E F O R E
H |
O
deductions? Include any
overtime pay, commissions,
Family responsibilities....................
•
O
In school or other train ing .............
O
O
© Other (Specify in notes) ..................
or tips usually received.
O
O
0 0 0
III
888
3 3 3
4-4-45 5 5
( Month)
One to five years ago ........................
More than 5 years ago ......................
O
O
Nev. worked full-time 2 wks. or more O
Never worked at a l l.............................
O
(SKIP to 23. I f layo ff entered in 21A, enter
Job, either fu ll or part time, from which laid
off. Else enter last fu ll time civilian job
lasting 2 weeks or more, or "never worked.')
5 5
(A sk 25D )
© Don't k n o w .....................................
Within last 12 months (^ e c /fy ).........
G G
O
o III health, physical disability...........
weeks or more?
GG
? ?
8 8
9 9
O
3 3
O
O
job or business lasting 2 consecutive
Ref.
l
o Couldn't find any w o r k .................
o Lacks nec. schooling.
training, skills or experience . . .
o Employers
think too young or too old.........
r
—
22F. When did . . . last work at a full-time
0
I
8
3
4*
5
G
?
8
9
Cents
3 3
4- 4*
5 5
ra
LJ
°
0
I
8
3
45
G
?
8
9
(Skip to 25D )
25C . H o w much does . . . earn
per hour?
for work?
(Mark each reason mentioned)
22E. Is there any reason why . . . could
not take a job LAST WEEK?
No O
O
O
O
O
O
o
o
o
o
o
o
o
(Go to 25C)
O
22D. Has . . . been looking for full-time or
r— j
OFFICE USE ONLY
A
B
O
No
24D. What are the reasons. . . is not looking
| Going to school...........
0 0
I I
2 8
3 3
‘h
3 5
G 6
? ? ?
88 8
0 9 9
O
Yes
8 8 88
^ I Temporary illness . . . .
O
O
Dollars
8
O
Own illness.................................
O
O
0 0
?
35 hours or more a week
O
25B . I s . . . . paid b y th e hour
on this job?
_
G o
?
No
O
- Bad weather.................................
How many extra
hours did . . . work?
O
O
^
looking for work?
9
8 8
3 3
3) How many weeks
6
?
9
O
^
ago did . . . start
21C. Does. . . usually work
G
?
O
^
c
3
4-
full- or part-time?
^
g
0
i i
8 8
0 55
24C. Does. . . want a regular job n
I
2) How many weeks
Yes O
O
^
has. . . been
o ff LAST WEEK?
O
Could find only part-time work . O
O
looking for work?
0
3
4*
O th er..............................................
O
------- z
this job?
E3
Slack work or business conditions
Temporary
noriseasonal job completed . . .
Unsatisfactory work
arrangements (Hours, pay, etc.)
O
^
d o e s . . . U S U A L L Y w o rk at
Retirement or old age....................
O
Quit jo b ............................
1) How many weeks
1 or 5 (Go to 25A)
2 5 A . H ow many hours per week
Seasonal job completed................
Wanted temporary work
O
Job terminated during week . . .
*
or quit a job at that time (pause)
22C.
°1
°
0
°J
O
0\
O/
2 .3 ,4 , 6, 7, or 8 (E nd questions)
O
H ea lth ............................................
Lost jo b .............................
21B. Is . . . getting wages or
salary for any of the time
O
Personal, family
(Inci. pregnancy) or school . . . .
or was there some other reason?
\
First digit o f S E G M E N T
24B. Why did . . . leave that job?
n
Left school........................
one job LAST WEEK?
Yes O
Never worked.............
O
22B. Why did . . . start looking for
overtime or at more than
O
5 or more years ago ..
O
Nothing (Skip to 24 )........................
Other (Specify in notes, e.g., CETA,
union or prof, register, etc.).........
20E. Did . . . work any
O
4 up to 5 years ago. . .
O
Placed or answered ads....................
Temporary layoff
(Under 30 days) O
\ (Skip
r to
Indefinite layoff
(30 days or more or O
2 up to 3 years ago. . .
3 up to 4 years ago . . .
O
friends or relatives. .
(Skip to
22B and
22C2)
1 up to 2 years ago . . .
O
employer directly . .
O
Labor dispute. . .
.
.
pub. employ, agency O
□
2 . 3 ,4 , 6, 7, or 8 (End questions)
24A. When did . . . last work for pay at a
regular job or business, either full- (
part-time?
Within past 12 months
work LAST WEEK?
Own illness.........
How many hours
(Correct 20A i f lost time
not already deducted;
I f 20A reduced below 35,
correct 20B and fill 20C;
otherwise, skip to 23.)
No O
O
number is:
Other (Specify in notes) .
hours LAST WEEK?
O
(Go to 24)
take off?
What is the reason . . .
(R o ta tion num ber)
0
d id . . .
worked less than 35
No
/
O
2 5 .IN T E R V IE W E R C H E C K IT E M
First dig it o f S E G M E N T number is:
4 weeks to find work? (Mark all
methods used; do not read list.)
Other (Specify).
or slack work?
Yes O
No
j
------------
( R otation num ber)
(Go to 22) 22A. What has. . . been doing in the last
21A. Why was . . . absent from
such as illness, holiday
20C. Does. . . USUALLY work 35
hours or more a week at this job?
O
s
24E. Does. . . intend to look for work of any
kind in the next 12 months?
Y e s .....................................
;
G G G
? ? ?
O
It depends (Specify in notes) O
N o ........................................
Don't know ........... ’. .........
8 8 8
9 9 9
O
O
( I f entry in 24B, describe job in 23
(End questions)
______ otherwise, en d questions.)_________
23. DESCRIPTION OF JOB OR BUSINESS
23 A. For whom did . . . work? (Name o f company, business, organization or other employer.)
23E. Was this person
An employee of PRIVATE Co ,
bus., or individual for wages, salary or comm. . . .
23B. What kind of business or industry is this? (For example: TV and radio mfg., retail shoe store, State Labor Dept., farm.)
2 3 F . IN T E R V IE W E R
C H E C K IT E M
P O
A FEDERAL government employee........................... F O
A STATE government employee................................. S O
/
(Go to
’ 23F)
Entry (or NA)
in item 20A
O
A LOCAL government employee................................. L O
23C. What kind of work was . . . doing? (For example: electrical engineer, stock clerk, typist, farmer.)
Self-empl. in OWN bus., prof, practice, or farm
/Y e s .................. I O
Is the business incorporated? < ..
. ,
.
\ No (or farm) . . SE O
23D. What were. . .'s most important activities or duties? (For example: types, keeps account books, files, sells cars, operates
printing press, finishes concrete.)
Working W ITHOUT PAY in fam. bus. or farm .........WP O
NEVER W ORKED......................................................NEV O
Pcjp 4
V
Entry (or NA)
12
(Go to 25
a t top
o f Page)
in item 21B O
(End
• ques
tions)
All other cases O
(£ n d
.
questions)
Chapter 2. Em ploym ent,
Hours, and Earnings from
the Establishment Survey
existed. This confusion stimulated efforts to develop
comprehensive estimates of total wage-and-salary
employment in nonagricultural industries, and, in 1936,
b l s survey data produced such a figure for the first
time.
Interest in employment statistics for States and areas
also grew. Even before b l s 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 add
ed by 1936. By 1940, estimates of total nonagricultural
employment for all 48 States and the District of Colum
bia were available.
Since 1949, the Current Employment Statistics (CES)
program has been a fully integrated Federal-State pro
ject which provides employment, hours, and earnings
information by industry on a national, State, and area
basis, b l s 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 na
tional, State, and area levels, from collection and pro
cessing of data through estimation and publication. In
1981, cooperative arrangements were in effect with all
50 States, the District of Columbia, Puerto Rico, and
the Virgin Islands.
Bls 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 1981, this sample included approximately 166,000
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 produc
tion or nonsupervisory workers; hours and earnings
data include average hourly earnings, average weekly
hours, and average weekly overtime hours. For many
series, seasonally adjusted data are also published.
Background
The first monthly studies of employment and payrolls
by BLS began in 1915 and covered four manufacturing
industries. Before 1915, the principal sources of
employment data in the United States were the census
surveys—the decennial Census of Population and the
quinquennial Census of Manufactures. No regular
employment data were compiled between the Census
dates.
In 1916, the survey was expanded to cover employ
ment and payrolls in 13 manufacturing industries; by
1923, the number had increased to 52, and by 1932, 91
manufacturing 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 in
crease 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 nonmanufac
turing categories.
During the Great Depression, there was controversy
concerning the actual number of unemployed people;
no reliable measures of employment or unemployment
Goimespts
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 activi
ty. 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. In the
collection of data on employment, payrolls, and hours,
BLS usually requests separate reports by establishment.
13
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 nonproduc
tion worker categories in manufacturing is shown on the
facsimile of the b l s 790 C shedule 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. Exclud
ed from this category are executive and managerial per
sonnel, 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 the ser
vice industries), data are collected for nonsupervisory
workers. Nonsupervisory workers include most
employees except those in top executive and managerial
positions. (See facsimile of b l s 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 an
nually. The basic source of benchmark data for the Cur
rent 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
nonagricultural payrolls are covered by the ui system.
The compilation and use of benchmark data are ex
plained in detail in later sections of this chapter.
However, when a company has more than one establish
ment engaged in the same activity in a geographic area,
these establishments may be covered by a combined
report.
Employment
Employment represents the total number of persons
employed full or part time in nonagricultural
establishments during a specified payroll period. Tem
porary 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 col
lecting employment data from business establishments.
However, national employment figures for Federal
Government establishments represent the number of
persons who occupied positions on the last day of the
calendar month; intermittent workers are counted if
they performed any service during the month.
Workers on an establishment payroll who are on paid
sick leave (when pay is received directly from the
employer), 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 duplication 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 proprietors, 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 includ
ed 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 supervisors,
who are engaged directly in the manufacture of the
product of the establishment. Among those excluded
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 forceaccount construction. (Force-account construction is
construction work performed by an establishment,
primarily engaged in some business other than construc
H@urs and earnings
The hours and earnings series are based on reports of
gross payrolls and the corresponding paid hours for
production workers, construction workers, or non
supervisory workers. (See facsimile of b l s 790 C.) (In
government and private educational institutions,
payroll data are for “ all employees.” )
Gross payrolls include pay before deductions for
social security, unemployment insurance, group in
surance, withholding tax, 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 payment in
kind are not included.
Total hours during the pay period include the hours
worked, overtime hours, hours paid for standby or
reporting time, and equivalent hours for which
14
their major product or activity as determined by the
establishments’ 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 classified in accordance with the Standard
Industrial Classification Manual, Office of Manage
ment and Budget, 1972, as modified by the 1977 Supple
ment. (See appendix B of this bulletin for a description
of this system.)
employees received pay directly from the employer for
sick leave, holidays, vacations, and other leave. Over
time or other premium pay hours are not converted to
straight-time equivalent hours. Total hours differ from
scheduled 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 dif
ferential, 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.
Gross average hourly earnings series, derived by
dividing gross payrolls by total hours, reflect the actual
earnings of workers, including premium pay. They dif
fer from wage rates, which are the amounts stipulated
for a given unit of work or time. Gross 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 gross average weekly
earnings by means of the Bureau’s Consumer Price In
dex, indicates the changes in the purchasing power of
money earnings as a result of changes in prices for con
sumer 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 in
come, the extension and incidence of various social serv
ices and benefits, and the duration and extent of
employment and unemployment. The long-term trends
of these earnings data are also affected by changing
mixes of full-time/part-time workers, high-paid/lowpaid workers, etc.
Straight-time average hourly earnings are approx
imated by adjusting gross average hourly earnings by
eliminating 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, late-shift work, and premium overtime rates
other than at time and one-half.
Data Sources and Collection Methods
Sample data
Each month, the State agencies cooperating with b l s
in the survey collect data by mail on employment,
payrolls, and hours paid for, from a sample of
establishments. 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 ( b l s form 790 series),
that is, one which is submitted each month by the
respondent, edited by the State agency, and returned to
the respondent 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. All aspects of the schedule—its format, the
wording of the requested items and definitions, and the
concepts embodied therein—are subjected to a continu
ing 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 consistency 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
operational 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 process is carried out in accordance with detailed
instructions from b l s . The State agencies use the infor
mation provided on the forms to develop State and area
Industrial classification
Industrial classification refers to the grouping of
reporting establishments into industries on the basis of
15
estimates of employment, hours, and earnings, and for
ward the data, either on the schedules themselves or in
machine-readable form, to BLS-Washington. At b l s ,
they are edited again by computer to detect processing
and reporting 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 in
formation for “ all employees” has been the periodic
tabulations 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
BLS-Washington each quarter. (See chapter 4.)
For the few industries exempt from mandatory ui
coverage, other sources are used for benchmark infor
mation. Data on employees covered under social securi
ty laws, published by the Bureau of the Census in Coun
ty 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, ui data on private elementary and secon
dary school employment are augmented by data from
the National Catholic Welfare Association for the
number of members of religious orders who teach in
such schools. Employment figures for religious
organizations are obtained from data provided by the
National Council of Churches and surveys of churches
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 b l s in the Current Employment
Statistics survey to collect data in most industries, since
full coverage would be prohibitively costly and time
consuming. The sampling plan for the program must:
(a) Provide for the preparation of reliable monthly
1 For a more detailed description o f the benchmarks, see Carol M. Utter and
John B. Farrell, “ BLS Establishment Estimates Revised to March 1980
Benchmarks,” Employment and Earnings, July 1981, pp. 7-13.
16
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
considerable industry detail for metropolitan areas,
States, and the Nation; (c) be appropriate for the ex
isting framework of operating procedures, ad
ministrative practices, resource availability, and other
institutional characteristics 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 distribution by stratifying the universe of
establishments into homogeneous groups. The strata are
arranged according to industry and size characteristics.
Under optimum allocation, a larger sample is usually re
quired 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 in
dustry. 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 in
dustries, establishments with 250 or more employees are
included in the sample with certainty; in many in
dustries, the cutoff is lower. In a manufacturing in
dustry 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. Consequently, the sample design for such in
dustries provides for a complete census of the large
establishments with only a few chosen from among the
smaller establishments. On the other hand, in an in
dustry where a large proportion of total employment is
in small establishments, the sample design calls for the
inclusion of all large establishments, and also for a
substantial number of the smaller establishments. Many
industries in the trade and service divisions fall into this
category.
This sample design, although aimed primarily at
meeting the needs of the national program, provides a
technical 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 sam
ple. 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
national sampling ratios in order to obtain a sufficient
sample. The additional reports needed for State and
area samples are added to the sample required by the na
tional design.
Estimating Procedures
Employment
To obtain “ all employee” estimates for a basic
estimating 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. (See earlier sections on bench
marks.)
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.
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 mulitplied 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
calculated from the sample data are modified using the
wedging technique described under “ Estimating
Procedures—Hours and earnings.”
The estimates for each type of series (all employees,
production workers, and women workers) for in
dividual basic estimating cells are summed to obtain the
corresponding 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
developing benchmarks, they are not available until at
least 16 months after the benchmark month (usually
March). For example, the revised estimates based on the
March 1980 benchmarks were released in August 1981.
The revision period extended from April 1979 through
July 1981.
To determine the appropriate revisons, the new
benchmarks for March are compared to the estimates
for that month based on the previous benchmarks. The
differences represent estimating errors that accumulated
since the previous benchmark revision. These dif
ferences 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 to the new benchmark level. Estimates for
women workers and production workers are recom
puted using the revised all-employee estimates.
Although most national all-employee series are ad
justed by this wedging technique, in some cases the c e s
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. National nonagricultural payroll employment
estimates, by industry division, as a percent of the benchmark
for 1978,1979, and 1980
Industry division
1979
1980
T o ta l...........................................
99.3
99.5
100.1
Mining......................................................
Construction...........................................
Manufacturing ......................................
Transportation and public utilities.........
Wholesale and retail t r a d e ...................
Finance, insurance, and real estate . . .
S ervices.................................................
Governm ent.................
17
1978
98.1
98.4
99.4
99.1
99.6
99.0
98.8
100.0
99.8
103.2
99.6
100.3
99.4
99.9
99.5
98.1
100.6
101.5
100.3
100.3
100.8
99.9
99.1
99.5
the panel of establishments reporting in the sample is
not fixed from_month to month, there may be differ
ences between Xp and Xp. A final estimate for the cur
rent month, Xc, is obtained by making use of both
pieces of information:
Hours and earnings
Independent benchmarks are not available for the
hours and earnings series. Consequently, the levels are
derived directly from the c e s 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 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.
If preliminary estimates were attempted using the full
stratification pattern, there would be a number of cells
with no sample reported. Experience indicates that
estimates of hours and earnings prepared from the c e s
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 revi
sions 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, North Central, South, and West). These
estimates are used as a standard against which the
published averages are compared. If this comparison in
dicates 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.
^
The procedure reflected in this formula has the fol
lowing advantages: (1) It uses matched sample data; (2)
it tapers the estimate for the previous month (Xp) to
wards the sample average for the previous month of the
current matched sample (xp); and (3) it promotes con
tinuity by heavily favoring the estimate for the previous
month (Xp) when applying the numerical factors.
The results of die 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 tne
product of the aggregate worker hours and average
hourly earnings. The payroll and worker hour aggre
gates for industry groups and divisions represent
the sum of the aggregates for the component industries.
Average weekly hours for industry groups are obtain
ed by dividing the worker hour aggregates by the
corresponding production worker estimates. Average
hourly earnings for industry groups are computed by
dividing the payroll aggregates by the worker hour ag
gregates. 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 in
dustry divisions, average weekly earnings are computed
by multiplying average hourly earnings by average
weekly hours.
Average 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
establishments classified in the cell is divided by the
total number of production workers reported for the
same establishments. In computing average hourly earn
ings, 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 re
porting 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 report
ing 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
= (0.9 Xp + 0.1 xp) + (xc - xp)
Overtime hours. Average weekly overtime hours are
estimated in basically the same way as gross 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 wedging technique and the summary level
estimating technique are also comparable to those used
to estimate gross average weekly hours.
18
ing all nonagricultural payroll employment in the
private sector.2
Gross average weekly earnings in 1977 dollars. Gross
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.
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 ad
justed series are published regularly for selected employ
ment, hours, and earnings series.3
The seasonally adjusted series for 2-digit industry
categories in manufacturing and for divisions in non
manufacturing are computed by multiplying the cor
responding 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 component in
dustries. Seasonally adjusted hours and earnings
averages for broader level industry groups are weighted
averages of the component series.
Average hourly earnings, excluding overtime, fo r
manufacturing industries. These are computed by
dividing the total production worker payroll for an in
dustry 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 straighttime hours. This method excludes overtime earnings at
an assumed rate of 1 1/2 times the straight-time rates;
no further adjustment is made for other premium pay
ment 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 ef
fects of fluctuations 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 im
pact of the fluctuations and varying trends in occupa
tional employment within industries and other factors
which also influence the trends in average hourly earn
ings. Hourly earnings indexes are published for “ total
private” and the major industry divisions except
government.
Presentation
At the national level, the program produces more
than 2,600 separate published series each month. Tables
2, 3, and 4 provide a summary of the national detail
which is published currently. Table 2 describes the
primary series produced by the program, that is, those
computed directly from the sample and benchmark
data. Table 3 indicates the special series which are ob
tained from the primary series by the application of
special adjustments, while table 4 lists the seasonally ad
justed series by type and industry division.
In addition to the series published on a current
monthly basis, a single figure for employment in March
of each year (based on benchmark data) is published for
a number of industries for which monthly estimates do
not currently meet established standards for publica
tion. In 1981, following revision to the 1980 benchmark,
data for 279 such industries were published.
In September 1981, employment, hours, and earnings
statistics were available for the Nation as a whole and
for all 50 States, the District of Columbia, Puerto Rico,
the Virgin Islands, and 263 areas.4 Approximately
10,000 employment series and about 11,000 hours and
earnings series (for 214 areas) are published for these
States and areas by the State agencies. The employment
series usually cover total nonagricultural employment,
Indexes o f aggregate weekly worker hours and payrolls.
These indexes are prepared by dividing the current
month’s aggregates by the average of the monthly ag
gregates 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 172 unpublished seasonally
adjusted employment series (2-digit nonmanufacturing
industries and 3-digit manufacturing industries) cover
2 For a detailed discussion o f these indexes, see “ Introduction o f Diffusion In
dexes,” in the December 1974 issue o f Employment and Earnings.
3 See appendix A o f this bulletin for a description o f the seasonal adjustment
method.
4 Puerto Rico and Virgin Islands data are not used in making national
estimates.
19
Table 2. Number of industries for which “primary” national series are published under the BLS Current Employment Statistics
program—employment, hours, and earnings, September 1981
All
employees
Production
workers1
Women
workers
Hours and
earnings2
Average weekly
overtime hours
Total.................................................................
512
453
440
454
323
Goods producing.........................................................
Mining .........................................................................
Construction ...............................................................
Manufacturing.............................................................
1
13
15
324
1
11
15
323
1
9
15
270
11
15
323
-
Service producing.......................................................
Private service producing ...........................................
Transportation and public u tilitie s ..............................
T ra d e ...........................................................................
Finance, insurance, and real estate............................
Services.......................................................................
Government...............................................................
1
1
24
54
19
35
23
Total private.................................................................
Total nonagricultural...................................................
1
1
Industry division
1
1
22
54
19
34
12
-
1
16
52
10
23
1Production workers in manufacturing and mining; construction workers in
construction; nonsupervisory employees in all other divisions.
-
-
19
52
10
23
“
-
1
-
1
1
1
-
323
2Average hourly earnings, average weekly hours, and average weekly
earnings.
Table 3. Number of national industries for which special series are published under the Current Employment Statistics
program—employment, hours, and earnings, September 1981
Index of aggregate
weekly worker
hours
Industry division
Index of
aggregate
weekly payrolls
Total private.................................................................
1
1
Goods producing.........................................................
Mining .........................................................................
Construction ...............................................................
Manufacturing.............................................................
1
1
1
23
1
1
3
1
1
Gross average
weekly earnings
(1977 dollars)
1
1
1
23
1
1
3
1
1
Service producing.......................................................
Transportation and public u tilitie s ..............................
T ra d e ...........................................................................
Finance, insurance, and real estate............................
Services.......................................................................
Average hourly
earnings (excluding
overtime)
1
Hourly
earnings
index
1
-
-
-
1
1
1
-
23
“
1
1
1
1
1
1
1
1
1
1
1
Table 4. Number of seasonally adjusted national series published under the Current Employment Statistics program employment, hours, and earnings, September 1981
Seasonally adjusted series
Industry division
All
employees
Women
workers
Production
workers
Average
weekly
earnings
Total nonagricultural.................................................
Total p riv a te .............................................................
1
1
1
1
1
Goods producing .....................................................
Mining.......................................................................
Construction.............................................................
Manufacturing .........................................................
1
1
1
23
1
1
1
23
1
1
1
23
_
-
Service producing.....................................................
Transportation and public utilities.............................
Trade.........................................................................
Finance, insurance, and real e sta te .........................
S ervices...................................................................
Government.............................................................
1
1
3
1
1
3
1
1
3
1
1
3
1
1
3
1
1
-
_
20
Average
hourly
earnings
1
-
-
-
-
Average
weekly
hours
1
1
_
_
-
-
Worker
hour
index
1
1
1
22
1
1
1
23
_
1
1
1
-
_
3
1
-
1
1
3
1
1
-
-
Average
overtime
hours
-
_
-
_
3
_
-
_
-
major industry divisions (e.g., mining, construction,
manufacturing), and major industry groups (e.g., textile
mill products, transportation equipment, retail trade)
for each State and area. Additional industry detail is
frequently provided for the larger States and areas, par
ticularly for industries which are locally important in
the various jurisdictions.
The series on employment, hours, and earnings ap
pear in several bls publications. The summary data are
first published each month in a news release which con
tains preliminary national estimates of nonagricultural
employment, average weekly hours, and gross average
weekly and hourly earnings in the preceding month, for
major industries. The release also includes seasonally
adjusted data on employment, average weekly hours,
and average weekly overtime hours. The preliminary
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
preceding years.
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
technical developments in the program. Many of the na
tional series are also published in the Monthly Labor
Review; data are shown for each series for the most re
cent 13 months.
Following each benchmark revision, a historical
volume, Employment and Earnings, United States (or a
supplement), is published. This provides monthly data
and annual averages from the beginning date of each
series, in a few instances as far back as 1909 (the sup
plements contain revised data for recent years). A com
panion volume, Employment and Earnings, States and
Areas, provides historical data (annual averages) on all
employees and on production-worker hours and earn
ings published by the State agencies for States and areas
from the beginning of these series, in some instances as
far back as 1939. This volume (or a supplement with
revised data for recent years) is published annually.
Detailed industry data are available each month in
releases published by the cooperating State agencies.
The data are also disseminated in the publications of
other Federal agencies; e.g., the Department of Com
merce, the Board of Governors of the Federal Reserve
21
System, and the Council of Economic Advisers. They
are also regularly republished in summary form or for
specific industries in many trade association journals,
the labor press, and in general reference works.
To facilitate the use of its data, bls has made them
available to the public in machine-readable form. More
than 2,800 national ces time series are available, as well
as over 24,000 State and area series. Employment data
for 250 areas and hours and earnings data for 210 areas
are included.
Comparison with the Current
Population Survey
Total employment in nonagricultural establishments
from the ces or payroll survey is not directly com
parable 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 of the Current Population Survey, or
household survey.) 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 household survey, but are not in
cluded 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
characteristics (sex, age, race) of the labor force. The
survey is not well suited to providing detailed informa
tion on the industrial and the geographic distribution of
employment. The establishment survey, while providing
limited information on personal characteristics of
workers, is an excellent vehicle for obtaining these
detailed industrial and geographic data; in addition, it
provides hours and earnings information which is
directly related to the employment figures. The payroll
and household surveys, therefore, should be regarded as
complementary.
Uses
The series are used by business firms, labor unions,
universities, trade associations, private research
the same schedules and procedures. While the estimates
are adjusted annually to new benchmarks, changes be
tween benchmark months are not reflected in the data—
new establishments, for example, or changes in the in
dustrial 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-meansquare error ( r m se ). This measure is the standard devia
tion adjusted for the bias in the estimates:
organizations, and many government agencies to study
economic conditions and to develop plans for the
future. Business firms, for example, use the employ
ment, hours, and earnings data for guidance in plant
location, sales, and purchases. Also, firms negotiating
long-term supply or construction contracts often use
escalation clauses based on the average hourly earnings
series as an aid in reaching equitable agreements; escala
tion clauses permit an adjustment 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.
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
negotiations. They not only furnish current and
historical information on a given industry but provide
comparative data on related industries.
RMSE = \/(Standard Deviation)2 + (Bias)2
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
benchmark revisions, although the broader industry
groupings may be affected slightly by changes in the
production-worker 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-meansquare errors for employment estimates and relative
errors for hours and earnings estimates are provided in
the Technical Note in Employment and Earnings.
Reliability of Estimates
Although the relatively large size of the c es sample
assures a high degree of accuracy, the estimates derived
from it may differ from the figures that would be ob
tained if it were possible to take a complete census using
Technical References
Bureau of Labor Statistics
Current Employment Statistics State Operating Manual,
A description of the impact o f a major benchmark
adjustment and of important technical innovations
on the Current Employment Statistics series.
June 1981.
Other
Early, John F. “ Introduction of Diffusion Indexes,” Employ
ment and Earnings, December 1974.
National Commission on Employment and Unemployment
Statistics. Counting the Labor Force, 1979.
A comprehensive review and critique o f the methods
and concepts used by various Federal Government
programs providing statistics on employment, unem
ployment, and the labor force in the United States.
Farrell, John B. “ BLS Establishment Estimates Revised to
March 1981 Benchmarks,” Employment and Earnings,
June 1982.
Weinberg, Edgar, “ bls Earnings Series as Applied to Price
Escalation,” Monthly Labor Review, July 1952.
A discussion of the use of bls average hourly earn
ings series in escalation clauses in contracts.
Shiskin, Julius; Young, Allan H.; and Musgrave, John C.
The X - ll Variant o f the Census Method II Seasonal
Adjustment Program. U .S. Department of Commerce,
Bureau of the Census, November 1967.
This article was reprinted by the U.S. Department
of Commerce, National Technical Information Service,
in 1976.
Wymer, John P. “ The Revised and Expanded Program of
Current Payroll Employment Statistics,” Employment
and Earnings, November 1961.
22
Bureau of Labor Statistics and the Employ
ment and Training Administration Report on
Employment, Payroll, and Hours—Trade
U.S. Department of Labor
T h is re p o rt Is authorized by law 2 9 U .S.C. 2. Y o u r vo lu n tary coo peration is needed to m ake the results o f this survey
Form A pproved
com prehensive, accurate, and tim e ly . The in fo rm a tio n collected on this fo rm by the Bureau o f Labor Statistics and th e
O .M .B . N o. 1 2 2 0 -0 0 1 1
States cooperating in its statistical program s w ill be held In con fidence and w ill be used fo r statistical purposes only.____________A p proval expires 1 /3 1 /8 4
Return promptly, each month in the enclosed envelope which requires no postage.
BLS Codes
State
Location of Establishment(s) Covered in this Report
Report Number
Industry
Number of Establishments
City
State
County
Return to:
SAMPLE CO W
L
J
(Change nam e and m ailing address if In correct—inclu de Z ip code.)
Before entering data see explanations on reverse side.
N u m b er o f Paid Em ployees
—
Y e ar and M o n th
Pay Per cd
W h ic h i n c lu d e s
th e 12 t i o f th e
m o n th .
N o . o f fa y s p a id
f o r in c l j d i n g h o iid a y s a n d v a c a t io n s
f o r t h e m a jo r it y
o f n o n s u p e rv is o r y em p J lo y e e s .
From— Through During
the
entire
B o th dates
pay
in ch isive
period
(2 )
(3 )
(4)
Nonsupervisor / Em ployees
A l l p e rs o n s w h o w o r k e d
d u r in g o r re c e iv e d p a y f o r
a n y p a r t o f t h e p a y p e r io d .
T h e n u m b e r o f n o n s u p e r v is o r y e m r lo y e e s w h o w o r k e d
d u r in g o r re c e i ve d p a y f o r a n y p a r t o f th e p e r io d rep o r t e d , th e pa>) e a rn e d ( b e f o r e d e d j c t i o n s b u t e x c lu d in g c o m m is s io r is ) , a n d h o u r s p a id t Dr. I n c lu d e p a y a n d
h o u r s f o r o v e r im e , s ic k le a v e , h o lic fa y s , a n d v a c a tio n s .
During the Both sexes
7 day
period
which
includes
the 12th
(6)
(5 )
Women
only
(7)
Number
of
nonsupervisory
employees
Total nonsuper
visory employee
payroll (excluding
commissions
reported in col. 9A)
(omit cents)
Total nonsuper
visory employee
hours
(omit fractions)
(8 )
(9 )
(10)
1981
$
Dec.
C o m m is s io is o f n o r
v is o r y e m p o y e e s
Amount
of com
missions
(omit
cents)
super
Period i il
which earned
From— Through
B o th dates
in c h isive
(9 A )
(9 B )
(9 0
$
19 8 2
Jan.
Feb.
Mar.
Apr.
May
June
July
Aug.
Sept.
Oct.
Nov.
Dec.
D o N o t Use (O ffic e Use O n ly )
P.R.
H.
(9X)
(10X)
1981
Dec. $
198 2
Jan.
Feb.
Y o u r C o m m en ts on Changes in E m p lo y m e n t, P a yro ll, o r Wage Rates—E nter below the main factors
responsible for significant month-to-month changes in employment, average hours worked, average
hourly earnings, etc. Examples are: M ore business, te m p o ra ry sum m er help, wage rate increase, over
Expl. U P
Code Code tim e, strike , fire , weather. For any general wage rate changes (not individual changes for length of
(11) (12) service, merit, or promotion), note the amount of increase or decrease (as + 2% —5 i ) , the effective
,
date of the change, and the approximate number of nonsupervisory employees affected.
Mar.
Apr.
May
June
July
Aug.
Sept.
Oct.
Nov.
Dec.
BLS 790 E (Rev. June 1981 )
A d d itio n a l Space fo r C om m en ts Provided on Reverse.
23
Explanations for Entering Data on Reverse Side
Columns 2 and 3. Pay Period Dates—Enter the beginning and ending dates
of your pay period which includes the 12th of the month.
The term "production and related workers" excludes employees engaged
in the following activities: Executive, purchasing, finance, accounting,
legal, personnel, cafeterias, medical, professional, and technical activities,
sales, sales-delivery (e.g., delivery and route workers), advertising, credit,
collection, and in installation and servicing of own products, routine office
function, factory supervision (above the working supervisors' level); and
force account construction employees on your payroll engaged in construc
tion of major additions or alterations to the plant who are utilized as a sep
arate work force. (Employees in the above activities should be excluded
from column 8 but included in column 6, Number of Paid Employees.)
Columns 4 and 5. Pay Period—Number of Days-Enter in column 4, for
the entire pay period reported, the number of days on which the majority
of production and related workers performed work plus the number of
holidays and vacation days during the period for which the majority were
paid. When the period is longer than a week, enter in column 5 the number
of such reported days worked or paid for during the 7-consecutive day per
iod which includes the 12th of the month and falls entirely within the per
iod reported in eslumns 2 and 3.
Column 9. Payroll—Enter amount of pay earned during the pay period by
the production and related workers reported in column 8. Payrolls should
be reported before deductions for old-age and unemployment insurance,
group insurance, withholding tax, bonds, and union dues. Include pay for
overtime and for holidays, vacations, and sick leave paid directly by your
firm to employees for the pay period reported.
Column 6. Number of Paid Employees—Both Sexes—Enter the total num
ber of persons on the pay roll (s) covered by this report who worked full- or
part-time or received pay for any part of the period reported. Include sala
ried officers of corporations and executives and their staffs, but exclude
proprietors, members of unincorporated firms, and unpaid family workers.
Include persons on vacations and sick leave who received pay directly from
your firm for the period reported, but exclude persons on leave without
company pay the entire period and pensioners and members of the Armed
Forces carried on the rolls but not working during the period reported.
Exclude bonuses (unless earned and paid regularly each pay period), or
other pay not earned in pay period reported (e.g., retroactive pay), and
value of free rent, fuel, meals, or other payment in kind.
Column 10. Total Hours—Enter the sum of (1) hours worked (including
overtime hours) during the pay period by the production and related work
ers reported in column 8, (2) hours paid for stand-by or reporting time,
and (3) equivalent hours for which employees received pay directly from
the employer for holidays, vacations, sick leave, or other leave paid to
these workers. Do not convert overtime or other premium paid hours to
straight-time equivalent hours.
Column 7. Number of Paid Employees—
Women Only—Report the number
of women employees included in column 6.
Column 8. Number of Production and Related Workers—Enter the number
of production and related workers, both full- and part-time, on yourpayroll(s), whether wa ga or salaried, who worked during or received pay for
any part of the pay period reported. Include persons on vacations or on
sick leave when paid directly by your firm.
Column 10Y. Overtime Hours—Enter the number of hours included in col
umn 10, for which premiums were paid because the hours were in excess
of the number of hours of either the straight-time workday or workweek.
Include Saturday and Sunday hours (or 6th and 7th day hours) only if
overtime premiums were paid. Holiday hours worked by employees should
be included only if payment for these hours is at more than the straighttime rate. Exclude hours for which only shift differential, hazard, incen
tive, or other similar types of premiums were paid. If none, enter "0" in
column 10Y.
The term "production and related workers" includes working supervisors
and all nonsupervisory workers (including group leaders and trainees) en
gaged in fabricating, processing, assembling, inspection, receiving, storage,
handling, packing, warehousing, shipping, trucking, hauling, maintenance,
repair, janitorial, guard services, product development, auxiliary produc
tion for plant's own use (e.g., power plant), and recordkeeping, and other
services closely associated with the above production operations.
Comments
1982
Jan.
Feb.
Mar.
Apr.
May
June
July
Aug.
Sept.
Oct.
Nov.
Dec.
G P O 882-608
24
Chapter 3. ©c©up@ti®nal
Em ploym ent S ta tis tic s
)
The Occupational Employment Statistics (OES) Survey
is a periodic mail survey conducted by State employ
ment 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, for each
State, and for 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 occupa
tional estimates for the Nation and for the cooperating
States. Similar surveys were inaugurated for non
manufacturing industries with the participation of addi
tional cooperating State agencies. State and local
governments were surveyed as well.
Between 1971 and 1981, three survey cycles, on
average, were conducted alternately for manufacturing;
nonmanufacturing; and trade, transportation, com
munications, utilities, and government services in
dustries. In addition to the regularly scheduled surveys,
which follow a 3-year cycle, surveys were conducted for
education services in 1978 and hospitals in 1980. Cur
rently, 48 State agencies (including the District of Col
umbia) are cooperating in this effort.
Industrial classification
The classification system currently used for compiling
and publishing data is that described in the 1972
Standard Industrial Classification (SIC) Manual as revis
ed in 1977. (See appendix B for detailed description of
this system.)
Reporting establishments are classified on the basis of
major product or activity for the previous calendar
year.
Concepts
Occupational classification
The o e s classification system is a combination of two
widely used systems. Titles and descriptions of occupa
tions used for data collection are derived primarily from
the Dictionary o f Occupational Titles, third and fourth
editions published in 1965 and 1977, respectively, by the
Department of Labor’s United States Employment Ser
vice. The 1970 Census o f Population, published by the
Department of Commerce, Bureau of the Census, is the
other major source for occupational classification. The
census is made up of about 400 categories reflecting
broad occupational coverage without definitions. The
Dictionary o f Occupational Titles, on the other hand, is
a more detailed classification system with about 12,000
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 activi
ty. 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);
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 ex
cludes proprietors (owners and partners) of unincor
porated firms; unpaid family workers; and workers on
extended leave (i.e., pensioners 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 oc
cupations which are most closely related to their work.
Part-time workers, learners, and apprentices are
reported in the occupation in which they ordinarily
work.
25
definitions of occupations and is organized to meet the
operating needs of the public employment service.
These two systems, plus information compiled from in
dustry officials and other sources, contributed to the
OES occupational classification system.
This system allows for the constant state of change
that occupational terminology and classification
undergo. This flexibility permits integration of the in
sight gained from each successive round of OES surveys.
In 1983, the oes occupational classification system will
be revised to provide compatibility with the 1980 Stan
dard Occupational Classification System published by
the Department of Commerce.
are returned, these additional occupations are coded ac
cording to the corresponding long-form occupation
content preparatory 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
completing the questionnaires. Normally, two mailings
follow the initial mailing and a subsample of residual
nonrespondents is contacted further by telephone.
Occupational employment data are requested for the
pay period including the 12th of the month, which is
standard for all Federal agencies collecting employment
data.
Data Sources arid Collection Methods
Sampling
Sources of occupational data reported by respondents
are personnel records and, especially for the small
reporting units, personal knowledge of persons com
pleting the reports.
Employment benchmarks for this survey are derived
from employment data tabulated from the reports of
the unemployment insurance program. In some non
manufacturing industries, supplemental sources are
used to obtain lists of establishments that are not
covered by unemployment insurance laws.
Employment information is currently being collected
for 1,700 occupations. A list of occupations has been
designed for each industry or for each group of in
dustries 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 addition, the questionnaire provides
descriptions of 3-digit sic industries to reduce industry
misclassifications.
The long form specifies an extensive list of occupa
tions 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 oc
cupations listed on the forms. When the questionnaires
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 re
porting 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
sample 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 Proeedyres
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 correspon
ding reporting establishments. These distributions are
multiplied by the corresponding benchmark estimates of
total employment in that size class. Estimates for oc
cupations in each industry group are derived by summ
ing all of the occupational size class estimates within
that industry group. Similarly, the estimates of com
bined industry groups are derived by summing the
individual industry components.
Presentation
A report on the results of each o es Survey is publish
ed by the cooperating State employment security agen
cies. b ls published national estimates for survey years
26
1971, 1977, 1978, 1979, and 1980. Each report consisted
of an 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 com
position 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 re
quirements by occupation and for vocational and
educational guidance. The occupational composition of
various industries is also needed to estimate the employ
ment implications of proposed new Government pro
grams, such as those in the fields of defense procure
ment, health, or mass transit. Local employment service
offices use information on the occupational patterns of
industries to locate employment opportunities. Finally,
occupational employment and patterns data are use in
analysis by the firms and in industrial management.
All surveys are subject to response and processing er
rors, although these are reduced through reviewing,
editing, and screening procedures and through contact
with respondents whose data are internally inconsistent
or appear to involve misinterpretation of definitions 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
Selected Nonmanufacturing Industries, Bulletin 2088,
Thompson, John, “ bls Job Cross-classification System Re
lates Information From Six Sources,” Monthly Labor
Review, November 1981.
Describes the relationships of several major classifica
tion systems to the Occupational Employment Statis
tics classification system.
1981.
Presents occupational employment data collected in
1978 for the mining; construction; finance, insurance,
and real estate; and services industries.
Bureau o f Labor Statistics. Occupational Employment in
Transportation, Communications, Utilities, and Trade,
U .S. Department of Commerce, Office o f Federal Sta
tistical Policy and Standards. Standard Occupational
Classification Manual, 1980.
Bulletin 2116, 1982.
Presents occupational employment data collected in
1979 for the transportation, communications, utilities,
and wholesale and retail trade industries.
U.S. Department o f Labor, Bureau of Labor Statistics. Oc
cupational Employment in Manufacturing Industries,
1977, Bulletin 2057, 1980.
U.S. Department o f Labor, Employment and Training Ad
ministration. Dictionary o f Occupational Titles, fourth
edition, 1977.
Comprehensive descriptions of 12,099 jobs coded by
work requirements and duties performed.
Presents occupational employment data collected in
1977 for manufacturing industries.
Bureau o f Labor Statistics. Occupational Employment in
27
Chapter 4. M easurem ent of
Unem ploym ent In States and
Lo cal A reas
tion controls from the Current Population Survey (CPS),
the Bureau of the 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, b l s
established, as a criterion for direct use of c p s data, a
maximum expected relative error of 10 percent for
unemployment given an expected unemployment rate of
6 percent. 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 ( s m s a ) and New
York City. All other State and area estimates are based
on the Handbook method controlled to c p s statewide
estimates as explained below.
B l s and the States also engaged in the u i data base
project to standardize for all States and areas the u i
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 cer
tification week. Currently, monthly estimates of
employment and unemployment are prepared in the
State agencies for some 5,000 geographic areas which
include all States, LMA’s, and counties and cities with
50,000 or more population.
Background
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 FederalState cooperative program, the Department of Labor
develops the concepts, definitions, and technical pro
cedures which are used by State agencies for the
preparation of labor force and unemployment
estimates.
Unemployment estimates have been developed for
Labor Market Areas ( l m a ’s) for over 35 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 dif
ficulties. 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
unemployment was established.
In 1950, the Department of Labor’s Bureau of Em
ployment Security (now Employment and Training Ad
ministration) published a handbook, Techniques fo r
Estimating Unemployment, in order that comparable
estimates of the unemployment rate could be produced
among the States. During the late 1950’s, their ex
periences led to the formulation of the Handbook
method, which is a series of computational steps
designed to produce total employment and unemploy
ment 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, employ
ment, and unemployment. In 1973, after extensive
research, a new system for developing labor force
estimates was introduced which combined the Hand
book method with the concepts, definitions, and estima
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 CPS. The Handbook presents
28
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 in
dustries covered 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
reentered 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 collec
ting ui benefits, have exhausted their benefits, have
been disqualified from receiving benefits, and have
delayed filing 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 esti
mates 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
unemployment is developed for each industry or classof-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. For some
subgroups, special factors, based on relationships de
rived from historical data, are used to adjust the State
covered rate.
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
Handbook estimate of entrants into the labor force is a
function 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 en
trants for a given month is derived from the following
equation:
ENT = A(X + E) + BX
where:
ENT= total entrant unemployment
E
= total employment
X
= total experienced unemployment
29
A, B= synthetic factors incorporating seasonal varia
tion, and assumed relationship between the
proportion o f youths in the working age pop
ulation and the historical relationship of
entrants to either the experienced unemployed
(B factor) or the experienced 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
to ta l num ber of em ployees on payrolls in
nonagricultural industries. Estimates of agricultural
workers, the self-employed, unpaid family workers, and
domestic workers are developed synthetically.
Methodological lmpr©¥@m@nts—
Adjystmamts t@ the Handbook
Research has established that the Handbook pro
cedures alone produce seriously biased estimates of
unemployment and employment as measured by the
CPS. These biases are caused, in part, by methodological
and definitional problems. For example, the employ
ment estimates in the Handbook method are based
primarily on establishment payroll data and are placeof-work estimates. The c p s estimates are based on a
survey of households in the area and are place-ofresidence estimates. A person on an unpaid absence is
excluded from the payroll estimate in the Handbook
method but is considered employed in the CPS. Also, a
person holding two jobs within the reference week is
counted twice in the payroll estimate but only once in
the c p s estimate.
The definitional and methodological differences be
tween the Handbook and c p s estimate of unemploy
ment are more difficult to reconcile. The Handbook
method does not count (or estimate) the number of per
sons 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 treat
ment of persons who fail to qualify for benefits for non
monetary reasons (quits, discharges, etc.), also vary
from State to State. More importantly, the c p s
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 equations developed primarily
from historical national data. These special equations
are subject to numerous errors related to the specifica
tion of functional form, the method of estimation, and
the use of national data which do not reflect interarea
differences in labor markets. While the differences be
tween 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 comparable 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, b l s has intro
duced a number of adjustment procedures and changes
in the previous estimating methodology. These are
described below.
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 disturb
ance 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
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 c p s . Estimated adjustment factors
for the major categories of employment in the Hand
book method were developed on the basis of employ
ment relationships which existed at the time of the 1970
Decennial Census, which will be updated when 1980
census data become available. 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.
Direct use of CPS data
The most fundamental change was the use of c p s 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 comprise each State. In the 10 largest States (and 2
large areas mentioned above), c p s data are used directly
on a monthly basis. In the 40 remaining States and the
District of Columbia, where the sample will not support
the direct monthly use of the c p s data, they are used as
follows.
Consistency/additivity adjustment
Monthly adjustment to CPS. 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 c p s estimate ending in the current month to the cor
responding 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:
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 ad
justed Handbook estimate for 40 States or the CPS
estimate for the 10 largest States—to produce the final
adjusted l m a estimates:
5
I UCPSs(t-K)
K =0
U s(t) = UHBs(t) * --------------------5
I UHBs(t-K)
K =0
UHBa(t)
U a(t) = Us(t)
where:
I U H B a(t)
a
a = area
S = State
where:
t
s
t = time
=tim e period
= State
U s(t)
= Official State estimate
UHBs(t)
The Handbook share procedure allocates the CPS 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 Califor
nia and New York which also have areas taken directly
from the c p s , the Handbook share ratio for the remain
ing areas is applied to the State total minus the CPS area.
= Handbook State estimate
UCPSs(t) = CPS State estimate
Annual benchmark adjustments. I^ach year, monthly
State employment and unemployment estimates
prepared by State employment security agencies using
the Handbook estimating procedure are adjusted, or
benchmarked, to the annual average c p s State estimate.
This is accomplished 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
Producing estimates for parts of IM A ’S
Current labor force estimates at the sub-LMA level are
required by several Federal programs. However, for
areas this small, the data required to compute indepen
dent Handbook estimates are generally not available.
Based on data availability, three alternative methods are
30
used to disaggregate the
level.
lma
Uses and Limitations
estimates to the subarea
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 Comprehensive Employment and
Training Act (ceta ), Public Works and Economic
D evelopm ent Act ( p w e d a ), the U rban D e
velopment Action Grant Program, and Labor Surplus
Area designation. Under ceta , unemployment data are
used with other data to determine the amount of funds
to be allocated; in the case of pw eda , the Urban
Development Action Grant Program, and Labor
Surplus Area designation, the data are used in the deter
mination of area eligibility for benefits.
The cps estimates used on an annual average basis to
control labor force estimates, at the State level for 40
States and monthly for 10 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 di
vided 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 cps ,
given an adequate sample size.
The population-claims method is the perferred techni
que. If residence based ui claims data are available for
the subareas within the lma , 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
estimates of unemployed entrants are allocated based on
the latest available census distribution of adult and
teenage population groups. Employment is disag
gregated using current population distributions
prepared by the Bureau of the Census.
If the necessary ui claims data are not available, but
decennial census data on subarea employment and
unemployment are available, then the census-share
method is used. This method disaggregates Handbook
employment and unemployment based on their 1970
census shares in total lma employment and unemploy
ment.
Finally, if both claims data and decennial census
employment and unemployment are lacking, the
population share method is used. This method
distributes the lma estimate to its subareas using shares
computed from the latest available population data
prepared by the Bureau of the Census.
Technical References
U .S.
Employment Structure and Trends.)
Department o f Labor, Bureau of Employment
Security. Handbook on Estimating Unemployment,
Employment Security Research Methods, Handbook
Series (BLS No. R-185), 1960. (Reprints are avail
able from the Bureau o f Labor Statistics, Office of
U.S. Department of Labor, Bureau of Labor Statistics.
Manual fo r Developing Local Area Unemployment
Statistics, July 1979.
31
Chapter 5. Em ploym ent and
W ages Cowered by
Unem ploym ent Insuranee
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 pro
gram 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 na
tional, State, and county levels. It provides a virtual
census of nonagricultural employees and their wages. In
addition, about 40 percent of workers in agriculture are
covered.
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 col
leges, 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.3
Ui coverage is broad and basically comparable from
State to State. In 1981, u i and u c f e covered 90,641,808
workers, or 90.3 percent of civilian employment.
Covered workers received $1,483 billion in pay or 95.7
percent of the wage and salary component of personal
income. The principal exclusions from coverage are
members of the Armed Forces, railroad employees, and
most domestic workers, agricultural employees, and
some employees of small nonprofit organizations. Also
excluded are the self-employed and unpaid family
members.
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
compensation 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 respon
sibilities of the U.S. Department of Labor’s Manpower
Administration, or its predecessor agencies. Semiannual
reports summarizing the data were issued until 1950,
when the periodical Employment and Wages began
quarterly publication. In 1972, b l s assumed respon
sibility and continued quarterly publication until 1975.
Employment and Wages then became an annual
publication until 1980, when quarterly issues were
resumed.
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
classification may be applied. Occasionally, a single
physical location encompasses two or more separate,
distinct, and significant activities, having separate
records, and classifiable in separate industrial codes.
Each activity unit is then properly reported as a separate
establishment.
Reporting units
A reporting unit is the economic unit for which the
employer submits a contribution report or identifies
Concepts and Methodology
1 Under the Unemployment Compensation for Federal Employees (UCFE)
program.
2 Under the Unemployment Compensation for Ex-servicemen (UCX) pro
gram.
3 The coverage given is, in all cases, the minimum required by Federal law.
State legislation often provides coverage for additional categories o f workers.
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
32
separate locations on a supplemental form that is in
cluded with the regular contribution report.
Most employers covered under State u i 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, employers 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 in
stallations. 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).
Employment
Employment data represent the number of workers
on the payroll during the pay period including the 12th
of the month.4 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 production 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 of
ficials, 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 establish
ment are counted each time reported. Workers are
counted even though their wages may be nontaxable for
purposes during that period (having reached the tax
able 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.
ui
Total wages
Total wages, for purposes of the quarterly u i 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 supplied. Total wages, however, do not include
employer contributions to old-age, survivors’, dis
ability, and health insurance ( o a s d h i ), unemployment
insurance, workers’ compensation, and private pension
and welfare funds.5
In most States, firms report the total wages paid dur
ing the calendar quarter, regardless of the timing of the
services performed. Under laws of a few States,
however, the employers report total wages earned dur
ing 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.
Depending on the method used by the Federal agency in
preparing its quarterly summary balance (cash or ac
crual basis), the gross amount of payrolls is either paid
or payable.
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 in
surance claims that have been filed against them. These
reimbursable accounts are not subject to the quarterly
4
The Department o f Defense is an exception. The employment count in in
stallations o f the Department o f Defense covers all persons employed on the last
workday o f the month plus all intermittent employees during the month. Inter
mittent workers are occasional workers who were employed at any time during
the month.
s 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.
assessment for Unemployment Insurance funds and,
therefore, their taxable wages and contributions are not
reported.
In mid-1982, approximately half the States required
that employers pay ui taxes on the first $6,000 of
employee wages—the minimum established by Federal
law. The remaining States established higher limits on
taxable earnings. The portion of wages subject to taxa
tion has varied substantially over time. In mid-1982
also, about half the States allowed employers to obtain
lower tax rates by making voluntary contributions to the
unemployment tax fund. A small number of States also
require contributions from employees. Such contribu
tions are included without separate identification.
Industrial classification
State employment security agencies use the current
Standard Industrial Classification (sic) Manual to
classify each reporting unit according to its primary ac
tivity. 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 installa
tions reporting more than one activity allocate the pro
per proportion of total production, revenue, sales, or
payroll costs (depending on the industry group) to each
activity. The State agency designates the proportion
ately largest activity as the primary activity. Occasional
ly, two or more relatively minor activities may be deter
mined 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 in
dustries and file separate reports.
Since 1938, the industrial classification of business
establishments and government installations has
undergone a number of modifications. (See table.) Until
1945, classification was based on the Social Security
Board (SSB) 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 non
manufacturing 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 non
manufacturing in 1968. (See table.) Statewide 4-digit
classifying for nonmanufacturing did not become man
datory until 1978. A few industry exceptions allow
3-digit coding (34 4-digit SIC’s are collapsed 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 in
formation sufficient to code at the 4-digit level.
Industrial classification of employment and wage data, 1938-81
Number of industry groups by:
Basis of industrial classification
Social Security
Board (SSB)
Period
2-digit
code
3-digit
code
20
21
21
21
21
21
20
20
146
150
150
148
148
143
143
60
56
58
62
64
64
64
256
236
235
277
277
277
4-digit
code
1939
edition
1942
edition
Standard Industrial Classification
(SIC)
1945
edition
1957
edition
1967
edition
1972
edition
1977
edition
Manufacturing
1938-41 ....................
1942-46....................
1947-55....................
1956-57....................
1958-67....................
1968-74....................
1975-78....................
1979-81 ....................
X
X
X
X
1
469
1
433
1
417
451
452
X
X
X
X
Nonmanufacturing
1938-41....................
1942-57....................
1958-67....................
1968-74....................
1975-77....................
1978 ........................
1979-81 ....................
1 January-March quarter only.
X
X
X
2
494
2
553
553
553
X
X
X
X
2 Not coded on a mandatory basis.
Census separately tabulates central administrative of
fices 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.
In addition, the census reports exclude public sector
employees (except in the Census o f Government). Some
censuses (but not County Business Patterns) include
self-employed persons in the retail trade, construction,
and service industries, who are not covered by ui. The
censuses also impute employment data, while ES-202
adheres to reported figures. The ES-202 data are more
frequently updated and consequently the program
maintains more continuity.
Collection methods
Approximately 4.5 million reporting units in the
nonagricultural private sector submit quarterly reports
to State agencies, with data on monthly employment,
quarterly total and taxable wages, and contributions. In
addition, the 53 State agencies receive reports from
about 33,000 reporting units of the Federal Government
for their civilian employees under the u c f e program
in each State; they also receive reports covering nearly
99 percent of State and 96 percent of local govern
ment employees, and about 40 percent of all farm
workers.
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. The
States have the option of either submitting two
tapes—(1) statewide by 4-digit industry and (2) county
by 2-digit industry—or one tape, 4-digit industry by
county. Most States provide the latter.
Bls, in turn, further summarizes these data at
county, State, and national levels, by industry and by
size of reporting unit, and publishes the summaries in
the quarterly and annual Employment and Wages
publication.
The individual States, which have a wide range of
uses for these data, usually publish their own ES-202
reports.
Current Employment Statistics
The Current Employment Statistics ( c e s ), or 790 pro
gram of b l s , employs a sample of 177,000
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 re
maining industries, 790 uses several other sources, in
cluding the Bureau of the Census’ County Business Pat
terns for certain salespersons and agents, the Interstate
Commerce Commission data for railroad workers, the
U.S. Department of Education and the National
Catholic Welfare Association data for private elemen
tary and secondary schools, and the National Council of
Churches data and State surveys for religious organiza
tions.
In addition to being sample-based as opposed to be
ing a universe count, the 790 program differs from
ES-202 in that it provides hourly earnings for produc
tion (nonsupervisory) workers only whereas ES-202 pro
vides total payroll data for all employees, unrelated to
hours.
Comparison of the ES-202 Program with
Other Series
A number of statistical data series, in addition to the
ES-202 program, produce employment and wage data
comparable in some respects to those obtained by
ES-202. These series all have certain applications,
strengths, and shortcomings. The ES-202 program,
because of its broad universe coverage, continuity, and
currency, is one of the most useful.
Office ©f Personnel Management
The Office of Personnel Management ( o p m ) main
tains a statistical series on Federal employment and
payroll information by agency, type of position and ap
pointment, and employee demographic characteristics.
Both the OPM and the ES-202 series exclude the Central
Intelligence Agency and the National Security Agency,
the Armed Forces, temporary emergency workers, and
crews of certain vessels. The o p m , but not ES-202, in
cludes employees working in foreign countries, workers
paid on a fee or commission basis, and paid patients, in
mates, 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 of the Agricultural Extension Service,
County Agricultural Stabilization and Conservation
Economic Census and County Business Patterns
The Bureau of the Census conducts a census of most
industries every 5 years. These data, along with the an
nual Company Organization Survey for multiunits and
data from the Internal Revenue Service and Social
Security Administration for single units, 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, particularly in some meas
urements, such as size of firm. The Bureau of the
Census uses a finer establishment breakdown than the
ES-202 reporting concept, so that numbers of units and
employment per unit may differ. The Bureau of the
35
employees, however, is fully disclosable.
In addition to published information, unpublished
data, such as county level data and 3-digit industry data
by State are available upon request. Depending on the
request, the data may be provided, for a nominal fee, on
microfiche or magnetic tape.
Committees, and State and Area Marketing Commit
tees.
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
Uses
The Current Population Survey (CPS) is a sample
survey of 60,000 households chosen to represent the en
tire civilian noninstitutional population and labor force.
Therefore, the sample includes categories of workers
which are entirely or partly excluded from the ES-202
program—certain farm and domestic workers, the selfemployed, 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 non
profit organizations, and railroads. The CPS also counts
employees uncompensated because of temporary
absence, but excludes workers under 16 years old.
Because the CPS is a sample and surveys households
rather than establishments, it cannot present employ
ment and wage data in the industrial and geographical
detail available under the ES-202 program, but it does
provide demographic characteristics.
As the most complete universe of monthly employ
ment and quarterly wage information by industry,
county, and State, the ES-202 series has broad economic
significance in evaluating labor trends and major in
dustry developments in time series analyses and industry
comparisons, and in special studies such as analyses of
wages by size of firm.
The program provides data necessary to both the
Employment and Training Administration and the
various State Employment Security Agencies in ad
ministering 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 b l s
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 establish
ment surveys by the Industry/Area Wage, Occupational
Employment Statistics, 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 na
tional 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 assump
tions and forecasting trends in the taxable wage base.
Finally, the ES-202 report is one of the best sources of
detailed employment and wage statistics used by
business and public and private research organizations.
Presentation
Employment and Wages, an annual and quarterly b l s
publication, presents State and national totals for
covered employment and wages by broad industry divi
sion and major industry group. Data for Federal
workers also are shown by agency, industry, and State.
For the first quarter of each year, the publication in
cludes distributions of employment and wages by size of
reporting unit for each major industry division within
each State and by industry for the United States as a
whole. These data are distributed in nine employmentsize categories.
To preserve the anonymity of establishments, b l s
withholds publication of data for any county, State, or
national 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 otherwise violate the State’s
disclosure provisions. Information concerning Federal
36
Technical References
Papers From North American Conference on Labor
Statistics, 1973. l/.S . Department of Labor, Bureau of
Armknecht, Paul A ., and Cartwright, David W. “ Statistical
Uses o f Administrative Records,” Selected Papers From
Annual Meeting o f the American Statistical Associa
tion, October 1979. U.S. Department of Health,
Labor Statistics.
Education, and Welfare, Social Security Administra
tion.
U .S. Department o f Labor, Manpower Administration (now
Employment and Training Administration). “Technical
Notes on Insured Unemployment, Covered Employ
ment, and Wage Statistics: Their Source, Nature, and
Limitations,” Summary o f Employment Security Sta
tistics Reports, 1975.
Bunke, Alfred L. Quarterly Report o f Employment, Wages,
and Contributions (ES-202), Selected Papers From
North American Conference on Labor Statistics, 1973.
U.S. Department of Labor, Bureau of Labor Statistics.
U .S. Department of Labor. “ Employment, Wages, and Con
tributions, ES-202,” Employment Security Manual,
Part III, Sections 0400-0599, as revised in 1972.
Ehrenhalt, Samuel M. “ Some Thoughts on Planning a Com
prehensive Employment Statistics Program,” Selected
37
Chapter (S ©©miiymtir
.
Expenditures ®
ndl
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 independent surveys, a Diary Survey and an Inter
view 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 p i .
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 re
quirement. The remainder of this chapter describes the
new continuing survey that extends the b l s tradition of
providing data describing the consumption behavior of
American families.
Consumer expenditure surveys are specialized family
living studies in which the primary emphasis is on collec
ting data relating to family expenditures for goods and
services used in day-to-day living. Expenditure surveys
of the Bureau of Labor Statistics also collect informa
tion 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’ in
comes and expenditures for all kinds of goods until
World War I. A third survey, spanning 1917-19, provid
ed weights for computing a cost-of-living index, now
known as the Consumer Price Index ( c p i ). (See Volume
II.) 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
welfare 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, covering only urban consumers. The seventh
survey, the 1960-61 Survey of Consumer Expenditures,
once again included both urban and rural families, pro
vided the basis for revising the c p i , 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
D@seripti®o of the Ongoing Syrwsy
Unlike previous surveys, the latest survey, initiated in
late 1979, is ongoing. Data will thereby be 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 kinds 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. There is an Interview
Panel Survey in which 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. In addition, there is
a Diary (or recordkeeping) Survey completed at home by
the respondent family for two consecutive 1-week
periods.
The unit for which expenditure reports are collected is
the set of eligible individuals comprising a consumer
unit, which is defined as (1) all members of a particular
housing unit who are related by blood, marriage, adop
38
tion, or some other legal arrangement, such as foster
children; and (2) a person living alone or sharing a
household 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 ongoing survey, students living in universitysponsored housing are also included in the sample as
separate CU’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 ex
penditures 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.
In the Diary Survey, detailed data are collected on all
expenditures made by consumer units during their par
ticipation in the survey. All data collected in both
surveys are subject to Census and b l s confidentiality re
quirem ents, which prevent the disclosure of
respondents’ identities.
iGiterwiew Survey
The Interview Panel 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 fair
ly regular basis, such as rent, utility bills, or insurance
premiums. Each occupied sample unit is interviewed
once per quarter for five consecutive quarters. For the
initial interview, inform ation 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 inter
views.
The second through fifth interviews use uniform
questionnaires to collect expenditure information in
each respective quarter. Data collected in these ques
tionnaires, which are arranged by major expenditure
component (e.g., housing, transportation, medical,
education), form the basis of the expenditure estimates
derived from the Interview Survey. In addition, infor
mation is obtained on the names of establishments (or
outlets) from which selected commodities or services are
purchased. Wage, salary, and other information on the
employment of each cu member is also collected in cer
tain of these interviews. In the fifth and final interview,
an annual supplement is used to obtain a financial pro
file of the consumer unit. This profile consists of infor
mation on the income of the cu as a whole, including
unemployment compensation; income from royalties,
39
dividends, and estates; alimony and child support, etc.
Information on occupational expenses and on
changes in assets and liabilities is also collected. This in
formation, along with the demographic information
collected in the first interview, links the Interview
Survey expenditure data to that of the Diary Survey for
publication in an integrated format. After the fifth in
terview, 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; hence, the rotating nature of
the survey.
The rotation of Interview Survey families into and
out of the sample is a new feature of the survey which
results from its being continuous. New families are in
troduced into the sample on a regular basis as other
families complete their participation. This rotating pro
cedure is designed to improve operational efficiency.
Another new feature of the current survey is that data
collected in each quarter are considered independently,
so that estimates are not dependent upon a family par
ticipating for a full five quarters in the survey.
Diary Survey
The primary objective of the Diary Survey is to obtain
reliable 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 sup
plies and services, nonprescription drugs, and personal
care products and services. The Diary Survey is not
limited to these types of expenditures, but rather, in
cludes all expenses which the consumer unit incurs dur
ing 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
R ecord of Daily Expenses. The H ousehold
Characteristics Questionnaire is used to record informa
tion pertaining to age, sex, race, marital status, and
family composition, as well as information on the work
experience and earnings of each cu member. This
socioeconomic information is used by b l s to classify the
consumer unit for publication of statistical tables and
for economic analysis. Household characteristic data
also provide the link in the integration of Diary expen
diture data with Interview expenditure data for
publishing a full profile of consumer expenditures by
demographic characteristics.
The daily expense record is designed as a selfreporting product-oriented diary on which respondents
record a detailed description of all expenses for two con
secutive 1-week periods. The diary is divided by day of
purchase and by broad classifications of goods and ser
vices—a breakdown designed to aid the respondent
and transmitted to Washington, and again cycled
through the computer preedit. This continues until er
rors identified by the preedit no longer appear. Once a
panel month’s preedit is complete, data necessary for
bounding are transcribed to the next quarter’s question
naire. 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 com
puter edits and adjustments which include the iden
tification and correction of data irregularities and in
consistencies throughout the questionnaire. Other ad
justments convert mortgage and vehicle payments into
principal and interest (given associated data on the in
terest 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
data from the Bureau of the Census, conducts an exten
sive review to ensure that severe data aberrations are
corrected. The review takes place in several stages: A
review of counts and means by region; a review of fami
ly relationship coding inconsistencies; a review of
selected extreme values (both high and low) for expen
diture and income categories; and a verification of the
various data transformations performed by b l s . Cases
of questionable data values or relationships are in
vestigated 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 en
tries. The routines, which affect all fields in the data
base except income and assets, are intended to improve
the estimates derived from the survey. In addition,
allocation routines are applied to the interview data in a
fashion similar to that for the diary data.
when recording daily purchases. The items reported are
subsequently coded by the Bureau of Census so that b l s
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 in
terview survey data are processed separately. Completed
diary questionnaires are manually reviewed for com
pleteness and consistency and are then transmitted to
the Census Processing Center in Washington, D.C.,
where further computer processing is performed. In ad
dition, missing or invalid demographic or workexperience data (except income) are imputed. The
families are assigned weights so that estimates can be
derived that represent the total population. Finally,
monthly diary data tapes are transmitted to the Bureau
of Labor Statistics.
As the monthly diary data tapes are received, b l s
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 af
fect the reasonableness of estimates after the data are
aggregated. All coding or extreme value errors are cor
rected before further processing is performed on the
quarterly data bases.
Selected portions of the diary data are also affected
by automated imputation and allocation routines when
respondents report insufficient detail to meet publica
tion requirements. The imputation routines assign
qualifying information to data items when there is clear
evidence of invalid nonresponse. For example, the
qualifiers classify food expenditures by type of process
ing and apparel expenditures by age and sex groupings.
Allocation routines are a means of transforming reports
of nonspecific 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.
Census processing of Interview Survey questionnaires
proceeds along similar lines. The questionnaires are
completed and returned to the regional offices, where
codes are applied to identify demographic charac
teristics, 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 process
ing, the data are keyed and transmitted to Washington
where they pass through a detailed computer preedit.
Inconsistencies, errors, and identification of missing
questionnaires are transmitted back to the regional of
fices for reconciliation by the field staff through office
review or interviewer followup. Corrections are keyed
Sample Design
Selection of households
■-
-
The Consumer Expenditure Survey is a national prob
ability sample of households designed to represent the
total civilian noninstitutional population. The selection
of households begins with primary sampling units
(PSU’s), which consist of counties (or parts thereof),
groups of counties, or independent cities. The set of
sample PSU’s used for the survey is composed of 101
areas, of which 85 urban areas have been previously
defined and selected by b l s for the Consumer Price In
dex program. These urban PSU’s are classified according
to the following four categories: “ A” PSU’s, which com
prise 27 certainty areas (i.e., they are self-representing)
that are primarily Standard Metropolitan Statistical
40
Areas (SMSA’s); “ B” p s u ’s, which comprise 20 s m s a ’s adjustment based upon cu family composition. In the
with a total 1970 population of over 400,000; “ C” p s u ’s case of the Diary, a pre-Christmas seasonal factor is
comprising 22 SMSA’s with a total 1970 population of also used in the estimation of weights.
400,000 or less; and “ D” p s u ’s, comprising 16 urban
places in all areas outside of s m s a ’s. These 16 urban
Presentation
p s u ’s were supplemented with 16 additional “ E” p s u ’s,
representing the rural population, in the original design.
Information from the ongoing Consumer Expen
However, as of the fourth quarter of 1981, the rural
diture Survey is available in bulletins, reports, analytical
portion was eliminated due to budget constraints im
papers, and on public use tapes. The publications may
posed on the Bureau.
be obtained through the b l s Office of Publications,
The sampling frame (i.e., the list from which housing
regional offices, or from the Government Printing Of
units are chosen) for this survey is generated from the
fice. Information on public use tapes can be obtained
1970 census’ 100-percent detail file, which is augmented
from the b l s Division of Living Conditions Studies.
by a sample drawn from new construction permits and
Publications from the consumer expenditures surveys
coverage improvement techniques to eliminate recog
generally include tabulations of average expenditures
nized deficiencies in that census. In addition, the
and income arrayed by family characteristics. Data
sampling frame for the Diary Survey was augmented
tabulated for a given year are shown at a relatively ag
during the time prior to the Christmas and New Year’s
gregated level due to the small sample size of the on
holidays to account for expenditure increases and dif
going survey. As the survey continues and more data
ferences during the period.
become available, however, estimates for several years
The Bureau of the Census establishes an address sam
may be combined to provide greater expenditure detail
ple of 6,800 households that are requested to participate
and additional classifications of families.
annually in the Diary Survey. This results in an effective
The public use tapes contain the actual expenditure
annual sample size of 4,800, since many interviews are
and income reports of each family but prevent iden
not completed due to refusals, vacancies, or the nonex
tification of the family. By eliminating selected
istence of the household address. The actual workload
geographic detail, the Bureau eliminates the possibility
of interviews is spaced over the 52 weeks of the year. As
that participating families may even be indirectly iden
to the Interview Survey, approximately 8,400 addresses
tified.
are contacted in each of the five calendar quarters.
Allowing for bounding interviews, which are not includ
Uses and Limitations
ed in estimates, and for nonresponse (including vacan
cies), the number of completed interviews per quarter is
As in the past, the revision of the Consumer Price In
targeted at 4,800. Each month, one-fifth of the units in dex remains a primary reason for undertaking such an
terviewed are new to the survey. This panel—and all extensive survey. The results of consumer expenditure
others—is interviewed for five consecutive quarters and surveys have been used to select new market baskets of
then dropped from the survey. This rotation of panels, goods and services for the index, to determine the
used by the Bureau of the Census in several other relative importance of the items selected, and to derive
continuing surveys, has the advantage of operational ef new cost weights for the baskets.
ficiency.
The survey data are of value to government and
private agencies interested in studying the welfare of
Weighting
particular segments of the population, such as the aged,
Each family included in the CES represents a given low-income families, urban families, and those receiv
number of families in the U.S. population, which is con ing food stamps. The Internal Revenue Service has used
sidered as the universe. The translation of sample the data as the basis for revising the average State sales
families into the universe of families is known as tax tables which taxpayers may use in filing Federal in
weighting. Several factors are involved in determining come tax returns. The survey data are used by economic
the weight for each consumer unit for which a usable policymakers interested in the effects of policy changes
report is received. One factor in assigning weights is the on levels of living among diverse socioeconomic groups.
inverse of the probability of selection of the housing Econometricians find the data useful in constructing
unit and the adjustment for subsampling in the field. economic models. Market researchers find them
For interviews which cannot be conducted in occupied valuable in analyzing the demand for groups of goods
sample households because of refusals or the fact that and services. The Department of Commerce uses the
no one is home, a complex noninterview adjustment is survey data as a source of information for revising its
made. Additional factors include a national ratio- benchmark estimates of some of the personal consump
estimate adjustment for age, sex, and race to known tion expenditure components of the gross national
civilian noninstitutional population controls and a final product.
41
Sample surveys are subject to two types of errors,
nonsampling 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 recording or coding the
data obtained, and other errors of collection, response,
processing, coverage, estimation for missing data, and
interviewer variability.
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. Standard error tables applicable to published
b l s data can be obtained from the b l s Division of Liv
ing Conditions Studies.
42
Chapter 7. Producer Prices
Part I. Traditional Methodology
Backgroymd
ing industries are covered by approximately 90,000 price
quotations and 6,000 published product indexes each
month.
Part I of this chapter describes the traditional Pro
ducer Price Index methodology, much of which is cur
rently being phased out. It includes a brief section on
the Industry Price Index methodology, which shares the
weaknesses of the traditional p p i methodology. Part II
describes the methodology of the Producer Price Index
Revision, including a discussion of some of the
problems with traditional p p i methodology which the
p p i r is designed to correct. Some of the principles and
practices which have guided the traditional p p i program
will, however, continue to be observed even after the
transition to the p p i r is complete.
The Producer Price Index ( p p i ), formerly known as
the Wholesale Price Index, is one of the oldest con
tinuous statistical series published by the Bureau of
Labor Statistics, as well as one of the oldest in 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 resolu
tion authorizing the Senate Committee on Finance to in
vestigate 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 in 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.
The m ost com prehensive overhaul of p p i
methodology ever planned has been underway since the
late 1.970’s in what is called the Producer Price Index
Revision ( p p i r ). By the mid-1980’s, all indexes in the
Producer Price Index will be calculated under this new
methodology. By January 1982, nearly one-third of the
total value of domestic mining and manufacturing pro
duction was being sampled to produce indexes
calculated with p p i r methodology. Commodity indexes
long established in the p p i are being continued under the
new methodology, and all newly introduced indexes for
products and industries are being estimated according to
p p i r methods. As of January 1982, approximately 3,450
commodities were included in the overall p p i program,
and the Bureau was receiving about 18,000 price quota
tions per month. This data base will be expanded at
6-month intervals until all 493 mining and manufactur
Deserspti®ii of Surrey
Concepts
1 Senate Committee on Finance, Wholesale Prices, Wages, and
Transportation, Senate Report No. 1394, “The Aldrich Report,” Part
I, 52d Congress, 2d sess., Mar. 3, 1893; and U.S. Department of
Labor, Course o f Wholesale Prices, 1890-1901, Bulletin No. 39,
March 1902, pp. 205-09.
43
The Wholesale Price Index was established as a
measure of price changes for goods sold in primary
markets in the United States, “ wholesale” referring to
sales in large quantities rather than prices received by
wholesalers, jobbers, or distributors. In the wake of
some semantic confusion, the name Producer Price In
dex was adopted in 1978 to emphasize that the index
measures changes in selling prices received by producers
from whoever makes the first commercial purchase.
From its inception, the All Commodities Index was
considered a general purpose index designed to measure
changes in the general price level in other than retail
markets. From the beginning of the index, however, at
tention was directed to some specific needs of users, and
indexes for individual commodities and for major com
modity groups were published. As early as 1903, two
special group indexes by stage of processing—raw com
modities and manufactured commodities—were
published to meet the needs of students of price
statistics.
Although most price quotations reported to the
Bureau are the selling prices of selected manufacturers
or producers, some prices are those quoted on organized
exchanges (spot prices) or at central markets. Prices for
imported commodities are those received by im
porters—the first commercial transaction involving the
commodity in the United States. Since the index is in
tended to measure “ pure” price change, that is, not in
fluenced by changes in quality, quantity, shipping
terms, product mix, etc., commodities included in the
index have been defined by precise specifications which
incorporate their principal price-determ ining
characteristics.2 So far as possible, prices are f.o.b.
(free on board) production point and have referred to
sales for immediate delivery. Prices applicable to longrun contracts have been historically excluded, except
where contract prices have dominated the market.
“ Futures” prices are not included.
day other than Tuesday is used because it is considered
more representative. Indexes for natural gas, most
refined petroleum products, and some industrial
chemicals are based on data from the entire month and,
therefore, are lagged 1 month in the indexes. A gasoline
index for November, for example, would reflect price
changes that occurred in October.
The Bureau attempts to base the PPI on actual
transaction prices. Companies are requested to report
prices less all discounts, allowances, rebates, free deals,
etc., so that the resulting net price is the actual selling
price of the commodity for the specified basis of quota
tion. b l s periodically emphasizes to reporters the need
to take into account all discounts and allowances. List,
or book, prices are used if transaction prices are unob
tainable. Some of these list prices are, of course, true
transaction prices. List prices have been used for only
about 20 percent of traditional p p i ’s . Rebates and other
forms of price concessions granted by producers to their
distributors reduce the proceeds received by producers
and, therefore, are reflected as decreases in the p p i , even
though such rebates are intended to be passed through
to the ultimate buyer. Conversely, terminations of
rebate programs are considered price increases in the
Um ers©
iiw
The PPI universe consists of all commodities sold in
commercial transactions in primary markets of the
United States, commodities produced in the United
States as well as those imported for sale. The universe
covers manufactured and processed goods and the out
put of industries classified as manufacturing,
agriculture, forestry, fishing, mining, gas and electrici
ty, public utilities, and goods competitive with those
made in the producing sector, such as waste and scrap
materials. All systematic production is represented, but
individually priced items, such as works of art, are ex
cluded. Also excluded are goods transferred between
establishments owned by the same company (interplant
or intracompany transfers). Goods sold at retail by
producer-owned retail establishments are excluded
because they conceptually belong to a retail (customers’)
universe, rather than to primary market transactions.
Civilian goods normally purchased by the Govern
ment are in the universe, but military goods are not.
Government sales of some commodities (e.g., electric
power) are included if they can be considered com
petitive with free market sales.
p p i.
Prices
To the extent possible, the prices used in constructing
the index are those that apply to the first significant
commercial transaction in the United States. Transac
tions for the same item at later stages of distribution are
not included. However, as raw materials are trans
formed into semifinished and finished goods, the
resulting products are represented.
With some exceptions, the prices refer to one par
ticular day of each month. In most cases, the pricing
date is Tuesday of the week containing the 13th day; but
for some commodities (farm products, particularly) a
2 An example of a commodity specification for steel strip is:
“ Strip, cold-rolled, carbon steel, coils, No. 4 temper, No. 2 finish,
No. 3 edge, base chemistry, 6” x 0.050” , in quantities of 10,000 to
19,999 lbs.; mill to user, f.o.b. mill, per 100 lbs.”
44
Prices are generally f.o.b. production or central
marketing point to avoid reflection of changes in
transportation costs. Delivered prices are included only
when the customary practice of the industry is to quote
on this basis and the Bureau can not obtain a price at the
production point. In such cases, adjustments for
changes in transportation costs are made whenever
possible. Subsidies to producers and excise taxes are ex
cluded since they are not considered part of the market
price; however, import duties are included as part of the
selling price of imported goods.
Although the same commodity usually is priced
month after month, it is necessary to provide a means
for bridging over changes in detailed specifications (or
descriptions of items priced) so that only real price
change will be measured. An adjustment is particularly
important when new commodities are introduced. Even
when specifications of existing commodities are changed,
care is exercised to insure that only price changes in
fluence the index. A new price series resulting from a
physical change in an article or a change in its selling
terms is substituted for the earlier series either by direct
comparison or by linking. The objective of the linking
procedure is to ensure that the index will reflect only
those changes due to actual price difference. Each time
a change in the item priced occurs, the Bureau appraises
its significance to ascertain whether an actual price
change occurred. If the specification change is minor
and does not involve price-making factors, the substitu
tion is effected by direct comparison, and any reported
price change between the old and the new specification
is reflected in the index. If the change is major and in
volves price-making factors, the substitution is made by
linking, and any price change associated with a quality
change is not permitted to affect the index level.3
When differences are major, an attempt is made to
obtain data from the reporters on the cost of the change
and to adjust the price index accordingly. This is par
ticularly important in the case of some durable goods,
such as automobiles, which have periodic model
changes. Also, price increases which result from the ad
dition of features that formerly sold at extra cost are not
reflected in the index. Conversely, price changes at
tributable to deletion of equipment which was formerly
standard are not treated as decreases.
The problem <pf devising adequate quality adjustment
techniques for changes in commodities has been par
ticularly troublesome for complex capital equipment,
where new machinery frequently incorporates a new
generation of technological advance without correspond
ing increases in costs. Any inability to reflect technical
progress embodied in new products imparts an upward
bias of unknown magnitude to p p i data. (Part II further
describes this problem and the Bureau’s response.)
Prices for specific commodities reported by in
dividual companies are generally averaged using an
unweighted mean. If five companies report prices for a
product, each company’s price counts one-fifth towards
the computation of the composite price, even though
the firms may be of widely different sizes. Reporting
companies are weighted equally in most cases. Monthto-month price change should be computed from
matched-company data. In order that a change in the
company-reporter sample itself does not affect the
measure of price change, the change for any month to
the succeeding month is calculated from identicalcompany data. A new report affects the index no earlier
than the second month after introduction.
Linking is also used for: (1) The addition to or dele
tion of commodities or groups of commodities from the
total index; (2) the addition to or deletion of a company
report from the sample of companies priced; or (3) oc
casionally, a change in the price source. Whenever a
new commodity is added to an existing commodity
grouping, linking of the new item to any one of the ex
isting items is not pertinent. Instead, the weights of the
entire group are redistributed to include the new items,
and the link is made at the group level instead of at the
commodity level. A similar procedure is used to handle
items that are dropped from the index without direct
replacement.
In the event production of a specified commodity is
discontinued by a reporter, or its importance is reduced,
the Bureau collects price data for a similar or a replace
ment item. Prices are obtained for the new and the
discontinued series for an overlap period. When pro
duction cost data are unavailable, the index is extended
by linking, and the difference, if any, between the new
item price and the original item price is taken as a
measure of the quality difference between the two items.
Classification
3 The following simplified example illustrates the principle of link
ing: The September price for a certain machine was $2,347.50. In Oc
tober, a new model of the machine was introduced, priced at
$2,562.60. The new model was considered essentially comparable with
the old, except that it had a more powerful motor and larger tires.
These were valued at $186.20 more than the value of those used on the
former model. For linking, the September price of the new model was
estimated at $2,533.70 ($2,347.50 September price of former model
plus $186.20 increase in value of motor and tires). The price com
parison between September and October was based on the estimated
September price of $2,533.70 and the reported October price of
$2,562.60. Thus, a 1.1-percent increase was reflected in the October
index, but the price change due to quality improvement (more power
ful motor and larger tires) was not reflected. Actual linking pro
cedures used in ppi calculations vary somewhat from those employed
in this example.
45
Producer Price Indexes can be organized in a number
of ways, such as by commodity or stage of processing.
The commodity classification organizes products by
similarity of end use or material composition. The
stage-of-processing ( s o p ) classification organizes pro
ducts by degree of fabrication (i.e., finished goods, in
termediate goods, and crude materials) and by class of
buyer (i.e., final demand or not).
The commodity classification structure of the p p i is
unique and does not match any standard classification
such as the Standard Industrial Classification (sic), the
Standard Commodity Classification, or the United Na
tions 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. Since 1952, 15 major
commodity groupings have comprised the All Com
modities Index; 2 major commodity groupings have
been combined to form the Farm Products and Pro
cessed Foods and Feeds Index, and the other 13 have
been grouped into the Industrial Commodities Index.
Each major commodity grouping includes (in de
scending order of aggregation) subgroups, product
classes, subproduct classes, and individual items.
Stage-of-processing (SOP) indexes regroup com
modities priced in the p p i at the subproduct class level
according to: (1) The amount of processing, manufac
turing, or assembling they undergo before entering com
mercial markets; and (2) the class of buyer. A single
subproduct class may appear in several different s o p
categories. For example, 31 percent of the value-weight
of the index for citrus fruits has been assigned to the in
dex for crude foodstuffs and feedstuffs to represent the
proportion of sales to food processors; most of the re
maining 69 percent has been assigned to the index for
finished consumer foods. The value-weights are the
same as those for the p p i subproduct classes. The alloca
tions among various s o p categories are currently based
on input-output studies for 1972 conducted by the
Bureau of Economic Analysis, U.S. Department of
Commerce.
Finished goods are commodities that will not undergo
further processing and are ready for sale to the ultimate
(final demand) user, either an individual consumer or a
business firm. Capital equipment (formerly called pro
ducer finished goods) includes commodities such as
motor trucks, farm equipment, and machine tools.
Finished consumer goods include foods and other types
of goods purchased by retailers and eventually used by
consumers. Finished consumer foods include unpro
cessed foods such as eggs and fresh vegetables, as well as
processed foods such as bakery products and meats.
Other finished consumer goods include durables such as
automobiles, household furniture, and jewelry, and
nondurables such as apparel and gasoline.
Intermediate materials, supplies, and components in
clude commodities that have been processed but require
further processing before they become finished goods.
Examples of such semifinished goods include flour, cot
ton yarns, and steel mill products. Commodities that are
physically complete but which are purchased by firms as
auxiliary products are also considered intermediate
goods. Examples would include residual fuel, paper
boxes, and motor vehicle parts. The proportion of
gasoline sold to businesses for their trucks or auto fleets
would also be classified as part of the intermediate
goods index since such sales are not considered part of
final demand.
Crude materials fo r further processing include pro
ducts entering the market for the first time which have
not been manufactured or fabricated but which will be
processed before becoming finished goods; scrap
materials are also included. Crude foodstuffs and
feedstuffs include items such as grains and livestock.
Crude nonfood materials include raw cotton, crude
petroleum, natural gas, hides and skins, and iron and
steel scrap.
As we shall see in more detail in Part II, stage-ofprocessing indexes are more useful than commodity
grouping indexes for analysis of general price trends.
Aggregated commodity indexes sometimes produce ex
aggerated or misleading signals of price changes by
multiple counting of the same price movement through
various stages of processing.
In addition to stage-of-processing and commodity
grouping indexes, indexes are available by durability of
product and for a number of special commodity group
ings, such as construction materials and copper and
copper products. Durability-of-product allocations of
individual commodities are based upon the Bureau of
the Census definition that products with an expected
lifetime of less than 3 years are classified as nondurable,
and products with a longer life expectancy are con
sidered durable goods. Special grouping indexes rear
range p p i data into different combinations of price
series, so that the appropriate prices and weights are
those of the p p i . Although most p p i ’s are based on a na
tional sample because most products are produced for a
national market, regional indexes are published for a
few series, notably the major refined petroleum pro
ducts, electric power, and coal. Industry price indexes
are described later in this chapter.
Data Sources and Collection Methods
Prices
Reporting of price data by companies through mail
questionnaires is voluntary and confidential. Most
prices are collected each month; for a few commodities
which change prices infrequently, however, a shuttle
questionnaire is mailed quarterly. While price data are
generally obtained directly from producing companies,
trade publications are sometimes used when they are ac
cepted as reliable by the Bureau and the industry; the
Bureau has been phasing out this practice as much as
possible, however. For fish and most agricultural pro
ducts, the Bureau uses prices collected and published by
other Government agencies.
Price reporting is initiated, whenever possible, by a
visit by a Bureau representative to the prospective
respondent. A detailed report describing all of the
pricemaking characteristics of the commodity is
prepared for each new price series. This commodity
price information form becomes a part of the perma
nent record for the series. After the initial collection of
prices, monthly information is collected by mail on a
shuttle schedule ( b l s 473, shown at the end of this
chapter).
Sampling
Traditional p p i ’s are based on nonstatistical, judg
ment samples of commodities, of specifications
(descriptions), and of reporters. Until p p i r sampling
techniques became standard, the sample of commodities
was chosen after a review of the data of the industrial
censuses and other statistics on the value of transac
tions. Generally, the commodities chosen were those
with the largest shipment values. New items were not
added until they became established in the market.
Samples of specifications and of reporters were
selected after consultation with trade associations or
other industry representatives and with staffs of other
Government agencies. Individual commodity specifica
tions were selected on the basis of net dollar sales. That
is, the volume seller of the industry (not of the com
pany) was preferred. The specification described not
only the physical characteristics but also the most com
mon quality, grade, level of distribution, and market.
46
However, terms of sales (discounts, etc.) were based on
the company’s own most common practice. For some
commodities, prices were quoted by producers and
sellers in terms of a single specification taken as stan
dard; all other prices were quoted as differentials from
the standard, as for some farm products such as wheat
and cotton. When no standard commodity basis existed,
the specification to be priced was selected with the help
of industry experts.
The number of reporters was determined, to some ex
tent, by the variation of price movements among them
and the degree of price leadership. Whenever possible, a
minimum of three companies was obtained to report on
each item so that data for specified commodities could
be published without disclosure of information supplied
by individual companies. For commodities with more
than one major production area and a definite regional
pattern, a larger sample was selected. Among these
commodities are electric power; refined petroleum pro
ducts; waste materials; bituminous coal; and building
materials such as brick, cement, and stone.
Estimating Procedures
Weights
The ppi weights
represent the total net selling value of
commodities produced, processed, or imported in this
country and flowing into primary markets. The values
are f.o.b. production point, exclusive of excise taxes.
The value of interplant transfers, military products, and
goods sold at retail directly from producing
establishments also is excluded. Thus, the definition of
the weights conforms to the universe definition.
Price data for individual commodities are combined
into indexes for various groupings by weights based on
the total value of commodity shipments. The major
sources of shipment-value data include: Bureau of the
Census—Census o f Manufactures and Census o f
Mineral Industries', Bureau of Mines—Minerals Year
book and other publications; U.S. Department of
Agriculture—Agricultural Statistics and other publica
tions; and Bureau of Fisheries—Fisheries o f the United
States and other publications. In additon, other sources
of data, such as trade associations, have been used.
Each commodity price series is considered represen
tative of a class of prices and is assigned its own weight
(the shipment value of the commodity) plus the weights
of other related commodities not directly priced but
whose prices are known or assumed to move similarly.
The assignment of price movements for priced com
modities to those for which quotations are not obtained
is referred to as imputation. For some commodities,
such as ships and some kinds of custom-made
machinery, it is not possible to obtain direct measures of
price movements. The weights for such items are as
signed to other commodities or groups of commodities
for which prices are available. Usually, this imputation
is made to priced commodities that have a similar
manufacturing process, on the assumption of similar
price movements.
P pi weights are revised when data from the quinquen
nial industrial censuses, as well as sufficient budgetary
resources, become available. Beginning in 1976, weights
have been based on the 1972 industrial censuses. Indexes
for 1947 through 1954 were based primarily on the 1947
censuses. In the January 1955 index, adjustments were
made to align the major group weight totals with
1952-53 average shipment values as reported in the
Surveys o f Manufactures. Weights based on the 1954
census shipment values were introduced in January
1958. From 1961 through 1966, weights were based on
1958 census values, and from 1967 through 1975, they
were based on 1963 census values.
The Bureau publishes the relative importance of each
item in the p p i , rather than the actual values used as
weights. The relative importance of an item represents
its basic value weight used in the index, including im
putations, multiplied by the relative of price change
from the weight date to a later date; the result is ex
pressed as a percent of the total for all commodities or
for some index grouping. Data showing the relative im
portance of commodity subproduct classes with respect
to the three major stage-of-processing groupings are
also available.
Bls calculates and releases relative importance data
each December. Except when entirely new weights are
introduced from the latest industrial censuses, relative
importance data usually change from December to
December prim arily because of relative price
movements. Thus, a commodity whose price rises faster
than the All Commodities Index from one December to
the next will automatically have a higher relative impor
tance; a commodity whose price falls or rises less than
the All Commodities Index will show a smaller relative
importance. Relative importance data, however, are not
used as fixed inputs by the Bureau to calculate indexes
for a whole year before the next set of relative impor
tance figures is calculated. Theoretically, the Bureau
could calculate and publish a new set of relative impor
tance data every month. Relative importance data for
any given grouping also change when its components
are subjected to a sample change.
Index calculation
In concept, the p p i is calculated according to a
modified Laspeyres formula:
Ii = (IQ aP /IQ aP o)xlG O
47
not maintain tables of historical dollar prices. Com
puter tapes and microfiche sets of p p i data are available
at cost from the Bureau. Articles by Bureau staff
analyzing the economic background for recent patterns
in p p i and c p i data appear in the Monthly Labor
Review.
where:
P0 is the price of a commodity in the comparison
period
P; is its price currently
Qa represents the quantity shipped during the weightbase period.
Seasonally adjusted data. Producer Price Indexes at a
variety of levels of aggregation may be seasonally ad
justed if statistical tests show there is a stable pattern of
seasonal price changes. A few of these seasonally ad
justed indexes, as well as a larger number of monthly or
quarterly seasonally adjusted percent changes, are
printed in the news release and detailed report. Official
analyses of the p p i customarily are based on seasonally
adjusted data, although unadjusted data are used when
tests show an absence of stable seasonality. Seasonal ad
justment factors are recalculated early each year, b l s
uses the Bureau of the Census’ X -ll seasonal adjust
ment method to compute these seasonal factors.
For analyzing general price trends in the economy,
seasonally adjusted data usually are preferred because
they eliminate the effect of changes that normally occur
at about the same time and in about the same magnitude
every year—such as price movements resulting from
normal weather patterns, regular production and
marketing cycles, and model changeovers. For this
reason, seasonally adjusted data reveal more clearly the
underlying cyclical trends. The unadjusted data are of
primary interest to users who need information which
can be related to the actual dollar values of transactions.
Unadjusted data generally are used in escalating sales
and purchase contracts, for example.
An alternative formula more closely approximates the
actual computation procedure:
I i = [ ( I Q aP o(P i/P o))/X Q aP o] X 100
In this form, the index is a weighted average of price
relatives for each item (P /P 0 The expression (QaP0)
).
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. Each value weight includes not only
the value of items priced but also the values of unpriced
items whose price movements are assumed to behave
similarly. When new weights are introduced, the index
with new weights is linked to the index constructed with
earlier weights. The weight adjustment itself, therefore,
affects only the later calculations of average price
change. When specifications or samples change, the
item relatives must be computed by linking (multiply
ing) the relatives for the separate periods for which the
data are precisely comparable.
Analysis and Presentation
The monthly p p i is published first in a news release,
usually issued in the second week of the month follow
ing the reference month. The release includes a narrative
on the most significant commodity price movements
within the major stage-of-processing groups, as well as
tables showing indexes and percent changes, both
seasonally adjusted and unadjusted, for all s o p groups
and for selected commodity groupings within these SOP
groups. Even though the release shows indexes for only
a limited number of series, all Producer Price Indexes
are available and officially considered published on the
day of the monthly news release.
The monthly detailed report, Producer Prices and
Price Indexes, is printed a few weeks after the news
release. It includes every publishable, not seasonally ad
justed index within the PPI complex, including industryoriented indexes from the p p i Revision program, as well
as some seasonally adjusted data. Prices for some in
dividual commodities calculated by the traditional
methodology are also published. An annual supplement
provides all publishable, not seasonally adjusted indexes
for the preceding calendar year, as well as tables of
relative importance data as of December of that year.
Tables of historical price indexes for any series are
available without charge on request; the Bureau does
Revised data. All p p i ’s are subject to revision 4 months
after original publication to reflect late reports and cor
rections by respondents. The news release and detailed
report print routinely revised data from the fourth
preceding month, as well as the current month’s in
dexes. Seasonally adjusted indexes for the preceding 5
years are also subject to revision early each year to
reflect the previous year’s impact on seasonal price pat
terns. Other revisions may be made infrequently, as
when all stage-of-processing indexes from January 1976
through December 1980 were revised in early 1981 to
reflect 1972 input-output relationships instead of the
1967 relationships on which the indexes had been based.
Other, infrequent revisions, including any corrections,
are also announced in the news release and detailed
report.
Uses and Limitations
Producer Price Indexes are used for many purposes.
The Finished Goods Price Index, the major focus of the
Bureau’s news releases and economic analyses of the p p i
48
since 1978, is one of the most widely cited indicators of
inflation in the overall economy. Fluctuations in this in
dex often presage changes in the Consumer Price Index
and the gross national product deflator, two other wide
ly followed measures of general inflation. Changes in
the Intermediate Goods Price Index frequently signal
similar changes to come in the Finished Goods Price In
dex. 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.
The state-of-processing structure thus facilitates
economic analysis of inflation transmission and
Government stabilization policies, p p i data are also used
in analyzing Government policies directed at specific in
dustries, such as energy and steel.
Private business firms use p p i data to assist their
operations in forecasting, market analysis, comparison
of costs paid for material inputs or prices received for
outputs with national averages, and so forth. One of the
most important uses of p p i data in recent years has been
in long-term sales and purchase contracts. In a 1976
survey of p p i data users, about half of the respondents
who used the p p i for escalation revealed that purchase
contracts worth nearly $100 billion were being escalated
with one or more Producer Price Indexes; the other half
declined to specify the dollar amounts.4 There is reason
to believe that the total then and today in the entire
economy is substantially higher than this amount.
Some p p i data, such as indexes for various kinds of
capital equipment, are used by the U.S. Department of
Commerce to help calculate the gross national product
deflator and many of its component deflators, p p i data
have also been used to deflate values expressed in cur
rent dollars to constant-dollar values for such time
series as inventories, sales, shipments, and capital
equipment replacement costs, for the total economy and
for industries and firms, p p i data are frequently used in
LIFO (last-in, first-out) inventory accounting systems by
firms wishing to avoid “ paper profits” which might ap
pear with a FIFO (first-in, first-out) system.
The Finished Goods Price Index can be used to
measure changes in the purchasing power of the dollar
in primary (but not retail) markets. This index, the Con
sumer Price Index, and the U.S. Department of
Agriculture’s index of prices received by farmers all
show relative changes from a base period; however,
comparisons among the levels of these indexes should
not be interpreted as a measure of the actual margins
between farm prices and manufacturing, or between
primary and retail markets.
The limitations of Producer Price Indexes calculated
by the traditional methodology are explained in detail in
Part II.
Industry and Produet Class
Price Indexes
An industry price index is a composite index con
sisting of price series that follow the general economic
pattern of a particular industry. It includes products,
sometimes of dissimilar types, grouped by industry of
origin. Thus, it differs from the traditional p p i com
modity indexes. An industry price index for a given in
dustry represents price indexes for a sample of the pro
ducts produced by that industry, averaged together ac
cording to the value of production of each sample pro
duct within the industry.
Industry price indexes use weights of gross shipments
of products “ made in the industry” for the deflation of
industry shipments. The product class indexes (used for
deflating Census product classes) are a set of output
price indexes of industry shipments classified by in
dustry, but weighted by shipments of the product pro
duced anywhere in the economy.
The Standard Industrial Classification (sic) system,
as revised in 1972, is used to define the scope of the in
dustry price index universe. Related products or services
are grouped together and assigned a 2-, 3-, or 4-digit in
dustry code according to the level of detail. Industry
price indexes are limited in scope to the value of primary
and secondary shipments. Total product shipments are
used (including interplant transfers), which inherently
include shipments to all customers and the value of ex
ported products.
The Sic provides the basis for the classification
scheme used in constructing industry price indexes.
Within this framework, individual products are given a
7-digit code by the Bureau of the Census. The product
indexes are then aggregated to 5-digit product classes.
Using these product-class indexes, 4-digit industry in
dexes are obtained using made-in-the-industry weights.
The Industry Price Index program has always
depended entirely upon price data primarily collected
for the p p i . Expansion of data for the industry price in
dex depended on the expansion of the p p i . This expan
sion was generally directed at those industries con
sidered to be most significant, based on such standards
as value of shipments, total employment, market impor
tance, etc. Under these general criteria, particular com
modities, specifications, and respondents were selected
judgmentally for the p p i and industry price indexes on
the basis of volume, market share, and price leadership.
Since January 1976, weights for the output indexes
are the 1972 value of shipments. Values include those
for interplant transfers, goods processed and consumed
in the same establishment, and goods sold for export;
values of imported commodities are not included. The
difference in the scope of the weights, as compared with
p p i commodity indexes, stems from the objective in this
4 Bureau of Labor Statistics, The bls Industrial Price Program: A
Survey o f Users, Report 509, 1977; and Escalation and Producer Price
Indexes: A Guide fo r Contracting Parties, Report 570, 1979.
49
modity indexes calculated with traditional method
ology. Because indexes from the p p i r program have
the advantages of industry price indexes without
the corresponding limitations, the Bureau is no longer
trying to expand the industry price index system as such.
Instead, the Bureau is phasing out this system by
eliminating publication of those industry indexes which
have been absorbed into the p p i r system.5
system to match price data with the scope of domestic
industry production.
The principal advantage of these price indexes has
been their industry orientation, an orientation shared by
many other economic time series. It has been easier and
more appropriate to use industry price indexes, rather
than p p i commodity indexes, in conjunction with data
on wages, productivity, employment and other series
that are expressed in terms of the sic and Census
product class extensions. At the same time, these in
dustry price indexes share the weaknesses of p p i com
5
For a more extended treatment of the traditional industry price
index methodology, see ch. 15 of the 1976 edition o f the Handbook o f
Methods, pp. 123-26.
Part II. Methodology of the Producer Price Index Revision
P pir indexes are classified according to the sic and,
therefore, are compatible with other industry-oriented
economic data, such as on employment, wages, and
productivity.
Imports are not priced for the p p i r , which is intended
to measure changes in the prices of the output of
domestic industries. Conversely, the share of domestic
industry’s output which is exported will be a part of the
p p i r . The traditional p p i system, aimed at measuring
changes in prices received in the first commercial transac
tion in this country, included imports and excluded ex
ports. The b l s International Price Program has assum
ed responsibility for collecting and publishing data on
import price changes. These data will eventually be in
corporated into the wherever-made product index
system.
Specifications fo r commodities priced for the p p i r
follow Bureau of the Census definitions and are con
siderably broader than most of the highly detailed and
narrow specifications for traditional p p i commodity in
dexes. Because companies in the p p i r are reporting
prices for a broader range of commodity and
transaction-term specifications within a given commodi
ty index, it is usually not possible to publish meaningful
average prices for individual commodities in the p p i r ,
as has been more frequently possible with the traditional
p p i . However, after the product sample is specified, a
detailed specification for each price report is followed
through time, p p i r indexes are usually calculated by
constructing an index for each reporting company’s
price and then averaging these company indexes to
derive the commodity index, rather than by taking an
average price directly from individual company prices.
In addition, averages of company price indexes in the
p p i r are usually weighted by the company’s relative
size, a departure from the traditional p p i practice of giv
ing equal weight to each reporter.
Indexes calculated with p p i r methodology emphasize
actual transaction prices at time o f shipment to
In 1978, BLS began publication of the first, pilot
group of indexes from the Producer Price Index Revi
sion program. This revision is the first comprehensive
overhaul of the entire theory, methods, and procedures
used by BLS to construct indexes measuring price
changes in nonretail markets.
Differences Between PPI and PPIR
Scientific (probability) sampling techniques are being
used instead of judgment sampling to select reporting
companies, products, and the price-determining trans
action terms. Companies of all sizes (not just larger
ones) are being asked to supply prices, and transactions
involving all kinds of output (not just volume-selling
items) are being priced each month. As a result,
statistical measures of index error and precision will
become available for the first time.
Coverage is being systematically expanded to include
all 493 sic industries in the mining and manufacturing
sectors; thousands of commodity indexes will thus
become available for the first time. Before the p p i r
began, indexes representing only about half of the total
value of mining and manufacturing output were incor
porated into the traditional p p i commodity system. Ex
panded coverage lessens the importance of imputations
of unpriced goods.
Indexes from the p p i r are industry oriented instead of
commodity oriented. The entire output of each industry
is sampled, including primary and secondary produc
tion and miscellaneous receipts. Traditional p p i com
modity indexes have been based on prices received by
producers without regard to the industry classification
of these producers. When data from more mining and
manufacturing industries are established with the p p i r ,
additional “ wherever-made” indexes will be con
structed for each product, regardless of the industry of
origin. In the meantime, p p i r indexes for products
made in the designated industries are being published.
50
minimize the use of list prices and order prices, which
occasionally have been used in traditional ppi commodi
ty and industry price indexes.
PPIR product indexes
Industries included in the ppir may be represented by
up to three kinds of product indexes. Every industry has
primary product indexes to show changes in prices
received by establishments classified in the industry for
products made primarily, but not exclusively, within
that industry. The industry under which an establish
ment is classified is determined by the Census coding of
those products which account for the largest share of its
total value of shipments. In addition, some industries
may have secondary product indexes to show changes in
prices received by establishments classified in the in
dustry for products primary to some other industry.
Finally, some industries have miscellaneous receipts in
dexes to show price changes in other sources of revenue
received by establishments within the industry which are
not derived from the sale of products. Because of the
distinction between primary and secondary products, an
index for a product made in one industry may differ
from the index for the same product made in another in
dustry.
Corresponding PPIR and PPI indexes
Some 7-digit Census products included in the ppir
correspond to 8-digit commodities published in the
traditional ppi commodity system. In this case,
movements in the traditional ppi ’s are governed by
movements of their counterparts in the ppir . Although
most such traditional ppi commodity indexes continue
to be published on their own original base period, the
corresponding indexes in the ppir are published on a
base of the month of their introduction. Therefore,
monthly percent changes for corresponding items will
be identical even when their respective index levels dif
fer.
The aggregation of traditional ppi commodity indexes
into commodity grouping indexes continues to follow
the traditional methodology; similarly, stage-ofprocessing price indexes are still calculated from com
modity grouping indexes. However, an entirely new
structure will eventually replace the traditional com
modity structure as the principal vehicle for releasing
and analyzing price changes at the primary market level.
One planned refinement is the split of the stage-ofprocessing index for intermediate materials into one for
primary processing and one for intermediate processing..
This change should enhance the analysis of the
transmission of price changes through the economy.
51
Economics of Producer Price
Measurement
The structure of the traditional PPI has not cor
responded to any meaningful economic construct; it has
been formulated in a rather ad hoc fashion without any
reference to a theoretical model. An appropriate
model—that of a fixed-input output price in
dex6
—derived from the theory of the firm has now been
developed and is being used in the ppir program.
Indexes based upon a fixed-input output price model
avoid one of the chief defects in the current aggregate
commodity indexes—multiple counting of price
changes. The All Commodities Index, as well as indexes
for major commodity groups, is calculated from price
changes of commodities at many stages of processing.
Each price change is weighted by its total gross value of
shipments in the base year. Suppose, as a simplified ex
ample, that the price of cotton were to rise sharply. If
this price increase were passed through by producers to
cotton yarn, then to gray cotton fabric, then to finished
cotton fabric, and finally to shirts, the price increase in
the raw material would have been included five times in
the All Commodities Index and four times in both the
Industrials category and the Textiles and apparel major
group. So long as prices for all items at all stages of pro
cessing are changing at about the same rate, this multi
ple counting produces no major distortion. But if there
is more rapid inflation in the prices for raw materials,
this multiple counting can produce aggregate rates of
price change that are misleading. More detailed indexes
which cover only a single stage of processing are free of
this defect.
The existing stage-of-processing indexes provide a
better basis for the analysis of price change than the
traditional major commodity grouping indexes,
although some sop indexes also suffer from some multi
ple counting. The Finished Goods Price Index and the
Crude Materials Price Index are both rather strictly
defined and are, therefore, largely free of multiple
counting problems. The Intermediate Materials Price
Index, however, is a residual category which still in
cludes several different stages within it—three such
stages in the shirt example.
The theoretically ideal output price index would allow
firms to adjust their outputs in response to relative price
changes. However, it is not feasible to collect and use
such changing output data in calculating ppi
movements. Instead, the ppi is a fixed-weighted, or
Laspeyres, index that approximates the theoretically
6
Robert B. Archibald, “ On the Theory of Industrial Price
Measurement: Output Price Indexes,” Annals o f Economic and
Social Measurement, Winter 1977, pp. 57-72; and Franklin M. Fisher
and Karl Schell, The Economic Theory o f Price Indexes (New York,
1972).
ideal index. Both the theoretical economic index and the
Laspeyres index measure the change in revenue from a
base period as the result of a change in prices. In the
economic index, the firm (or other entity) is allowed to
adjust its output in response to these price changes to
maximize its profit. As a result, it will have revenue at
least as large as (and possibly larger than) if it were con
strained to maintained relative proportions among its
outputs as in the Laspeyres index. The Laspeyres index
is the lower bound of the true economic output price in
dex.
duction in an industry (the production associated with
the definition of that industry), secondary production
and miscellaneous receipts are also priced. Thus,
refrigerators made in a home laundry equipment
establishment are part of the home laundry equipment
output price index. Prices received for the sale of scrap,
contract work, or property leasing are also included.
Technological change
The theoretical model for a fixed-input output price
index assumes no technological change. However, pro
duction technology does change, and new and improved
products are frequently introduced into the
marketplace. As a result, it is often impossible to obtain
prices for identical items over long periods of time. In
addition, at any given time, b l s is usually most in
terested in price changes of output that is in some sense
representative of current production. For most pur
poses, information on price change during the 1980’s in
outputs stemming from the technology of the 1940’s is
not very useful.
Three procedures are being used to deal with these
problems. One is a periodic substitution of the set of
particular commodities to be priced and the weight each
is to represent in the index. (In this context, a commodi
ty is defined as a class of similar goods or services at
about the same level of detail as represented by the Cen
sus o f Manufactures 7-digit product code or the p p i
8-digit commodity code.) Second is a periodic update of
the sample of companies that report prices to b l s . Once
the comprehensive p p i revision methodology has been
extended to encompass all industries in the mining and
manufacturing sectors, b l s will revise company and
item samples in one-fifth of the industries each year.
Finally, since individual items are frequently replaced
by new models, a replacement procedure is needed to
reflect a dynamic economy. This replacement procedure
is one of the most difficult methodological challenges in
price index construction. The process—usually called
quality adjustment—seeks to solve the following type of
problem: If a firm stops producing a model A-l
hydraulic crane which has a 30-ton lifting capacity and
replaces it with a model A-2 with a 32-ton lifting capaci
ty for which the company charges $5,000 more, how
much of that price increase represents the value of the
extra 2 tons of lifting capacity, and how much
represents “ pure price increase’’? The current
methodology for evaluating this change in specification
would be to value it at the cost of the change to the pro
ducer. There has been a substantial body of economic
literature dealing with this problem,7 and the Bureau
Net output
In an industry output price index, items are only
selected for pricing to the extent that they are actually
sold outside the industry. This is called a net output ap
proach to price measurement. In the shirt example, an
output price index for the cotton weaving mill industry
would contain only the outputs of the industry—in this
case, gray cotton fabric. An index for the combined tex
tile and apparel industry would reflect price changes
only for shirts; none of the other items actually leaves
the industry. Such fabric as does leave the industry, for
example, for home use, would be included in this latter
index, but any fabric used within the industry would be
excluded.
In constructing an output price index for a firm,
someone could sit at the plant gate and count the
amount and price of each item shipped. Because an in
dustry is a set of establishments making similar pro
ducts, an output price index could be constructed for
the industry by placing a fence around all the
establishments, and counting and pricing every transac
tion that passed through that fence. Transactions
among establishments inside the fence would not be
measured. Thus, for constructing an industry output
price index, weights would be used that reflect the new
output of each product, that is, total shipments of that
product by establishments in the industry to buyers out
side the industry. In the construction of an output price
index for an aggregate industry composed of several
smaller industries, the shipments among the component
industries would be excluded, and only those shipments
outside the aggregate would be included in the index.
Construction of net output weights for the p p i r in
dexes is more complex than the use of gross shipment
weights in the traditional p p i commodity structure. In
addition to the value of shipments data currently used
which have been supplied by the Census o f Manufac
tures, b l s uses the value of materials consumed (also
from the Census), data on detailed industry flows from
the input-output table constructed by the Bureau of
Economic Analysis, and other detailed industry data to
construct appropriate net output weights.
In preparing industry output price indexes, all sources
of revenue must be priced. In addition to primary pro
7
Jack E. Triplett, “ The Measurement of Inflation: A Survey of
Research on the Accuracy of Price Indexes,” in Paul H. Earl,
Analysis oflnflation (Lexington, Mass., 1975).
52
has been among those conducting research to improve
quality adjustment procedures.8
A system ®f prie® indexes
The output price index which has been described here
is the major organizing element for the work of the p p i r
program. It is the long-run objective to produce output
price indexes for every industry in the private economy.
The first stage, now well underway, is to include all pro
duction of both goods and services in the mining and
manufacturing sectors. The next phase will cover the
output of agriculture and contract construction. Final
ly, some work has already begun on calculating servicesector price indexes. Price indexes for railroad freight,
telephone services, and postal services are already being
published, and developmental work is proceeding on in
dexes for the life insurance industry. Development of
service industry indexes will be slow, but the goal is
eventually to cover this entire sector.
The overall relationships among these indexes as will
as with other derivative indexes can best be understood
in an input-output framework. Figure 1 is a schematic
input-output table. Along the left side are arrayed the
industries producing goods and services in the economy,
including imports. Along the top are the consuming
elements in the economy—both intermediate demand
(industry purchases of materials and services) and final
demand (composed of personal consumption expen
ditures, gross private domestic investment, net exports,
and government purchases).
Each cell of figure 1 represents the value of sales from
the producing industry to the consuming element. For
example, cell A contains the value of sales from
manufacturing industry 2 to mining industry 1; cell L
represents the value of sales from manufacturing in
dustry 2 to personal consumption expenditures.
An output price index for manufacturing industry 2
8 Jack E. Triplett and Richard J. McDonald, “ Assessing the Quali
ty Error in Output Measures: The Case of Refrigerators,” Review o f
Income and Wealth, June 1977.
Figure 1. Schem atic input-output table
53
will cover all of the transactions in cells A through O,
except cell E, which represents the value of sales within
the industry. The Standard Industrial Classification is
used to classify each producing and consuming in
dustry. As noted earlier, the PPI revision covering min
ing and manufacturing will produce output price in
dexes for each 4-digit sic industry in those two sectors, a
total of 493. Each industry is being measured by its own
survey, and industry indexes are being published as they
become available. As of January 1982, output price in
dexes were being published for 114 industries, account
ing for about one-third of the total value of the output
of the mining and manufacturing sectors.
Various aggregate output price indexes can be for
mulated. Output price indexes for aggregate 3- and
2-digit sic industries can be constructed which price all
transactions originating in the industry, except those
transactions within the aggregate industry. For exam
ple, an output price index of the combination of
manufacturing industries 1 and 2 would exclude
transactions in cells D, E, S, and AA. Similarly, ag
gregate output price indexes can be constructed for min
ing, for manufacturing, and for the two sectors com
bined. When all agriculture, construction, and service in
dustries are completed, it will be possible to construct an
aggregate output price index for all production in the
U.S. economy. The currently published Finished Goods
Price Index roughly corresponds in concept to an index
for items originating in the mining, manufacturing, and
agricultural sectors.
An input-output table, also allows the inputs to an in
dustry to be identified. Pricing the transactions in cells
P through Z (E excluded) allows the construction of an
input price index for manufacturing industry 2 covering
its materials and purchased services. An index of
material input prices could be constructed from cells P
through W and the goods portion of Z.
Fundamental to constructing both input and output
price indexes are the detailed commodity indexes of
items made in or consumed by an industry. A given
stages could also be divided into other categories of in
terest, such as food, nonfood, fuels, consumer goods,
and capital equipment.
The existing p p i stage-of-processing indexes are
commodity-based, not industry-based, and, therefore,
differ from those described here. The existing Finished
Goods Price Index corresponds approximately to transac
tions in figure 2 in cells R through Z, AA through CC,
G through J, and DD through GG. There is currently no
input price index for the finished goods sector. It is im
plicitly assumed that intermediate materials are the in
put, but there are difficulties with that assumption: (1)
The intermediate materials price index, as presently
calculated, has substantial multiple counting because it
includes sales within the intermediate stage; and (2) in
puts to the final production sector may also come from
other sources.
commodity may be produced as a primary product in
one industry and as a secondary product in several dif
ferent industries. The combination of indexes for the
same commodity, irrespective of its source, will give
detailed commodity indexes. About 6,000 of these
wherever-made indexes eventually will be available for
mined and manufactured commodities.
Finally, data from the p p i r program can be arranged
to form new stage-of-processing indexes, with an in
dustry rather than a commodity basis.9 The first step
in developing new SOP indexes is to rearrange the inputoutput table so as to minimize the amount of
“ backflows” —transactions in which the buyer is in an
earlier stage of processing than the seller. Ideally, once
the table is rearranged, industries would sell only to
other industries to the right of them on the table. In
practice, some backflows will remain. The existence of
these backflows does not reduce the quality of the
resulting price indexes but makes analysis a little more
involved. Once the 4-digit industries are rearranged,
they can be grouped into summary s o p industries.
Figure 2 illustrates how that might be done.
Just as was done for the Sic classification approach,
an output price index for the finished goods sector could
be constructed from transactions in cells A through J.
The shaded cells represent within-industry sales. A
materials input price index for finished goods industries
can be constructed from transactions in cells K, M, N,
and O. Other input and output price indexes could be
constructed for each stage of processing. Each of these
Typ© of pries
9 Joel Popkin, “ Integration of a System of Price and Quantity
Statistics with Data on Related Variables,” Review o f Income and
Wealth, March 1978; “An Integrated Model of Final and In
termediate Demand by Stage of Process: A Progress Report,” Pro
ceedings o f the American Economic Association, February 1977; and
“ Consumer and Wholesale Prices in a Model of Price Behavior by
Stage of Processing,” Review o f Economics and Statistics, November
1974.
Clearly, when b l s prices the output of a production
unit, the price needed is that which the producer actual
ly received for selling the item—the transaction price.
This is not a change from the traditional p p i program
(with one exception to be noted later). As in the tradi
tional p p i , the desired price for the p p i r reflects all ap
plicable discounts, extras, and surcharges. The price is
f.o.b. the seller’s freight dock and excludes all direct ex
cise taxes and transportation charges. There is increased
emphasis on pricing all the different types of transac
tions in which a particular item may be sold: Items may
be sold “ off the shelf,” by a single delivery contract, or
by a multiple delivery contract. It is important to price
all these types of sales at the time o f shipment. The time
of pricing is very critical, especially because of the in
creasing use of escalator clauses that adjust the final
price in long-term sales and purchase contracts.
One major area in the traditional p p i in which clear
Figure 2. Idealized stage of processing input-output table
Consuming industry
Producing industry
Services
to
business
Final demand
Inter
Finished
Crude
Primary
mediate
materials processing processing goods
Other
service
Con
struction
Personal
con
Gross
sumption
private
expend investment
itures
Exports
Govern
ment
purchases
EE
FF
GG
Imports......................................
K
DD
Services to business ................
L
HH
Crude m aterials........................
M
R
S
T
U
Primary processing..................
N
V
W
X
Y
Intermediate processing..........
0
Z
AA
BB
CC
G
H
I
J
Finished goods..........................
A
B
C
D
E
F
Other services..........................
P
II
Construction ............................
Q
JJ
54
shipment prices are not obtained is machinery, where
prices reported for major equipment sales are, for the
most part, order prices for later delivery. While these
usually represent prices at which orders are actually
received, they do not necessarily represent the price at
which the shipment finally occurs, since the order may
not be filled or the final price may be escalated. For
measuring current price change, for deflating produc
tion or inventories, and for constructing models of price
change, shipment (transaction) prices are required.
In some industries, the list price and the transaction
price are the same. The same physical item may be sold
by its producer at different prices, depending on the cir
cumstances of the transaction. Long-term contracts
have different prices from spot sales; large orders, dif
ferent from small; preferred customers, different from
occasional. All of the conditions of the sale must be
identified and priced through time, a procedure consis
tent with the theory of the index in that the technologies
of production and distribution will change if order sizes
or leadtimes in production are changed. The index
assumes constant technology. One method used by
others to obtain transaction prices is the average real
ized price, or unit value procedure, in which the total
revenue for an item is divided by the total number sold.
However, data are frequently not available for a unique
item, but rather for an entire class. For example, a unit
value index for nails might decline as production shifted
to smaller, cheaper sizes, even though the price for each
individual size rose. If this problem is overcome by the
use of sufficiently detailed data, there is still a
“ customer mix” problem: If the proportion of sales to
large customers were to rise, the average price across all
customers might fall, even if the price to each customer
rose. For these reasons, the Bureau tries to avoid resort
ing to the average realized price approach.
Another suggested procedure is to obtain transaction
prices from the buyers. Buyers’ prices have been tested
by Stigler and Kindahl and have been used by bls in
some series.1 While buyers’ prices may continue to be
0
used in special cases, they are not usually the best ap
proach. In the first place, buyers surveyed may report
prices paid at a different stage of distribution, with
changes in taxes, transportation, insurance, and
distributor markups being mixed with the price change.
In addition, the prices reported by a buyer may be from
different manufacturers over time, reflecting the
buyer’s minimization of input costs. This violates the in
dex assumptions of fixed production technology anc
maximization of production revenue. If the buyers who
buy directly from the manufacturer are identified and
surveyed, and if they report on prices from the same
1 George J. Stigler and James K. Kindahl, The Behavior of In
0
dustrial Prices (New York, National Bureau of Economic Research,
1970).
manufacturer, these problems conceivably could be
overcome. However, the Bureau’s operational costs
would be very high. There is no universe list of the
buyers of an industry’s output from which a sample can
be drawn. Except in those few instances where the
sellers of a product outnumber the buyers, it is con
siderably more expensive to obtain the same precision
from a buyers’ price index as from a sellers’ price index.
Sample Designi
In order to produce the described indexes with a
reasonably small expenditure of money, it is necessary
to price only a small number of the transactions through
which an industry sells its output. In the traditional p p i ,
this procedure of selecting items to be priced was “ pur
posive.” Major product areas were selected for pricing,
and representative products were selected within these
areas. Usually, the largest producers were selected to
report on these products, with some smaller ones
sometimes added. Each company was then asked to
report on the volume-selling variety of that product.
This procedure resulted in an index heavily composed of
volume sellers by major producers.
The industrial price revision is aimed at systematically
producing output price indexes and detailed product in
dexes for each of the 493 4-digit sic industries in mining
and manufacturing. Each industry has an individually
designed sample. The first step in selecting the sample is
to construct a list of establishments in the industry. The
primary source for this list is the data of the Unemploy
ment Insurance System, since almost every employer is,
by law, a member. Supplementary information is ob
tained from multiple, publicly available lists.
An establishment is a production entity in a single
location. Two establishments may occupy the same or
adjacent space if they are separable by physical iden
tification or recordkeeping, or both. Establishments are
the units for which production and employment data
are collected. However, establishments are frequently
not the proper unit for the collection of price data. As
discussed earlier, the model for an output price index
assumes a profit-maximizing production unit. An
establishment may be one of several owned by a single
company engaged in similar production. This group, or
cluster, of establishments often constitutes a profit
center in the company; in such a case, profits are max
imized over the cluster as a whole. The outputs of the
establishments are undifferentiated and are priced
homogeneously. The second step in sampling consists of
clustering establishments into price-forming units that
conform as closely as possible to the theoretical concept
of the firm. The major deviation from this concept is
that establishments in a real-world profit center may be
classified in different industries. To meet the need for
industry indexes, the members of a price-forming unit
cluster must all belong to the same industry.
Once a list of price-forming units in an industry is
constructed, it is stratified by variables that are ap
propriate for that industry. The criterion for identifying
the sampling strata is whether price trends may be dif
ferent 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 and input costs. Some industries
are characterized by geographically independent
markets, which should be strata. Other characteristics
may also lead to strata, such as the presence of
cooperatives in the industry. Within each stratum, units
are usually ordered by geography or size, or both, to en
sure a proportionate distribution of the sample. The
more successful the identification of the proper strata,
the lower is the sampling error of the estimated indexes.
The next 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, they wili
be assigned a greater proportion of the sample than
their value would require. As a further step to reduce
the sampling error of the estimate, each price-forming
unit is selected systematically with a probability propor
tionate to its size. The proper measure of size would be
the total revenue of the unit; however, because that in
formation is not contained in available universe lists,
employment is used as a proxy. As an overall proxy for
revenue, employment is rather poor. However, it is
usually necessary to use it only within strata within in
dustries; as a result, the production technologies should
be relatively homogeneous, and the proxy fit should be
fairly good.
Once an establishment or cluster of establishments is
selected for pricing, a bls field economist visits the
reporting unit to select a few transactions to be priced
through time from among all activities which produce
revenue. To avoid bias and to permit estimates of the
precision of the indexes, a probability technique called
disaggregation is used.1 This multistage procedure
1
assigns to each category of items shipped and to each
category of other types of receipts a probability of selec
tion proportionate to its value within the reporting unit.
The categories selected are broken into additional detail
in the subsequent stage until unique items or unique
types of other receipts are identified.
Even after a unique item is selected, it may be
necessary to disaggregate further. If the same item is
sold at more than one price, then the conditions that
determine that price—such as the size of the order, the
11 b l s successfully developed and used this technique in its revi
sion of the Consumer Price Index.
56
type of customer, or whether it is a spot or contract
sale—must also be selected on a probability basis. This
method for identification of terms of sale serves two im
portant functions: It ensures that the same type of trans
action is priced through time, and it eliminates any bias
in the selection of the terms.
There are a number of practical problems that must
also be dealt with in the execution of this survey. One is
company cooperation, which remains completely volun
tary, as it has always been in the traditional p p i pro
gram. Companies that cooperate with the program
agree to one interview and disaggregation process which
averages about 2 hours. They also agree to supply prices
for the items selected every month (b ls 473P, shown at
the end of this chapter). In general, companies asked to
supply prices for the p p ir program have been quite
cooperative, with only about 20 percent of these firms
declining to participate. The most serious effect of non
cooperation is to increase the bias in the indexes. In ad
dition, not all cooperating companies have been as time
ly and as accurate in their price reports as the firms par
ticipating in the traditional p p i program. Because of this
and the fact that publication criteria in the p p ir are
often more difficult to meet than in the traditional pro
gram, a higher proportion of indexes may not be
publishable in any given month.
Sometimes the data required for the index are not
available for the sample unit (cluster or establishment)
but only for some larger aggregation, such as a com
pany division. If the larger aggregation (called a report
ing unit) is classified totally within the industry being
sampled, the report can be made from it, with some ad
justments to the probability of selection. However, if
the reporting unit includes establishments classified in
other industries, some adjustments can be made, but
some nonsampling error will remain. Because individual
establishments must report value-of-shipment data for
the Census o f Manufactures, at least the value-ofshipment data required for proper weighting of the in
dex could be obtained for each sample unit.
After the items and terms of sale have been selected,
the reporter supplies to bls each month the transaction
price for that item and its terms of sale effective for the
Tuesday of the week including the 13th. These data are
combined with data from other reporters in the same in
dustry to produce the Laspeyres fixed-weight industry
output price index, as well as commodity price indexes
and stage-of-processing price indexes.
The reporter also supplies descriptions of any changes
that may occur in the physical specification or terms of
sale that are available. Such information is used to
evaluate these changes so that comparisons are made
only between comparable items and transaction terms.
Incremental model changes to essentially the same item
can readily be made in this manner; radical changes in a
reporter’s production may require that some of the
approxim ately 4 w eeks after the pricing date. T hey p ro
vid e a m uch m ore com prehensive and accurate picture
o f industrial price changes than has ever b efore been
available.
item s reported be reselected if w hole categories o f item s
are n o longer produced.
A s is the case w ith the traditional P roducer P rice In
d exes, the new price indexes are being released m onthly,
Technical References
Archibald, Robert B. “ On the Theory o f Industrial Price
Measurement: Output Price Indexes,” Annals o f Eco
nomic and Social Measurement, Winter 1977.
National Bureau of Economic Research. The Price Statistics
o f the Federal Government: Review, Appraisal, and
Recommendations. Washington, D .C ., nber General
Series, Number 73, 1961.
Clorety, Joseph A ., Jr. “ Measuring Changes in Industrial
Prices,” Monthly Labor Review, November 1970.
Popkin, Joel. “ Integration of a System o f Price and Quantity
Statistics With Data on Related Variables,” Review o f
Income and Wealth, March 1978; “ An Integrated
Model o f Final and Intermediate Demand by Stage of
Process: A Progress Report,” Proceedings o f the
American Economic Association, February 1977; and
“ Consumer and Wholesale Prices in a Model o f Price
Behavior by Stage o f P rocessing,” Review o f
Economics and Statistics, November 1974.
Council on Wage and Price Stability. The Wholesale Price
Index, June 1977.
Early, John F. “ Improving the Measurement of Producer
Price Change,” Monthly Labor Review, April 1978.
Early, John F. “ The Producer Price Index Revision: Overview
and Pilot Survey Results,” Monthly Labor Review,
December 1979.
Stigler, George J., and Kindahl, James K. The Behavior o f
Industrial Prices. New York, National Bureau of Eco
nomic Research, 1970.
Fisher, Franklin M ., and Schell, Karl. The Economic Theory
o f Price Indexes. New York, 1972.
Triplett, Jack E. “ The Measurement o f Inflation: A Survey of
Research on the Accuracy o f Price Indexes,” in Earl,
Paul H ., Analysis o f Inflation. Lexington, Mass., 1975.
Howell, Craig. “ Changes in the Presentation and Analysis of
Price Movements at the Producer Level,” Statistical
Reporter, January 1979.
Triplett, Jack E., and McDonald, Richard J. “ Assessing the
Quality Error in Output Measures: The Case o f Refri
gerators,” Review o f Income and Wealth, June 1977.
Howell, Craig, and Thomas, William. Escalation and Pro
ducer Price Indexes: A Guide fo r Contracting Parties,
Report 570. U.S. Department of labor, Bureau of
Labor Statistics, 1979.
U.S. Congress, Joint Economic Committee. Government
Price Statistics. Hearings: Subcommittee on Economic
Statistics, 87th Congress, 1st sess., Part 1, Jan. 24,1961;
Part 2, May 1-5, 1961.
The b l s Industrial Price Program: A
Survey o f Users, Report 509. Bureau of Labor Statis
Moeller, Dereck.
tics, 1977.
57
Form Approved
O.M.B. No. 44R0194
MJ 473
(Rev. Jan. 1974)
U.S. DEPARTMENT OF LABOR
Bureau of Labor Statistics
Washington, D.C. 20212
INFORMATION FOR THE WHOLESALE PRICE INDEX
ALL REPORTS WILL BE HELD IN CONFIDENCE
Dear Sir:
H ie price data which you provide is used in com puting the Wholesale Price Index which is the
officially accepted indicator o f primary market price m ovem ents. The index is widely used by
industry and government.
These voluntary reports, submitted by you and other businessmen, are the major source o f infor
mation used in preparing this index. The information you provide is strictly confidential and open
to inspection only to sworn em ployees o f the Bureau o f Labor Statistics.
Please use the enclosed envelope, which requires no postage, for returning this schedule.
continued cooperation is greatly appreciated.
COMMISSIONER OF LABOR STATISTICS
IMPORTANT INSTRUCTIONS
In the boxes provided on the other side, please be sure to indicate all changes in
COMMODITY DESCRIPTION, BASIS OF QUOTATION, DISCOUNTS, ALLOWANCES, AND TAXES
that may have occurred since your last report.
Your cooperation in keeping all information current
is a great aid in computing a reliable, accurate Wholesale Price Index.
Remarks
58
Your
INFORMATION FOR THE WHOLESALE PRtOE INDEX
_________________________ ____ ______________________________ ________________________
1. COMMODITY DESCRIPTION (Please indicate ail changes.) — ------ ---
Code No.
CHANGES
Give date, nature, and
estimated value o f change
Date and nature o f change
2. BASIS OF QUOTATION (Please indicate all changes.)
Unit
C la s s o f s e lle r a n d c u s t o m e r
S iz e o f o r d e r
S h ip p in g t e r m s
O th e r ( S p e c i f y )
3. DISCOUNTS, ALLOWANCES, AND TAXES Indicate all discounts, allowances, and taxes applicable to above-basis oLquotation. This information is needed to arrive at the ACTUAL SELLING PRICE. (Please indicate all changes.)
YES
Date and nature of change
NO
Quantity discount
Trade discount
Have any indicated
► discounts been deducted
from the reported price?
Cash discount
Seasonal discount
Other discount
Have any o f these
been included?
Other charges
Excise taxes
4. PRICE INFORMATION For the commodity described in item 1, please enter below the current price for the date indicated, on the basis quoted in
item 2.
r ............
P R IC IN G
PRICE
DATE
Price as of Sept. 9,1975
DATE OF PRICE
CHANGE (I f a n y )
PRICING
DATE
PRICE
Oct. 14, 1975
Apr. 13,1976
Nov. 11, 1975
May 11,1976
Dec. 9,1975
June 15,1976
Ja n .13,1976
July 13,1976
Feb. 10,1976
Aug. 10,1976
Mar. 9,1976
DATE OF PRICE
CHANGE (I f a n y)
Sept. 14,1976
r
L
n
J
PERMANENT OFFICE RECORD
KINDLY RETURN
THIS FORM PROMPTLY
Information for the
Producer Price Indexes
U.S. Department of Labor
Bureau of Labor Statistics
600 E Street, N.W. Code 47
Washington, D.C. 20212
T h is f o r m w ill b e r e a d b y a n o p t ic a l s c a n n e r . It is im p o r ta n t th a t y o u re a d
GENERAL INSTRUCTIONS:
b o th s e c tio n s a n d f o llo w th e
in s tr u c tio n s c a re fu lly . T h e re ve rse o f th is fo rm p ro v id e s a d d itio n a l in s tr u c tio n s . If y o u h a ve a n y q u e s tio n s ,
p le a se c a ll
c o lle c t a t
T h e in fo r m a tio n in S e c tio n I w a s p ro v id e d b y y o u r c o m p a n y . P le a se d e te rm in e if it is c u r re n tly a p p lic a b le o r re q u ire s c h a n g e s . If
a n y c h a n g e s a re n e c e s s a ry , p le a se c ro s s o u t th e in c o rr e c t p o r tio n s a n d w r ite th e n e w in fo r m a tio n in th e c lo s e s t o p e n a re a w h ic h
is e ith e r b e lo w o r t o th e r ig h t o f th e c h a n g e .
s
Product and Transaction Description: I f y o u m a k e a n y c h a n g e s in th is d e s c r ip tio n , p le a s e a ls o in d ic a te th e d a te th a t t h is 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 c h a n g e ( c h a n g e in p r o d u c tio n c o s t p lu s s ta n d a rd m a rk -u p ).
E
C
T
S
O
N
Previous Price Information: T h e se a re th e p ric e s th a t y o u h a v e s u p p lie d in p re v io u s m o n th s . If th e p r ic e is b la n k f o r a n y o f th e
d a te s lis te d , th e n p le a s e e n te r o n e if it is a v a ila b le . A ls o , in th e s p a c e p r o v id e d , c o r re c t a n y in c o r r e c t p ric e s th a t a re s h o w n .
The PRICE was
$
$
$
this DATE
CORRECTIONS
on
on
on
The PRICE was
$
$
$
this DATE
CORRECTIONS
on
on
on
Net transaction prices a re th e m o s t d e s ira b le ty p e s o f p ric e s .
T h e ty p e o f p r ic e th a t y o u n o w re p o r t is
If t h is is in c o rr e c t, o r if y o u c h a n g e t o a d iff e r e n t ty p e o f p ric e , p le a s e in d ic a te th e c u r re n t ty p e o f p ric e .
Yes
IMPORTANT!
Please Answer BOTH Questions
Black Pen/Pencil Is Preferred
I□
□
2. Did the price change betw een
I□
□
and
Yes
s
E
C
T
S
0
N
No
1. Have you m ade any changes or entries anywhere in Section I?
Complete Section II ONLY i f you answered YES to Question 2.
No
For BLS Use Only
Current Price Information:
If th e re w a s a s h ip m e n t b e tw e e n
and
p le a s e re p o r t
th e p r ic e f o r th e la s t s h ip m e n t d u r in g th a t p e rio d .
If th e re w a s n o t a s h ip m e n t, p le a se e s tim a te th e p r ic e t h a t y o u w o u ld h a ve c h a rg e d
P le a s e e n t e r t h e p r i c e b e lo w w r it in g t h e n u m b e r s lik e th is
IU2l3N5lbl7l8N0l .
Dollars
Is th is a closeout p r ic e o n a p r o d u c t w h ic h is b e in g p h a s e d o u t ? .
PLEASE RETURN THIS FORM WITHIN 5 BUSINESS DAYS EVEN IF THERE ARE NO CHANGES.
60
only.
Do n o t type.
U s e b la c k p e n lp e n c il
Cents
Bureau of Labor Statistics
Information for the Producer Price Indexes
The inform ation collected on this form by the Bureau
of Labor Statistics w ill be held in the strictest confidence and w ill be used fo r statistical purposes only.
U.S. Department of Labor
This report is authorized b y la w 2 9 U.S.C. 2. Your
voluntary cooperation is needed to m ake the results
o f this survey comprehensive, accurate, and tim ely.
Form Approved
O.M.B. No. 1220-0008
Approval expires 12/31/83
Dear Respondent,
Thank you for your continuing participation in the Producer Price Index (PPI) program. The data which you provide are used in com
puting the Producer Price Indexes and constitute the basis for analyzing industrial price changes.
Please use the enclosed postage free envelope to return this form. Your continued cooperation is greatly appreciated.
Commissioner of Labor Statistics
instructions for completing this form:_________________________________________________________________________________________
SECTION I
The information on the reverse side of this form in Section I was furnished by your firm in previous pricing periods. Review the informa
tion carefully to verify that it remains current. Cross out any incorrect information and write in all corrections and additions that are
necessary.
Product and Transaction Description
If either the Product Description or the Transaction Terms, or both, no longer apply, a substitute product or substitute transaction
terms should be selected by you. Product substitution should only occur when the product previously reported is no longer available
because it is being or has been permanently discontinued. The substitute product should be as similar as possible in materials and
production technology to the current product, and should be expected to remain in production for some time. The substitute trans
action terms should likewise be as similar as possible to the discontinued transaction terms.
Report these changes where necessary in the space provided in Section I, and provide current price information in Section II.
Adjustments to Price
Following is a list of the more common adjustments to price. The specific adjustments on the reverse side o f this form were selected
originally and should be changed only when either the level of an existing adjustment changes or a new adjustment becomes applica
ble to the product and transaction described.
Deductions from price include:
1.
2.
3.
4.
Standard discounts (Cash, Seasonal, Cumulative Volume, Quantity, and Trade)
Rebates
Other recurring discounts
Other nonrecurring discounts (Competitive and Negotiated)
Additions to price include:
1. Surcharges
2. Other recurring charges added to price
3. Other nonrecurring charges added to price
Taxes should always be excluded from the price. If this exclusion is not possible, note this in Remarks.
Freight charges should be excluded from the price, unless delivery was selected originally as part of the product. Make changes if the
currently described freight terms no longer exist.
QUESTIONS
Answer Question 1, which refers to Section I. Mark 'X ' in either the Yes or the No box depending on whether you have ('YES') or have
not ('NO') made any changes or entries within Section I.
For Question 2, mark 'X ' in either the Yes or the No box depending on whether the shipment price of the product described in Section I
changed ( YES') between the two dates listed or whether the shipment price did not change ('NO') during this time period.
If the answer to Question 2 is NO, the form has been completed and is ready for mailing. If the answer to Question 2 is YES, please con
tinue to Section II.
SECTION I I
The preferred price is a net transaction price of an actual shipment made during the pricing period and as near to the pricing date as pos
sible. The price should reflect all deductions and additions to the price, including competitive price reductions that reflect current market
conditions.
Please complete and return within 5 business days all o f the forms that are mailed to you, even i f there are no changes.
I f you are anticipating a change in any o f the information that you provide, please indicate in Remarks. List the anticipated change and
when i t w ill occur.
Any questions that you have regarding this form o r its completion may be resolved by calling the Bureau o f Labor Statistics, Division o f
Industrial Prices and Price Indexes, Washington, D.C.; call collect by using the telephone number at the top o f the reverse side o f this form.
BLS 473P (Rev. Sept. 1981)
61
©ihupfeir §= International
Price lotax©®
price trends of other countries. Indexes are not prepared
for s it c categories with small values of trade.
Sa©kgr®yndl
The International Price Program (i p p ) provides in
dexes of prices for U.S. exports and imports. The pro
gram grew out of a longstanding need for accurate
measures of price changes in the foreign trade sector of
the U.S. economy. In the period immediately following
World War II, b l s began to develop export and import
price indexes. The program was terminated in 1948 due
to budget reductions and lay dormant until 1967 when
b l s began research on the feasibility of reintroducing
export and import price indexes.1This research resulted
in funds being allocated for the program in FY 1970.
Export price indexes were first published in 1971 and
import price indexes in 1973. Data were collected an
nually for June of each year until 1974 when collection
and publication began on a quarterly basis. Coverage as
of June 1982 accounted for 71 percent of the value of
exports and 96 percent of the value of imports. Plans
are to extend coverage to all exports and imports of
nonmilitary products by late 1983.
Prodoet universe
The product universe of the export price indexes con
sists of all products sold by U.S. residents to foreign
buyers. (“ Residents” refers to the national income ac
count definition; it includes corporations, businesses,
and individuals but does not require either U.S. owner
ship or U.S. citizenship.) The product universe of the
import price index covers all products purchased from
abroad by U.S. residents. The universe in the case of
each of these indexes includes raw materials,
agricultural products, semifinished manufactures, and
finished manufactures, including both capital goods
(electrical machinery, agricultural equipment, textile
equipment, etc.) and consumer goods (appliances, elec
tronic equipment, clothing, etc.).
Military goods are not priced in the indexes except to
the extent that some products may be purchased on the
open market for military use; e.g., automobiles,
clothing, nonspecialized hardware, fuel, etc. A few
items (works of art, ships, etc.) are not included because
of the difficulty of obtaining time series for comparable
products in their categories.
Dtsseiriiptta ® Sorway
fi
C®n©@pts
The export and import price indexes being developed
cover transactions in nonmilitary goods between the
United States and the rest of the world. The export price
index provides a measure of price change for U.S. prod
ucts sold to other countries. The import price index pro
vides a measure of price change for goods purchased
from other countries by U.S. residents.
In addition to general indexes of prices for U.S. ex
ports and imports, indexes are also published for detail
ed product categories of exports and imports. These
categories are defined by the 4-digit level of detail of the
Standard Industrial Trade Classification System ( s it c ).
The calculation of U.S. export and import price indexes
by s it c category facilitates the comparison of U.S. price
trends and sector production with the export and import
P itas
1 This research was among the early major products o f the Division o f Price
and Index Number Research established in 1966 on the recommendation o f the
Price Statistics Review Committee o f the National Bureau o f Economic
Research.
62
Prices are collected according to the specification
method. To the extent possible, they refer to prices at
the U.S. border for exports and at both the foreign
border and the U.S. border for imports. For nearly all
products, the prices refer to transactions completed dur
ing the first 2 weeks of the third month of each calendar
quarter. If a firm had no transactions in a product dur
ing the 2-week period, prices for a transaction up to 2
weeks earlier or later may be used.
Respondents 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 price for
which the product was bought or sold. During the
quarterly repricing process, respondents are reminded
of this requirement through a combination of personal
visits, telelphone calls, correspondence, and special
enclosures with the reporting forms.
For the export price indexes, the preferred pricing
basis is f.a.s. (free alongside ship) U.S. port of exporta
tion. When firms report export prices f.o.b. (free on
board), production point information is collected which
enables the Bureau to calculate a shipment cost to the
specified port of U.S. exportation. This information in
cludes location of production point and port of exporta
tion, size and weight of shipment, name of carrier, and
routing. For finished manufactures, respondents fre
quently report export prices on an f.o.b. factory basis.
As a result, many of the export price indexes are
published on this basis pending conversion to an f.a.s.
basis.
For imports, two prices are collected. The first is the
import price f.o.b. at the foreign port of exportation.
This is consistent with the basis for valuation of im
ports in the national accounts. The second is the import
price c.i.f. (cost, insurance, and freight) at the U.S. port
of importation. 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. The import duty on the product, if any, is col
lected as a separate piece of information.
Since a price index depends on the same items being
priced from period to period, it is necessary to recognize
when a product’s specifications or terms of transaction
have been modified. The specifications collected for
each product include detailed descriptions of the
physical and functional characteristics of the product.
The terms of transaction include information on the
number of units bought or sold, discounts, credit terms,
packaging, class of buyer or seller, etc.
When there are changes in either the specifications or
terms of transaction of a product, the dollar value of
each change is deleted from the total price change in
order to obtain the “ pure” change. Once this value is
determined, a linking procedure is employed which
allows for the continued repricing of the item.2
Average prices are not published or calculated
because even within the narrowest category of prod
ucts, i.e., 7-digit level of detail, products vary among
respondents.
However, when prices are collected, the products are
classified by the basic product classification systems for
recording U.S. foreign trade: For exports—the 7-digit
Schedule B classification system of the U.S. Department
of Commerce;4 for imports—the 7-digit Tariff Schedule
o f the United States Annotated ( t s u s a ) . 5 Concordance
schemes are used for classifying Schedule B or t s u s a
categories into the s i t c . 6 By maintaining the detailed
7-digit product classification, it is possible to prepare in
dexes for analytical purposes for groupings other than
those afforded by the s i t c ; e.g., sic and end-use.
Data Sources and Calculation Methods
Prie@s
Price data are collected quarterly by mail question
naire and reporting is voluntary and confidential. In
nearly all cases, price data are collected directly from
the exporter or importer, although in a few cases, prices
are obtained from brokers. Price reporting by firms is
initiated by a visit from a Bureau representative, at
which 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. No
index is published in such a way as to reveal the name,
price, or price behavior of any respondent. In all
published indexes, the number of respondents is greater
than the minimum of three established for this purpose.
Information initially provided by a firm usually con
tains data for the current and previous quarter. Subse
quent current prices are collected quarterly by mail.
Telephone contact is maintained each quarter with the
person at the firm who is responsible for providing the
price information. In addition, respondents are revisited
periodically in order to review reporting practices and
requirements and to reselect products for which prices
are reported. Frequently during these revisits, some
products are dropped from further reporting and new
items are added.
Sampling
The objective of the i p p sample design is to provide an
unbiased measure of price change in each published in
dex. 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 repeatedly
priced over time. The survey design is responsive to both
cost and reporter burden constraints. The cost con
straints impose limits on the number of distinct estab-
Glassification
Published indexes are described by the nomenclature
of Revision 2 of the s it c system of the United Nations.3
The s it c is made up of 10 sections at the 1-digit level, 63
divisions at the 2-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
SITC.
4 U .S. Bureau o f the Census, Statistical Classification o f Domestic and
Foreign Commodities Exported From the United States, Schedule B, Jan. 1,
2 The linking procedure is similar to that used for the Producer Price Index.
(See chapter 7.)
3 United Nations Statistical Office, Standard International Trade Classifica
tion, Revised, Statistical Papers, Series M, No. 34/Rev. 2 (New York, United
Nations, 1975).
1978, edition and revisions.
5 U .S. International Trade Commission, Tariff Schedule o f the United States
Annotated, 1980.
6 U .S. Bureau o f the Census, U.S. Foreign Trade Statistics Classifications and
Cross Classifications, 1980.
63
lishments that are selected in a sample, while the
number of items priced in each establishment is con
trolled to limit respondent burden.
The two universes for the i p p are all exporters and all
importers and their respective products. A sampling
frame for each universe is constructed from all docu
ments filed during a specified reference period (gen
erally 1 year). In the case of exports, these are the Ship
per’s Export Declarations ( s e d ’s) and, in the case of im
ports, the Consumption Entry and Warehouse With
drawal Documents. These documents contain brief
product descriptions, 7-digit product classification
codes, value, quantity (where required), date, origin or
destination, company name and address, and, in the
case of importers, a company identification code.
The import frame, obtained from the Customs
Bureau, contains a record of every import transaction.
The availability of a coded company identifier on these
records makes it possible to incorporate frequency of
trade (consistency) into the import design. Companies
are designated as either consistent or inconsistent im
porters of particular products. This information is used
in each stage of sampling with the result being an in
creased yield of useable prices.
The export frame is drawn by the Bureau of the Cen
sus according to a sample design provided by b l s . In
contrast to imports, this frame is a sample of the total
universe. The selection of s e d ’s is done in each
publishable stratum with probabilities proportionate to
size ( p p s ) , using the dollar value of any line or lines on
the s e d within that stratum as a measure of size for the
s e d ’s chance of selection. Once selected, all lines on that
SED are in the sample. Because the s e d ’s do not contain
a company identifier, the name and address of each ex
porter chosen in the initial sample is obtained
manually.7 Once chosen, these exporter names and ad
dresses are linked to the s e d line information resulting
in the export frame utilized by i p p .
The sample design for both imports and exports con
sists of three stages. The first stage selects estab
lishments. The second stage selects Entry Level Items
( e l i ’ s —commodity classes within a publishable
stratum). The third stage is the selection of specific
items in the e l i . The system is identical for both imports
and exports unless otherwise noted.
The first step is to generate the measure of size (maxprob) for an establishment as follows. The dollar value
(weighted dollar value in exports) on each document is
aggregated to company-ELi, company-publishable
stratum, and company levels. It is also aggregated
within an e l i , and within a stratum, across all com
panies. A proportion is then calculated for each
company-stratum by dividing the aggregated company7 Of necessity, the frame used in sampling is a subset of the total universe of
export transactions.
stratum dollar value by the aggregated dollar value
within the stratum. This “ company-stratum prob” is
the proportion of dollar value that the company con
tributes to the particular stratum. The max-prob for
each company is the maximum company-stratum prob
for that company over all strata. In addition to a maxprob, a max-prob stratum (the stratum associated with
the max-prob) is assigned to a company. The companies
are than implicitly stratified by max-prob strata and a
systematic p p s selection of companies is made using the
max-prob as the company’s measure of size. The prin
cipal 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 appealing since it is
product category indexes which are being produced.
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 control respondent
burden, it is then necessary to select a second-stage sam
ple of e l i ’s within each company. The first step is to en
sure that publication requirements are met by selecting
company e l i ’s with certainty in some strata. The re
maining e l i ’s in each company are then sampled using a
systematic p p s technique. The measure of size is the e l i prob (the company-stratum prob distributed among the
e l i ’s in the company in proportion to their dollar value
contributions). This constitutes the sample of respond
ents 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 prob
ability selection method was introduced. Under this
method, the e l i is partitioned into subclasses and a p p s
selection is made among the subclasses using their pro
portion of trade in the establishment as the measure of
size.8The process continues through successive subdivi
sions of each selected subclass until an identifiable item
that can be priced over time is obtained.
At this time, most published indexes are composed of
products selected prior to the implementation of this
probability method. The judgmental criteria that were
previously in place required that items selected in each
company be repriceable and that their price movement
be representative of the respondent’s other products in
the same ELI. As maintenance samples are conducted on
a 4-year timetable, the number of these judgmentally
selected products will decrease. Therefore, by 1986, all
published indexes will be constructed using probability
selected products.
Estimating Procedures
Formula
The export and import price indexes are weighted in
dexes of the Laspeyres type. Price relatives are assigned
8 This “proportion of trade” estimate is provided by the respondent.
64
equal importance within each weight category and are
than aggregated to the SITC index level.
Analysis and Presentation
The export and import price indexes are published
quarterly in bls news releases 5 weeks after the
reference month. The indexes, which are not seasonally
adjusted, are published by sitc categories and are also
shown in terms of percent change. The reference period
is 1977 = 100, where possible. In numerous cases,
however, price data do not extend back to 1977, and
these indexes use a more recent reference period. For
calculation purposes, as noted above, the weight base
remains 1980. Following the calculation, indexes are set
equal to 100 in the reference period.
In addition to the export and import price indexes, a
quarterly report is prepared that updates bls Bulletin
2046, Comparisons o f U.S,, German, and Japanese Ex
port Indexes. This bulletin presents index comparisons
between the United States and the Federal Republic of
Germany, and the United States and Japan. These data
are useful in measuring the U.S. export price movement
of a given commodity area in comparison to the price
movement of competitive products exported from Ger
many and Japan. The methodology is explained in
detail in the bulletin and is accompanied by 34 reference
tables of U.S.-Germany comparisons, and 26 reference
tables of U.S.-Japan comparisons.
The trend of prices of U.S. and other industrial coun
tries’ exports to opec is analyzed in bls Bulletin 1969,
Estimating Price Trends o f Industrial Countries’ Ex
ports to o p e c . These are compared with opec crude oil
prices. The recently released U.S. import price index for
crude oil is discussed in “ Import Price Indexes for
Crude Petroleum,” Monthly Labor Review, November
1982.
2 2
j i
Po
1
n.
J
2
j
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
nj
t
= product within j
= number of price relatives within j
= time
wj
=
share o f value o f jth category in group x in
the base year
pV P°
=
price relative of product i in year t to base
year o
Weights
The values assigned to each weight category are based
on trade value figures compiled by the Bureau of the
Census.9 In the case of the export price index, price
relatives in each 7-digit Schedule B category are
weighted by the dollar value of exports in that category
during the base period. For the import price index, price
relatives are weighted by the dollar value of imports in
each 7-digit tsusa category during the base period. For
both indexes, the weight period is 1980.
Each value weight for a 7-digit category covers un
priced items as well as the priced items which have been
selected in that category for estimating indexes. Not all
7-digit categories are included in each sitc group for
which an index is published. Instead, a sample of 7-digit
categories represents all of the 7-digit categories within
the index group. When new 7-digit categories, i.e.
weights, are introduced into an index, the index, in
cluding the new categories, is linked to the earlier index.
As the judgmentally selected items in establishments are
replaced by probability selected items through sample
rotation, the weight system will be altered to the ap
propriate sampling weights.
, Uses
9 The value data for 7-digit categories are compiled by the Bureau o f the Cen
sus using Shipper’s Export Declarations and Import Entry Documents. This in
formation is available on magnetic tape and can be found in the following
Bureau o f the Census publications:
Exports: U.S. Exports—Schedule B Commodity by Country, Report
FT-410, December o f each year.
Imports: U.S. Imports fo r Consumption and General Imports, Report
FT-246, annual.
65
The indexes published in this program are the only in
dexes 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 analysts and policymakers to assess the effect of
export and import price changes in 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 real exports and imports in the aggregate and
for product groups. They provide a basis for measuring
changes in the prices of U.S. products in relation to
price trends of comparable products of other major in
dustrial countries with which the United States com
petes for markets, and for assessing changes in U.S.
price competitiveness.
Technical References
Carpenter, J. Finley; Bishop, Troy M.; and Goudie, Ginger S.
“ System for Matching Company Documents,” Pro
Kravis, Irving, and Lipsey, Robert E. Price Competitiveness
in World Trade. New York, Columbia University Press
for the National Bureau of Economic Research, 1971.
ceedings o f the Section on Survey Research Methods.
Washington, American Statistical Association, 1978.
Carpenter, J. Finley. “ Error Analysis in the International
Price Program.” Paper presented at the American Sta
tistical Quality Control Technical Conference. Chicago,
1978.
Pratt, Richard J., and Ferguson, Gwyn R. “ Alternative
Sample Designs in the International Price Program,”
Proceedings o f the Section on Survey Research
Methods. Washington, American Statistical Associa
tion, 1980.
Creamer, D. “ Some Recommendations for Data Improve
ment in the GNP Accounts,” Statistical Reporter.
Washington, U .S. Office o f Management and Budget,
January 1975.
Suomela, John W. “ The Meaning and Measurement o f Inter
national Price Competitiveness,” Proceedings o f the
Section on Survey Research Methods. Washington,
American Statistical Association, 1978.
Interagency Committee on Measurement o f Real Output,
Subcommittee on Prices. Report on Criteria fo r Choice
o f Unit Values or Wholesale Prices in Deflators.
U.S. Congress, Joint Economic Commmittee. Government
Price Statistics. Hearings. Subcommittee on Economic
Statistics, 87 Congress, 1 Sess.; Part 1, Jan. 24, 1961;
Part 2, May 1-5, 1961.
Mimeographed. Washington, Bureau o f the Budget,
June 17, 1970.
Kasper, Marvin, and Pratt, Richard J. “ Surveying Inter
national Prices,” Proceedings o f the Section on Survey
Research Methods. Washington, American Statistical
Association, 1978.
U.S. Department o f Labor, Bureau of Labor Statistics.
Comparisons o f United States, German, and Japanese
Export Price Indexes, Bulletin 2046, February 1980.
U .S. Department of Labor, Bureau o f Labor Statistics.
Kravis, Irving, and Lipsey, Robert E. “ International Prices
and Price Proxies” in Ruggles, N .E ., et al. The Role o f
Estimating Price Trends o f Industrial Countries’
Exports to OPEC, Bulletin 1969, 1977.
the Computer in Economic and Social Research in Latin
America. New York, National Bureau of Economic
Research, 1974.
66
Chapter 9. O eeupatienal Fay
and Supplem entary B enefits
parability 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,
based upon employer records. The Bureau’s first such
study, growing out of a study by the U.S. Senate in
1891, resulted in a wage rate record extending back con
tinuously to 1860. 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. A large survey
program undertaken for the War Industries Board in
1919 produced occupational pay rates by industry and
State, and (for some industries) by city. Between 1934
and 1940, the selection of industries studied was deter
mined largely by administrative needs under the Na
tional Industrial Recovery Act, Public Contracts Act,
and the Fair Labor Standards Act, with emphasis on na
tionwide data for relatively low-wage industries.
Survey activity shifted in the early 1940’s defense
period to heavy industries essential to war production.
Implementation 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, the
Description of Syrweys
Although differing in industrial, geographic, and oc
cupational coverage, the surveys described form an in
tegrated program of occupational wage surveys based
upon a common set of administrative forms, a single
manual of procedures, and common concepts and
definitions. Employer cooperation in surveys is on a
voluntary basis. Confidential individual establishment
data compiled by the Bureau’s field representatives are
grouped in published reports in a manner that will avoid
possible disclosure of an establishment’s rates. In all
surveys, establishments are classified by industry as
defined in the 1972 edition of 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. Standard Metropolitan
Statistical Area definitions are employed in all pro
grams.2
Industry Wage Surveys provide data for occupations
selected to represent a range of activities performed by
workers. Consideration is given to prevalence in the in
dustry, definiteness and clarity of duties, and impor
tance as reference points in collective bargaining.
In addition to collecting straight-time first-shift rates
(or hours and earnings for incentive workers) for in
dividuals in the selected occupations, surveys in most in
dustries also establish the wage frequency distribution
for broad employment groups, such as production and
related workers or nonsupervisory workers.
Weekly work schedules; shift operations and dif
ferentials; paid holiday and vacation practices; and
health, insurance, and pension plans are included in the
information collected, along with other items applicable
to a particular industry. The studies also provide
estimates of labor-management agreement coverage,
proportions employed under incentive pay plans, and
the extent to which establishments provide a single rate
or a range of rates for individual job categories.
Bureau developed new types o f w age 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 com
mon to a wide variety of manufacturing and non
manufacturing industries within metropolitan areas.
This survey program was firmly established and tem
porarily expanded for use in the wage stablization 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 pro
jected to represent all metropolitan areas combined.
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 cur
rently being used in administering the Federal Pay Com
1 See appendix B.
2 See appendix C.
67
are directed to the Scope and Method of Survey appen
dix in the published bulletins for a description of current
practice.
workers meeting the basic requirements established for
the job are included.
In applying these job descriptions, the Bureau’s
field representatives exclude working supervisors, ap
prentices, learners, beginners, trainees, handicapped
workers, part-time or temporary workers, and proba
tionary workers unless provision for their inclusion is
specifically stated.
Tabulations of paid holidays, paid vacations, and
health, insurance, and pension plans are based on the
assumption that plans are applicable to all 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 insurance and
pension plans are limited to plans for which at least a
part of the cost is borne by the employer. Informal pro
visions are excluded.
Concepts
The Bureau’s occupational wage surveys summarize a
highly specific wage measure—the rate of pay, ex
cluding premium pay for overtime and for work on
weekends, holidays, and late shifts, for individual
workers. For workers paid under piecework or other
types of production incentive pay plans, an earned rate
is computed by dividing straight-time earnings for a
time period by corresponding hours worked. For all
workers, production bonuses, commissions, and costof-living bonuses are counted as earnings. In general,
bonuses that depend on factors other than the output of
the individual worker or group of workers are excluded;
examples of such nonproduction payments are safety,
attendance, year-end or Christmas bonuses, and cash
distributions under profit-sharing plans.
Unless stated otherwise, rates do not include tips or
allowances for the value of meals, room, uniform, etc.
The earnings figures, thus, represent cash wages (prior
to deductions for social security, taxes, savings 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 his
regular straight-time salary.
Occupations are defined in advance of the survey.
Because of the emphasis on comparability of occupa
tional content, the Bureau’s job descriptions may differ
significantly from those in use in individual estab
lishments 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 concept of the jo b .3
Although work arrangements in any one establish
ment may not correspond precisely to those described,
Survey Methods
Planning. 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 importantly supple
ment comments and suggestions received from the
regional offices at the conclusion of 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 devel
oped in conjunction with the Office of Management and
Budget and the Office of Personnel Management.
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 Stan
dard 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
salary survey of professional, administrative, technical,
and clerical jobs. The needs of major users are a prime
consideration in designing the multipurpose occupa
tional studies.
The minimum establishment size included in a survey
is set at a point where the possible contribution of the
3 An example o f a job description:
M ACHINIST, M AINTENANCE
Produces replacement parts and new parts in making repairs o f metal parts of
mechanical equipment operated in an establishment. Work involves most o f the
following: Interpreting written instructions and specifications; planning and lay
ing out o f work; using a variety o f machinists’s handtools and precision measur
ing instruments; setting up and operating standard machine tools; shaping of
metal parts to close tolerances; making standard shop computations relating to
dimensions o f work, tooling, feeds, and speeds o f machining; knowledge of the
working properties o f 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 ap
prenticeship or equivalent training and experience.
4
In general, workers are included in a classification if the described duties are
performed a major part o f the time and the remainder is spent on related duties
requiring similar or lesser skill and responsibility. However, in some jobs, par
ticularly office and skilled production-worker categories, workers may regularly
perform a combination o f duties involving more than one occupation. Unless in
dicated otherwise in the description, in these situations consideration for
classification purposes is given to those elements o f the job which are most im
portant in determining its level for pay purposes. Thus, a worker meets the basic
concept o f the stenographer classification if taking dictation is a regular require
ment o f the job even though a majority o f the time is spent on routine typing.
68
Forty manufacturing and 25 nonmanufacturing in
dustries, accounting for about 27 million employees, are
surveyed. A majority are studied on a 5-year cycle, but a
number of comparatively low-wage industries are on a
3-year cycle. Most surveys are at the 4-digit sic level of
detail.
Nearly all of the manufacturing, utilities, and mining
industries are studied on a nationwide basis, and
estimates are provided also for regions and major areas
of concentration. Surveys in trade, finance, and service
industries usually are limited to a number of
metropolitan areas. Nationwide surveys generally
develop separate estimates by size of establishment,
size of community, labor-management agreement
coverage, and type of product or plant group.
Area Wage Surveys annually provide data for selected
office clerical, professional, technical, maintenance,
toolroom, powerplant, material movement, and
custodial occupations common to a wide variety of in
dustries in the areas surveyed. The occupations studied
provide representation of the range of duties and
responsibilities associated with white-collar, skilled
maintenance trades, and other nonproduction manual
jobs. Weekly salaries reported for individuals in whitecollar 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 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 in
dustries. Establishments employing fewer than 50
workers are excluded—with a minimum of 100 applying
to manufacturing; transportation, communication, and
other public utilities; and to retail trade in the 13 largest
areas.
In addition to the all-industry averages and distribu
tions 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 two areas. There were 70
Standard Metropolitan Statistical Areas in this survey
program as of 1981, selected to represent all metro
politan areas of the United States, excluding Alaska
and Hawaii. In 31 of the larger areas, wage data are
presented separately for establishments that have 500
workers or more.
Data on weekly work schedules; paid holiday and
vacation practices; and health, insurance, and pension
plans are recorded separately for nonsupervisory office
workers and production and related workers (non
office). Data on minimum entrance rates for inex
perienced office workers are collected in all industries.
While the wage data are collected annually, these items
are studied every 3 years.
69
Cross-industry area wage surveys have also been con
ducted annually since 1967 at the request of the Employ
ment Standards Administration for use in administering
the Service Contract Act of 1965. Survey scope and
method are the same as for the Bureau’s regular
surveys, but a more limited number of occupations and
related benefits are studied and data are published only
for all industries combined.
In addition to the cross-industry surveys, special in
dustry studies are conducted for the Employment Stan
dards Administration. These studies provide informa
tion on hourly earnings for moving and storage, refuse
hauling, laundry and dry cleaning, and food serv
ice jobs. For both the cross-industry surveys and special
industry studies, data on incidence of paid holidays;
vacation practices; and health, insurance, and pension
plans are provided every 3 years.
The National Survey o f Professional, Administrative,
Technical, and Clerical Pay provides a fund of broadly
based information on salary levels and distributions in
private employment. Approximately 100 occupation
work levels were studied in 1981 selected from the
following fields: Accounting, legal services, personnel
management, engineering and chemistry, purchasing,
photography, drafting, computer processing, and
clerical. Definitions for these occupations provide for
classification of employees according to appropriate
work levels (or classes). Although reflecting duties and
responsibilities in industry, the definitions were de
signed to be translatable to specific pay grades of
Federal white-collar employees. This survey, thereby,
provides information in a form suitable for use in com
paring the pay of salaried employees in the Federal civil
service with pay in private industry.
Monthly and annual average salaries are reported for
all occupations. Data relate to the standard salaries that
were paid for standard work schedules; i.e., to the
straight-time salary corresponding to the employee’s
normal work schedule, excluding overtime hours. Na
tionwide salary distributions and averages are presented
for men and women combined. Averages also are
presented for establishments employing 2,500 workers
or more.
Industry divisions included are: (1) Mining; (2) con
struction; (3) manufacturing; (4) transportation, com
munication, electric, gas, and sanitary services; (5)
wholesale trade; (6) retail trade; (7) finance, insurance,
and real estate; and (8) selected services.
Limited to the Nation’s metropolitan areas during
1960-64, the annual survey was expanded in 1965 to in
clude nonmetropolitan counties. The minimum estab
lishment size included in the, survey is 50, 100, or 250,
depending on the industry. The minimum establishment
size has been adjusted at various times since 1961 in
response to the specifications of the President’s Pay
Agent. Since the survey scope is subject to change, users
salary rates (or hours and earnings, when needed) from
payroll or other records and data on the selected
employer practices and supplementary benefits from
company officials, company booklets, and labormanagement agreements.
Area wage surveys in all areas involve visits every
third year with partial collection 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 respon
dent. Visits are made to establishments not responding
to the mail or telephone request and to those reporting
unusual changes from previous-year data.
The work of field representatives is checked for quali
ty of reporting and accuracy in job matching. Revisits
are made by supervisory and senior representatives.
Systematic technical audits of the validity of survey
definitions, made by staff with specialized training, also
are maintained for the technically complex nationwide
white-collar salary survey.
excluded establishments is regarded as negligible for
most of the occupations surveyed. Another practical
reason for the adoption of size limitations is the diffi
culty encountered in classifying workers in small
establishments where they do not perform the spe
cialized duties indicated in the job definitions.
Considerations in timing of industry surveys include
date of expiration of major labor-management agree
ments, deferred wage adjustments, seasonality of pro
duction, and interests of users. Wherever possible, area
wage surveys are timed to follow major wage set
tlements as well as to meet the needs of government
agencies engaged in wage administration as required by
law.
The types of occupations studied and the criteria used
in their selection are identified in the description 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 estab
lished hierarchy of job rates to be found within
establishments and industries permits 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 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.
Sampling
All surveys are conducted on a sample basis using a
suitable sampling “ frame,” a list of establishments
which fall within the designated scope of the survey.
The frame is as close to the universe as possible but is
often incomplete. B l s uses frames primarily compiled
from lists provided by regulatory government agencies
(primarily State unemployment insurance agencies).
These are supplemented by data from directories, trade
associations, labor unions, and other sources.
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 in
dependently. Within these broad groupings, a finer
stratification by product (or other pertinent attribute)
and size of establishment is made. Stratification may be
carried still further in certain industries: Textile mills,
for instance, are classified by whether they spin only,
weave only, or do both. Such stratification is important
if the occupational structure of the various industry
segements differs widely.
The sample for each industry-area group is a prob
ability sample, each establishment having 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 frac
tion used in the various strata ranges downward from all
large establishments through progressively declining
proportions of the establishments in each smaller size
group. This procedure follows the principles of op
timum allocation where the standard deviation of the
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 schedule, overtime premiums,
paid holidays and vacations, insurance and pension
plans, union contract coverage, and other items ap
plicable to the establishment. The second ( b l s 2753G) is
used in recording occupation, sex, method of wage pay
ment, hours (where needed), and pay rate or earnings
for each worker studied. Supplementary forms are used
to meet particular needs.
Collection. Bureau field representatives collect data by
visits to each of the sample establishments. Job func
tions and factors in the establishment are carefully com
pared with those included in the Bureau job definitions.
The job matching may involve review of records such as
pay structure plans and organizational charts, company
position descriptions, interviews with appropriate of
ficials, and, on occasion, observation of jobs within
plants. A satisfactory completion of job matching per
mits acceptance of company-prepared reports where
this procedure is preferred by the respondent. Gen
erally, however, the field representative secures wage or
70
characteristic being estimated is assumed to be propor
tional to the average employment in the stratum. 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 estima
tion employed yields unbiased estimates by the assign
ment 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 estab
lishments by size, and the degree of accuracy required.
Area wage surveys are limited to selected metro
politan areas, which form a sample of all such areas
and, when properly combined (weighted), yield esti
mates of 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 of stratification
are region, type of industrial activity as measured by
percent of manufacturing employment, and major in
dustries. Each area is selected with its probability
of selection proportionate to its nonagricultural em
ployment. The largest metropolitan areas are selfrepresenting; i.e., each one forms a stratum by itself and
is certain of inclusion in the area sample.
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 occupa
tion:
Workers in occupa
tion in
sample
establishments
Establishment
A ....................
B ......................
C ......................
2
1
4
40
50
10
$10.40
11.20
10.60
2x40
1x50
4x10
2x40x$10.40
lx50x 11.20
4x1 O 10.60
x
170
Estimated universe
$1,816.00
A similar method applies to any characteristic
estimated from the sample. To 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 proportion of the estimated employ
ment in the 2-week category is computed. Using the
three establishments in the previous example, this can be
illustrated as follows:
Estimating Procedures
Estimated average earnings (hourly, weekly,
monthly, or annual) for an industry or an occupation
are computed as the arithmetic mean of individual em
ployee 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
computed on an establishment basis. Supplementary
benefit provisions which apply to a majority of the pro
duction or office workers in an establishment are con
sidered to apply to all production or office workers and
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
rate for the stratum from which it was selected; e.g.,
if a third of the establishments in one stratum are se
lected, 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 cell or 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 proba
bility of 1) and is thus given a weight of 1. Establish
ment C was taken from the remaining group where 1 of
Weight
Actual
Estimate o f total
employ- Average
in stratum
ment in hourly
occupa- earnings
tion
Workers Earnings
Establishment
Weight
Actual total
establishment
employment
A ...................
B ........................
C ........................
2
1
4
100
500
75
Estimated universe
Weighted
Vacation
employ- provisions
ment
after 2 years
200
500
300
1 week
2weeks
1 week
1,000
Thus, the estimated percentage of workers in
establishments granting 2 weeks’ vacation after 2 years
of service is 500 or 50 percent.
1,000
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 totals to
regional and/or national estimates. 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.
This procedure provides the ratio of nonagricultural
employment in the stratum to that in the sample area
(one in the case of the large self-representing areas).
Summing all such estimated stratum totals yields the
earnings and employment totals for the regional and the
national estimates.
71
Analysis and Presentation
the suitability of job offers. Knowledge of levels and
trends of pay rates by occupation, industry, locality,
and region is required in the analysis of current
economic developments and in studies relating to wage
dispersion and differentials.
Bureau data are used in connection with private wage
or salary determinations by employers or through the
collective bargaining process. To the extent that wages
are a factor, survey data also are considered by em
ployers in the selection of location 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 selection and definition of industries, of geographic
units for which estimates are developed, of occupations
and associated items studied, and in determination of
periodicity and timing of particular surveys. Depending
upon his needs, the user may find it necessary to inter
polate for occupations or areas missing from the survey
on the basis of knowledge of pay relationships.
Because of interestablishment variation in the propor
tion 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 relationships found
in the majority of establishments.
The incidence of incentive methods of payment may
vary greatly among the occupations and establishments
studied. Since hourly averages for incentive workers
generally exceed those for hourly rated workers in the
same job, averages for some incentive-paid jobs may
equal or exceed averages for jobs positioned higher on a
job evaluation basis but normally paid on a time basis.
Wherever possible, data are shown separately for time
workers and incentive workers in the 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 year-to-year changes in 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 changes in the labor
force resulting from labor turnover, labor force expan
sions and reductions for other reasons, as well as
changes in the proportion of workers employed in
establishments with different pay levels. A labor force
expansion might increase the proportion of lower paid
workers and thereby 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 surveys by holding establishment employment
constant while computing percent increases in earnings.
That is, the previous and current-year earnings of each
establishment are weighted by that establishment’s
previous year’s employment.
Where an industry survey is designed to yield
estimates for selected States or areas, these are pub
lished separately as information becomes available from
all sample firms in the State or area unit. Industry
surveys limited to selected areas do not provide a basis
for the examinations of pay levels by size of community,
size of establishment, product, or labor-management
agreement coverage that generally are included in
reports on nationwide surveys. Regardless of geo
graphic scope, industry reports record the incidence of
incentive pay plans and, to the extent possible, average
pay levels separately for time and incentive workers.
Area wage survey percent increases, adjusted for
changes in employment, are computed for broad oc
cupational groups; e.g., office clerical, electronic data
processing, skilled maintenance, and unskilled plant.
These increases are computed annually, separately for
all industries, manufacturing, and nonmanufacturing,
for each metropolitan area studied, for all metropolitan
areas combined, and for four broad Census regions.
Pay relatives for the four occupational categories, ex
pressing area pay as a percentage of the national
average, are published annually, permitting ready com
parisons of average pay levels among areas. Estimates
of labor-management agreement coverage are presented
every third year. Occupational pay relationships within
individual establishments are summarized annually.
Bulletins on the National Survey of Professional, Ad
ministrative, Technical, and Clerical Pay present oc
cupational averages and distributions on an all-industry
basis, nationwide and separately for all metropolitan
areas combined, and for establishments employing
2,500 workers or more. Average pay levels for industry
division are shown as percentages of the all-industry
averages. Year-to-year salary trend estimates for oc
cupations are reported.
Industry wage and area wage survey reports are issued
throughout the year as the surveys are completed. The
bulletin on the National Survey of Professional, Ad
ministrative, Technical, and Clerical Pay is available in
October.
Summaries of the data in the bulletins and special
analyses appear also in the Monthly Labor Review.
Uses and Limitations
Occupational wage data developed in these 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 unemployment compensation agencies in judging
72
Reliability o f surveys. Results of the surveys generally
will be subject to sampling error. This error will not be
uniform, since, for most occupations, the dispersion of
earnings among establishments and the frequency of oc
currence of the occupation differ. In general, the sample
is designed so that the chances are 9 out of 10 that the
published average does not differ by more than 5 per
cent from the average that would be obtained by
enumeration of all establishments in the universe.
The sampling error of the percentages of workers
receiving any given supplementary benefit differs with
the size of the percentage. However, the error is such
that rankings of predominant practices almost always
will appear in their true position. Small percentages may
be subject to considerable error but will always remain
in the same scale of magnitude. For instance, the pro
portion of employees in establishments providing more
than 5 weeks’ paid vacation 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 oc
cupation are subject to considerable sampling error, due
to the wide variation among establishments in the pro
portion of workers found in individual occupations. (It
is not unusual to find these estimates subject to sampl-
ing error of as much as 20 percent.) Hence, the
estimated number of workers can be interpreted only as
a rough measure 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 ac
tual fieldwork, and the universe estimates are revised ac
cordingly. Those firms which should have been included
but were classified erroneously in other industries can
not 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
repetition of the survey in an establishment with dif
ferent interviewers and respondents would undoubtedly
produce 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.
T(§©taB©ai References
identifies the area sample as originally determined for
the labor market survey program.
Cohen, Samuel E. “ Studies o f Occupational Wages and
Supplementary Benefits,” Monthly Labor Review,
March 1954.
An early description of the methods o f wage surveys.
Stelluto, George L. “ Federal Pay Comparability: Facts to
Temper the Debate,” Monthly Labor Review, June
1979.
A review o f the methods used and decisions made in
setting Federal white-collar pay.
Douty, H.M. “ Survey Methods and Wage Comparisons,”
Labor Law Journal, April 1964.
A discussion of the uses of wage survey results and the
pitfalls to be avoided. A short discussion of factors
affecting survey methods is included.
Talbot, Deborah B. “ Improved Area Wage Survey Indexes,”
Monthly Labor Review, May 1975.
A discussion of differences in computing Area Wage
Survey pay increases by the matched and unmatched
sample techniques.
Houff, James N. “ Improving Area Wage Survey Indexes,”
Monthly Labor Review, January 1973.
Kanninen, Toivo P. “ New Dimensions in bls Wage Survey
W ork,” Monthly Labor Review, October 1959.
An outline o f the occupational wage survey programs,
as expanded in fiscal 1960. Lists the type of survey and
cycle for each of 70 industries studied separately,
Ward, Virginia L. “ Area Sample Changes in the Area Wage
Survey Program,” Monthly Labor Review, May 1975.
A description of the Area Wage Survey program and
changes in the area sample.
73
C h a p tsr 10= IMagototed Wag®
asud B® fDt Changes
rti®
more in all industries and 1,000 workers or more in con
struction.
Contracts covering multiplant firms are included if
the agreement as a whole covers 1,000 workers even
though each plant employs fewer workers. Also includ
ed are contracts 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 in
dividually representing fewer than 1,000, negotiate
essentially identical contracts with one or more firms,
the workers involved are considered to constitute one
bargaining unit.
The Bureau of Labor Statistics prepares information
on current changes in wages and supplementary benefits
agreed to in collective bargaining. The information in
cludes 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.1 b l s also prepares quarterly and annual
statistical summaries of negotiated wage changes in all
major collective bargaining situations in private in
dustry, and semiannual summaries for State and local
government bargaining units.
Background
Bls began publication of the monthly listing of
settlements 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 in
dustry. 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. Begin
ning with 1979, similar data have been published for
State and local government bargaining units.
State and local government agreements. This series sum
marizes general 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; (3) wages are deter
mined by collective bargaining; and (4) at least 5,000
workers are covered by the bargaining agreement. As of
1980, almost one-fourth of all State and local govern
ment employees covered by collective bargaining
agreements were included in the series.
Data presented
Description ©f Statistical Series
Wage changes. Two types of information are presented
on wage changes. Settlement data measure wage
changes specified in the bargaining settlements reached
during a particular time period (e.g., quarter or year).
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. Both the changes scheduled during
the first 12 months of the contract (first-year changes)
and the total of wage changes scheduled over the life of
the contract, expressed as an annual rate, are presented.
Effective wage change data measure all wage changes
effective in the period stemming from current settle
ments and also from deferred wage changes specified
in earlier settlements and from c o l a adjustments. Con
C@verag©
Private industry agreements. The series summarizes
general wage rate changes in major collective bargaining
settlements (settlements covering 1,000 workers or
more) for production and related workers in manufac
turing and nonsupervisory workers in nonmanufactur
ing. b l s currently follows about 1,900 bargaining situa
tions, for virtually complete coverage of major
agreements. Changes in total compensation are
measured for agreements covering 5,000 workers or
1 Where information is available, unilateral management deci
sions in nonunion situations also are listed.
74
tracts providing no wage adjustments during the period
also are taken into account.
wage and benefit components; i.e., to measuring the ef
fect of settlements on employer outlays for employee
compensation. Included are: Changes in wage rates;
modifications in premium pay, bonuses, paid leave, and
severance pay; and adjustments in employer payments
for pension, health and welfare, and supplemental
unemployment benefits, excluding the costs of ad
ministering 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 compensation, are not nor
mally considered part of compensation, such as per
diem payments, moving expense reimbursements,
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 non
union workers in the same firm or to members of other
bargaining units are not considered. Similarly, although
the cost of providing lengthened vacation 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 ef
ficiency as well as employer payments, are disregarded.
Compensation changes. Although at one time the
economic terms of collective bargaining settlements in
volved wage rates almost exclusively, today, changes in
a wide variety of benefits also must be considered.
“ Compensation” refers to the total of pay and benefits.
As with wage change data, the Bureau publishes com
pensation change 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
construction. Changes scheduled for the first year of the
contract, and those scheduled over the entire contract
term, expressed as an annual rate, are published.
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 in
clude average hourly earnings in the establishment, and
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 re
leased outside the Bureau. Other data sources for these
calculations include the file of union contracts main
tained 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 publica
tions, are used in producing listings of agreements.
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 ex
ample, analysts assume that methods of financing pen
sions will not change, and that expenditures for in
surance 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 Con
sumer Price Index.
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 of the workers involved recognize that
the choice in incorporating alternative benefit changes
into contracts is affected by their costs, which, in turn,
are affected by the character of the work force. For ex
ample, 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
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 priced
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 effi
ciency. Such effects, however, are not measurable. Con
sequently, the b l s program is confined to measuring the
75
percent to the annual cost of straight-time pay for work
ing time when half of the workers have been employed
for 15 years or more.
Changes in wage rates affect costs for certain benefits
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 ad
justments; 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-perhour figures are used as the costs of the settlement pro
visions. Percentage 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
additional 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 fre
quency of layoffs, which at best, is hazardous to
estimate. Pension improvement costs are particularly
difficult to estimate because employers often have con
siderable discretion in funding their obligations, b l s
assumes that a pension benefit change will change ex
isting expenditures for current service proportionately.
tercompany comparisons by eliminating 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 in
surance). The overall percentage change generated by
each settlement is weighted by the number of workers
affected (the pricing of individual settlements is not
disclosed). The sum of the worker-weighted changes is
divided by the total number of workers affected, to
determine average percent change. Effective wage
change data are handled in similar fashion. Since collec
tive bargaining agreements generally are for 2-year
periods or longer, b l s expresses the total percent change
over the contract term at an annual rate to permit com
parison 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 increase
take into account the compounding of successive
changes. In addition, the Bureau computes first-year
changes as a percent of current hourly wages (or com
pensation). Generally, the first-year increase is larger
than the average annual increase over the full term of
the agreement.
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
reopening clause is treated as a new settlement.
Sometimes, the parties to a contract agree to an
Since em ployer contributions for pensions frequently
unscheduled contract reopening. B eginning w ith full-
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 settle
ment provision. 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 accompanied by additional overtime work;
unless this overtime is specified in the agreement, it is
ignored in the cost estimate.
year data for 1981 (published in January 1982), com
pensation 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 chang
ed 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.
Prss0 ntati©!n)
The listing of current changes in wages and benefits is
published monthly in the periodical Current Wage
Developments ( c w d ). Grouped by industry, 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 ).
Expressing <s@
sts
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 in
76
Statistical summaries of preliminary data on set
tlements and total effective wage and benefit changes 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 in c w d each year.
Statistical summaries of State and local government
bargaining settlements are issued in news releases semi
annually and also appear in c w d .
Uses and limitations
The series on wage and compensation changes
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
compensation, 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
Mediation and Conciliation Service; State and local
government agencies; employer and employee organiza
tions; economic consultants; and researchers and prac
titioners in industrial relations, collective bargaining,
and economic forecasting.
77
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 ex
isting 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 earn
ings, 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 technological developments independent of collec
tive bargaining and may influence the cost of the unionmanagement settlement.
Public and private sector negotiated compensation
changes are not strictly comparable because some fac
tors (e.g., pension benefits) are not subject to collective
bargaining in many State and local government jurisdic
tions.
Glhap'figr 11. Em ploym ent
G@sf index
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
series now represents the civilian nonfarm economy, ex
cluding households and the Federal Government.
Future development of the e c i will include increases
in the number of published series and expansion to in
clude the Federal Government.
The Employment Cost Index ( e c i ) measures the rate
of change in employee compensation, which includes
wages, salaries, and employers’ cost for employee
benefits.1 The e c i was developed in response to a fre
quently expressed need for such a statistical series. Ex
isting measures, 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 e c i 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.2 It measures the
change in a fixed set of labor costs so that it is not af
fected over time by changes in the composition of the
labor force. The survey is timely in that statistics are
published quarterly, approximately 2 months after their
reference date. The e c i also enables users to compare
rates of change in detailed occupational, industrial,
geographic, union coverage, and ownership (publicprivate) submeasures.
D<§s©ript0 d ®f the ECU
®)
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
e c i industries reported in the 1970 Census of Popula
tion; the weights remain fixed from period to period
pending a major index revision, next scheduled to occur
when the results of the 1980 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
occupations and establishments in the economy. The
wage and salary component of the index is represented
by average straight-time hourly earnings in the occupa
tion. Straight-time earnings are defined as total earnings
before deductions, excluding premium payments for
overtime, weekend, and late-shift work. Earnings in
clude production bonuses, commissions, and cost-ofliving allowances but exclude nonproduction bonuses,
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. Earn
ings of salaried employees and those paid under incen
tive 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 e c i .
Ba©kgr®yndl
The ECI survey is being implemented in stages. Initial
ly, 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
November 1978, the survey was expanded to include
establishments in Alaska and Hawaii, and an additional
13 statistical series (union manufacturing and non
manufacturing, for example) were published.
The second major stage was completed in 1980 with
the publication of quarterly changes in total employee
compensation.
1 The measure was initially referred to as the General Wage Index; the term
Employment Cost Index was substituted as a more accurate description. See
Norman J. Samuels, “ Developing a General Wage Index,” Monthly Labor
Review, March 1971, pp. 3-8.
2 Coverage o f the private sector is limited to the private nonfarm economy, ex
cluding private household workers. Public sector coverage includes employees of
State and local governments, but excludes workers in the Federal Government.
78
Since pay rates generally relate to the job rather than
to the incumbent worker, the basic unit of data collec
tion is an occupation in an establishment. The occupa
tion is comprised of all those workers employed in jobs
classified under an e c i occupation in an establishment.
While shifts in the types of workers within the oc
cupation in an establishment may affect wage
movements, shifts in employment among occupations
and establishments are controlled by measuring wage
change for the same occupations in the same
establishments and applying fixed employment weights
to the results. The unit of observation is standardized to
a certain extent below the occupation level by measuring
only certain types of labor within the occupation; e.g.,
full or part time, incentive or time rated, depending on
the predominant type.
The benefit data portion of the e c i encompasses 23
distinct benefit categories, which can be grouped as
follows:
Merchandise discounts
23. Merchandise discounts (retail trade—
department stores only)
The benefit data supplied by respondents normally
consist of data elements which are used to compute the
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 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, information is also collected on
scheduled daily and weekly hours and annual weeks.
The following example illustrates the calculation of the
cents-per-hour-worked cost for a benefit:
CALCULATING THE COST PER HOUR WORKED OF A
BENEFIT—Example
Data element—2.8 average weeks of vacation
Hours-reSated benefits
1. Premium pay for overtime and
work on holidays and weekends
2. Vacations
3. Holidays
4. Sick leave
5. Other paid leave
Scheduled weekly hours—40
Straight-time average hourly rate—$6.95
Annual hours
system)—1,950
Shift differentials
Nonproduction bonuses
Severance pay
Supplemental unemployment benefit
funds
data
processing
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)
10. Life insurance
11. Health benefits
12. Sickness and accident insurance
($778.40/year) / (1,950 hours/year) = $0,399 (cost per hour worked
for vacation)
Note that average annual hours of vacation are also used by the
data processing system to compute annual hours worked.
Pension and savings plans
13. Pension and retirement benefits
14. Savings and thrift plans
The nature of the data collected varies somewhat
depending upon the particular benefit. For hoursrelated 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 in
surance 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 ele
ment, the benefit cost is always converted to cents per
hour worked.
Legally required benefits
Social security
Railroad retirement
Railroad supplemental retirement
Railroad unemployment insurance
Federal Unemployment Tax Act
State unemployment insurance
Workers’ compensation
Other legally required benefits
by
This equation can be broken into the following steps:
Insuranee
15.
16.
17.
18.
19.
20.
21.
22.
(computed
(2.8 weeks/year x 40 hours/week x $6.95/hour
----------------------------------------------------------------- -- $0.399/hour
1,950 hours/year
Supplemental pay
6.
7.
8.
9.
worked
79
(establishments located in a Standard Metropolitan
Statistical Area) and for other areas.
Occupational classification
The e c i occupational classification system is based on
the classification system used for the 1970 Census of
Population.3 The Census system classifies all occupa
tions reported into 441 3-digit occupational categories
(such as accountant, stockhandler, etc.) which are then
combined into 12 major occupation groups:
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) set
tlements are embodied in signed, mutually binding col
lective bargaining contracts.
Professional, technical, and kindred workers
Managers and administrators, except farm
Sales workers
Clerical and kindred workers
Craft and kindred workers
Operatives, except transport
Transport equipment operatives
Laborers, except farm
Farmers and farm managers
Farm laborers and farm supervisors
Service workers, except private household
Private household workers
Data Sources and Collection Methods
The wage, salary, and benefit cost data from which
the e c i is computed are obtained quarterly from a sam
ple of about 2,800 establishments and a sample of oc
cupations within those establishments.
Data collection is initiated by a b l s field represen
tative 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; perform job matches; develop
establishment reporting procedures; and complete the
first schedule. Quarterly reports thereafter are normally
collected by mail or telephone by the b l s regional of
fice.
A major task in the initial contact is job matching;
that is, determining which establishment jobs and
workers match the occupation definitions in the survey.
At this time, characteristics of the occupations are
determined—whether the majority of incumbents are
full or part time, time or incentive workers, or covered
by collective bargaining agreements.
The wage data are collected on a “ shuttle” form (see
e c i Wage Data Form at the end of this chapter) which is
sent to the respondent each quarter for the addition of
new data. The survey months are March, June,
September, 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, in
surance, and other benefits. Then each quarter, the
benefit provisions are summarized and sent to the
respondents to review the information and report any
changes which have occurred since the prior quarter.
For example, in the prior quarter, the respondent might
have reported that 9 of the 10 employees in a surveyed
occupation 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 in
creased to $129 per month. In both the prior and current
Only 9 of these 12 major occupational groups are cur
rently within the scope of the survey. Farmers and farm
managers, farm laborers and farm supervisors, and
private household workers are excluded.
For e c i samples initiated after 1976, many of the
3-digit Census occupation categories were combined in
to broader occupational groups, designated e l o ’s. More
detailed occupations within the e l o ’s are then selected
on a probability basis by b l s field representatives from
data provided by the respondents. It is for these detailed
occupations that wage and benefit data are collected in
itially and on a continuing basis.
The Census occupational classification system only
lists occupations to be included under each of the 441
occupation categories. For data collection purposes,
definitions of the Census occupations and e l o ’s have
been developed.4
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.
Geographic classification
The geographic coverage of the ECI includes all States
and the District of Columbia. Rates of change in wages
and salaries are published using the four-region
classification system which is defined in appendix C.
Statistics are also published for metropolitan areas
3 Classified Index o f Industries and Occupations, 1970 Census o f Population
(Bureau o f the Census, 1971).
4 Employment Cost Index Occupation Classification System Manual (Bureau
o f Labor Statistics, June 1981).
80
quarter, the employer assumed 50 percent of the plan’s
cost. For e c i 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 par
ticipating in the plan is $57.50. Ninety percent of the
workers participate, $57.50 x 0.9 = $51.75.) The cur
rent quarter’s cost of the plan would equal $58.05
($64.50 x 0.9 = $58.05). Note that the 90-percent par
ticipation rate was held constant. This would be chang
ed only if the employee contribution rate (50 percent of
plan cost) increased or decreased. Holding the participa
tion rate constant eliminates the effects of forces such as
shifts in work force composition from affecting the
measurement of the cost change. Similarly, when an
employer changes an overtime pay provision, new over
time 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 overtime cost to
cost changes caused by the adoption of a new overtime
rate and eliminates the effect of changes in the number
of hours of overtime worked from period to period.
The Syrw©y Desago
The current e c i survey design consists of three parts:
(a) The private sector, initiated in 1975; (b) the public
sector, initiated in 1981; and (c) the replenishment
samples, to be initiated from 1981 through 1984.
Prwat® s@<gtor= respondent universe and
sampie design
Principal features of the design for the private sector
are:
1. Selection of a set of sample occupations by
industry based on the 1970 census occupational
employment distributions.
2. Establishment sampling in two phases. The first
phase consisted of about 10,000 employing units,
the second phase of about 2,200. Collection of
employment data for the occupations selected in
(1) was undertaken in the first phase. Use was
made of available data from b l s Occupational
Employment Surveys whenever feasible.
3. A two-way controlled selection5 of sample
establishments and occupations after the second
phase of establishment sampling. The resulting
sample was used for wage and benefit data col
lection.
5 R. Goodman and L. Kish, “ Controlled Selection, A Technique in Probabili
ty Sampling,” Journal o f the American Statistical Association, Vol. 45, 1950,
pp. 350-72.
81
4.
Data collection by mail survey in the first
establishment phase, by personal visit in the final
initiation, and by mail or telephone thereafter.
In the private sector, a sample of approximately
establishments6 was selected from a larger b l s
sample of 200,000 establishments drawn from the
unemployment insurance universe and other sup
plementary files. Prior to selection of the 10,000
establishments, the larger sample was first ordered by
State, within a State by 2- or 3-digit sic, and within sic
by size class. The 10,000 establishments were then
systematically selected with probability of selection pro
portionate to the measures of size. The measures of size
used were the Unemployment Insurance File employ
ment weighted by the larger sample weight.
Establishments with greater than 4,000 employees were
selected with certainty.
A survey to determine occupational employment
(phase 1 survey) was taken of approximately 23 occupa
tions selected for each 2-digit sic. The basis for this oc
cupational selection was the 1970 Census of Population.
In each 2-digit sic, five occupations with the largest
employment were taken with certainty. Then one to
four occupations (usually two) were selected, again pro
portionate to their employment as determined in the
1970 census, within each major occupation group
( m o g ). This was done to insure the publishability of the
m o g indexes. Since there was not a 100-percent response
to the phase 1 survey, it was necessary to impute for par
tial responses and total refusals. The imputed employ
ment used for a given nonresponding establishment/occupation was the estimated proportion of total employ
ment for that occupation multiplied by the establish
m e n t’s em ploym ent. This resu lted in every
establishment/occupation having a probability of being
selected unless it had specifically reported zero employ
ment.
The occupational employment was then used in
calculating the measures of size for the next phase of the
selection. These measures of size enhanced the prob
ability of retaining any establishment with a large pro
portion of the occupational employment in any 1 of the
23 occupations. Approxmately 2,000 establishments
were selected in this fashion, selection being done
separately within each 2-digit sic.
Finally, the specific occupations for a given establish
ment were selected within each 2-digit sic using a twoway controlled selection technique. Generally, larger
establishments tended to have a larger sample of oc
cupations.
1 0 ,0 0 0
6
The term establishment generally indicates a single, physical location. This is
particularly true in the private sector. 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 o f units in several different locations.
Subsequent to the selection of the basic sample, a
number of supplemental samples were constructed. The
purpose of most of those was to bolster the level of
response in certain industries and, in one instance, to ex
pand the scope of the survey to include Alaska and
Hawaii. The survey designs for these supplemental
samples were somewhat different from the design used
for the basic sample. In the construction supplement,
for example, which replaced the original sample of con
struction establishments, sample occupations were more
broadly defined than the level of the 441 Census occupa
tions used in the original sample. These broader occupa
tions, designated e l o ’s, were constructed by combining
several similar Census occupations within the same
MOG. This was done to increase the probability of get
ting an occupational match at the time of initiation.
Once the broader occupation was matched, a single,
more detailed job title within the occupation was
selected with probability proportionate to its employ
ment. The selected job title became the unit for which
data were requested.
The expansion of the e c i to include Alaska and
Hawaii consisted of a single phase sample similar to the
rest of the private sector; i.e., systematically with prob
ability proportionate to size of employment from a
frame ordered by 2-digit sic.
sional schools, and junior colleges). Establishments
were stratified by 3-digit SIC; then, with a certainty
cutoff,7 a sample was selected with probability of selec
tion proportionate to enrollment within the school. The
sampling frame was ordered by region, and within
region, by size of enrollment. When ELO’s for schools
were defined, a phase 1 survey was conducted by mail to
determine the employment within each of the ELO’s 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
sector, with the exception of the subsampling of e l o ’s at
the time of initiation. The final sample consisted of 260
establishments.
Hospitals
No survey of occupational employment was under
taken 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 proportionate to size and
with a certainty cutoff. The establishment sampling
frame was the 1976 HEW list of public hospitals. Two
sets of occupations were selected: One for State
hospitals and one for local government hospitals. The
occupational selection was essentially a systematic sam
ple within each establishment. The 106 establishments in
the final sample were then requested to supply data for
the appropriate occupations.
Public sector— respondent universe and
sample design
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
governments. Each has a somewhat different survey
design.
As in the case of the construction supplement, Census
occupations were combined into e l o ’s. When an e l o is
matched in an establishment, a single, more detailed job
title is selected using probability proportionate to size
(employment) sampling procedures. The use of e l o ’s
and the sampling to a specific job title increases the
probability of finding occupational matches while re
taining the advantages of surveying narrowly defined
occupations. Different sets of e l o ’s were chosen for
States than for local governments.
State and large local governments
No universe listing of establishments was available
for State and large local governments; it was, therefore,
necessary to conduct a refinement survey to develop a
list of potential sample units. This refinement was ac
complished through personal visits and allowed for the
selection, based on the respondent’s criteria, of iden
tifiable units within each jurisdiction and the assign
ment of major industry division 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 pro
vided 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 and
over employees. The 30,000 and over units were picked
with certainty and the other two groups were further
classified into four Census regions. This provided eight
probability strata from which a probability-
Schools
The sampling frame for public elementary and secon
dary schools was the 1973-74 National Center for
Education Statistics (n c e s ) listing of all State and local
schools. The frame includes 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 sampling 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, profes
7
Certainty cutoff indicates that all units with a measure o f size greater than a
specified figure are automatically selected.
82
proportionate-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.
The selection of States that underwent refinement 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 probability of selection proportionate to size.
The above procedures resulted in the selection of ap
proximately 780 units for which phase 1 data on occupa
tional employment were collected. 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 1
survey was done for small local governments (units with
fewer than 100 employees). Any refinement required
was accomplished by b l s field representatives at the
time the data were collected. The universe of small local
governments 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 probabilityproportionate-to-size sample selection was done in each
stratum. Thirty final sample units were selected.
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 ma
jor industry division, m o g , and size class. The applica
tion of the nonresponse adjustment factors compensates
for the loss of data due to base period refusals only.
Although the adjustment factors are calculated and ap
plied only once, their effects on the estimates are cons
tant for the duration of the specific samples.
For wage change estimation after the base period,
values are imputed when there is a temporary
nonresponse. The basic assum ption is th a t
nonrespondents have, on the average, the same value
that respondents have. Therefore, for a temporary
nonresponse, an establishment/occupation’s prior
quarter data are moved by the average occupational
wage change estimated from similar establishment/oc
cupations. Establishment/occupations are considered
similar if the establishments are in the same 2-digit sic
and the occupations are in the same m o g . If there are
not sufficient data at this level, a broader level of ag
gregation is used. Prior quarter data are not adjusted
when nonresponse is the result of the 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
establishment/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 e c i is simply a weighted average of
the cumulative average wage changes within each esti
mation cell, with base-period wage bills as the fixed
weights. The simplified formula is:
Sample replenishment and sample rotation
Beginning in 1981, the existing sample of private sec
tor establishments is gradually being replaced by a new
sample. A few very large establishments may be includ
ed in both the old and new sample. Sample replenish
ment is necessary to ensure sufficient data to allow con
tinued publication of detailed occupation and industry
rates of change in employee compensation. Replenish
ment will be done in stages, with part of the sample be
ing replenished each quarter.
As in the case of the construction supplement and the
public sector, in the new replenishment samples, detail
ed Census occupations within MOG’s are being combined
into ELO’s.
Planning is also being done to redesign the e c i survey
and build sample rotation into the program. The cycle
of sample rotation has not been established but a 3- to
5-year rotation plan is under consideration.
Adjustments for sample nonresponse
I W 0,i
It = — -------------- * 100
s w 0si
i
8
In actual practice, the ECI computational formulas and procedures differ
somewhat from those presented here, which have been simplified for illustrative
purposes.
83
The index computation involves essentially five steps:
where:
Mt,i = Mt - l , i * Rt i and
It is the symbol for the index.
1. Establishment/occupation sample weights are ap
plied to the occupational earnings to obtain weight
ed average earnings for each estimation cell for the
current and prior survey periods. The estimation
cell is defined on the basis of owner/industry/occupation. For the private sector, 62 SIC industries
have been identified, most at the 2-digit level. For
the public sector, separate cells are identified for
State and for 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 estima
tion cell is identified as State government/public
administration/clerical workers.
2. Each quarter, the ratio of the current quarter weight
ed average wage to the prior quarter weighted aver
age wage is, in effect, multiplied by the prior
quarter cumulative average wage change for the
cell. The product is a measure of the cumulative
percentage wage change in the cell since the base
period.
The other variables are defined as follows:
WG i 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.
Mt j is the cumulative average wage change in the ith
cell from time o (base period) to time t (current
quarter).
Rt j is the ratio of the current quarter weighted
average wage in the cell to the prior quarter
weighted average wage in the cell, both calculated
in the current quarter using matched establishment/occupation wage quotations. The weights
applied are the sample weights described in the
next section.
All wage indexes are computed from the following data:
3.
4.
Average straight-time hourly earnings for 3-digit Census
code occupations, or groups of those occupations, in
those sample establishments for which data are avail
able for both the current and prior survey periods. The
occupational wage data are identified by major occupa
tional group, industry, geographic location, metropoli
tan area, and union status.
Employment, in 1970, in the 3-digit Census code oc
cupation or group of occupations in an industry, ob
tained from the decennial census.
Sample weights derived from an occupational em
ployment survey or the initial employment reported on
the survey schedule. These weights reflect both em
ployment in each establishment/occupation surveyed
and the probability of selection of that establishment/
occupation.
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:
Current
quarter
cumulative
change
(a x d)
Base
period
wage
bill
Current
quarter
wage
bill
(t x e)
Prior
quarter
wage
bill
(f x a)
(d)
(e)
(f)
(g)
(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
Prior
quarter
cumulative
change
Current
quarter
weighted
average
earnings
Prior
quarter
weighted
average
earnings
(a)
(b)
(c)
Electricians
1.23567
$5.50
Carpenters
1 1.15435
7.20
Occupation
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.
Relative
(b/c)
Total
j__________
84
in August. Initially, the statistics are presented in a news
release which includes descriptions of quarter-toquarter and year-to-year trends, tables with the
statistics, and an explanatory note about the survey.
The data are published later in Current Wage
Developments and the Monthly Labor Review, monthly
b l s periodicals.
it = (g/f)*ioo =
($25,995.28/$20,929.77)* = 124.2
lt - 1 = (h/f)*100 =
($25,186.00/$20,929.77)*100 = 120.3
Link relative (percent change) It/I t _ i = 1.032
(3.2 percent)
It refers to the index for the current quarter.
It - i is the index for the prior quarter.
Uses and limitations
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 in
dustries and regions for each group; the wage bills for
the industry divisions 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 in
dexes 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
distribution in the sample between, for example, union
and nonunion 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 im
portance of the union sector in each cell is not held cons
tant over time. Since the relative weights of the region,
the union, and the metropolitan area subcells are allow
ed to vary over time, it is not possible to calculate
Laspeyres indexes for the nonnational series.
Presentation
e c i statistics are published quarterly in the second
month after the survey period. For example, statistics
computed from the survey data for June are published
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 in
dexes are also provided for union status, geographic
region, and metropolitan area status. The ECI is used 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 e c i is particularly important in studies of
the relationships between prices, productivity, labor
cost, and employment and is used as an escalator of
wage costs in long-term purchase contracts.
The limitations of the index must be kept in mind.
Because the e c i is an index, it measures changes in
employee compensation rather than levels of employee
compensation. Further, the index is not a measure of
the total cost of employing 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 employers and employees, although it
does cover nearly all workers in the civilian (nonFederal) nonfarm economy. Finally, the index is not an
exact measure of wage or compensation 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
U.S. Department o f Labor, Bureau o f Labor Statistics. Em
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,” Monthly Labor Review,
January 1978.
U .S. Department o f Commerce, Bureau o f the Census. Classi
ployment Cost Index Occupation Classification System
Manual, 1981.
Bureau o f Labor Statistics, Employment Cost Index Manual
o f Benefit Descriptions, 1981.
W ood, G. Donald. “ Estimation Procedures for the
Employment Cost Index,” Monthly Labor Review, May
1982.
fied Index o f Industrial Occupations, 1970 Census of
Population, 1971.
85
Bureau of Labor Statistics
ECI Wage Data Form
E stablishm ent Nam e
P a g e _____
Schedule N u m b er
Line
No.
BLS
Id e n tific a tio n o f Survey O ccupations, Establishm ent Jobs, o r Individuals fo r w h o m Wage In fo rm a tio n is being reported on each line
o f _____
R eference D ate
Occ.
, 19
C ode
H o u rly
R ate
Hours and Earnings
q
R
(1 )
(2)
1
(3)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
* P le a se u s e t h e b a c k p a g e o f t h is f o r m t o e x p l a in s ig n if ic a n t e a r n in g s c h a n g e s ( i.e ., d e c re a s e s o r la rg e in c re a s e s in t h e a v e ra g e r a te o f p a y f o r a n o c c u p a t io n ) f r o m o n e r e p o r t in g p e r io d t o t h e n e x t .
B L S 3 0 3 8 B (R e v . J u l y 1 9 7 8 )
N u m b er o f
W orkers
Per Line
(4)
VACATION_______________________________________________________________________________________________________________________________
Plan
Company Plan Identification
Eligibility Requirement
Most Recent Change
1. Description
Occupational Distribution
Occu. code
Length
of
Service
|§ |ii| f
11
Weeks □ Wkrs.
or
Percent □
Wt. Wkrs.
Wt. Wkrs.
Wkrs.
Wkrs.
Wt. Wkrs.
Wkrs.
Wt. Wkrs.
Wkrs.
Wt. Wkrs.
Totals
l a W
t o
1
Average
vacation weeks
i
2. Computation
Weeks X Workers = Weighted Workers
Total Weighted Workers -r Total Workers = Average Vacation Weeks
OR
Percent X Workers = Weighted Workers
(Total Weighted Workers-f Total Workers) t 2 = Average Vacation Weeks
Total Weighted Workers -5- Total Workers = Value Entry
3.
Data Entries
(1-9)
Control Information0
7
Benefit
Code
(10-13)
(14-15) (16-18)
02
I
I
I
I
I
I
Conversion
Code
Average Vacation Weeks
(28-29)
(30-35)
I
I '
I
I
-
I
l
I
I
I
I
I
I
02
|
|
I
I
I
I____ [
02
I
I
I
02
Value Entry
(19-27)
Status Code
Occupational
Code
|
|
I
I
!
I
I
02
I
l
I
I
.
I
I
I
I
I
I
I
I
l
.
I
i
|
i
I
I
i
i
I
I
|
|
|
|
|
i
I.
|
I
I
i
i
i
i
i
■
End of card = © » ■ 80
87
07
Chapter 12. Employee Benefit
Plans
Background
ample, major provisions of selected health, insurance,
and pension 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),4studies 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, in
cluding 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 provisions, 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 adjustments 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, Ad
ministrative, Technical, and Clerical Pay ( p a t c ) pro
vides the data on private industry salaries used in ad
ministering this legislation.
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
consequences of accident, sickness, death, and old age.
The adoption of pension and welfare plans was en
couraged 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 sup
plementary benefit improvements while denying wage
increases, (2) National Labor Relations Board decisions
bringing pensions 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. In the
pre-1940 period, key developments in the growth of
benefit plans occurred in nonunion environments; after
the outbreak of World War II, many initiatives in sup
plementary benefits emerged through collective bar
gaining.
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.1
2
Analysis of employee benefit plans continued to ex
pand. One group of studies emphasized provisions of
individual plans. Based on small samples, these analyses
were designed to provide information about the par
ticular benefit plans studied rather than to report on the
overall incidence of plans or plan provisions.3 For ex
4
In accordance with the act, administrators of welfare and pension plans, ex
cluding those for government workers and employees o f nonprofit organizations
having at least 26 participants filed with the Department o f Labor detailed
descriptions o f their plans, including all amendments. Administrators o f plans
having at least 100 participants also had to file annual statistical reports on the
financial status o f their plans. In general, similar filings are now required by the
Department of Labor in accordance with the Employee Retirement Income
Security Act o f 1974 (ERISA), which replaced the Welfare and Pension Plans
Disclosure Act.
3 Early Retirement Provisions o f Pension Plans, 1971, Report 429 (Bureau o f
Labor Statistics, 1974).
6 Employee Compensation in the Private Nonfarm Economy, 1977, Summary
80-5 (Bureau o f Labor Statistics, 1980).
7 For a description o f the occupational wage studies, see
chapter 9.
1 Trade Agreements in 1923 and 1924, Bulletin 393 (Bureau o f Labor
Statistics, 1925); and Beneficial Activities o f American Trade Unions, Bulletin
465 (Bureau o f Labor Statistics, 1928). Some detail on contract provisions for
supplementary benefits is contained in Union Agreement Provisions, Bulletin
686 (Bureau o f Labor Statistics, 1942).
2 Wage Structure o f the Machinery Industries, January 1945, Series 2, N o. 1
(Bureau o f Labor Statistics, 1946).
3 Health-Benefit Programs Established Through Collective Bargaining,
Bulletin 841 (Bureau o f Labor Statistics, 1945).
88
The rapid growth of employee benefits has raised
questions about the validity of a comparability process
limited to wages and salaries alone. Expenditures for
supplementary benefits accounted for nearly one-fourth
of all employer outlays for employee compensation in
1977, the last year they were measured by the Bureau. In
the 1970’s, the General Accounting Office and two
Presidential review groups recommended that the com
parability system be expanded to include both pay and
benefits.
In response to these recommendations, the Office of
Personnel Management (opm ) initiated its Total Com
pensation Comparability (tcc) project. It developed a
method of evaluating and comparing benefits—known
as the “ level of benefits” or “ standardized costing”
method. Under this approach, the provisions of each
benefit plan being evaluated are measured against a
common standard to determine their value; the standard
used is the Federal work force. This “ standard cost”
approach determines the cost of providing non-Federal
sector benefit plans to the Federal work force, and com
pares this with the cost of providing Federal benefit
plans to the same work force. Comparisons of benefit
plans, therefore, are not affected by differences in the
characteristics of the work forces involved, employer
financing procedures, or economic assumptions. Only
actual differences in provisions can affect the relative
worth of the plans. Plans with identical provisions will
always be evaluated as having the same worth.
Because of the Bureau’s long experience in studying
employee benefits, opm asked the Bureau to participate
in the gathering of data on plan provisions and
characteristics. The Bureau participated in a series of
tests to determine the feasibility of collecting and
analyzing the provisions of non-Federal benefits in suf
ficient detail to meet the requirements of opm ’s cost
estimating models.
In 1979, the first full-scale test was conducted in con
junction with the patc survey. Data were collected on
plan provisions and participation for six paid leave
items, including sick, holiday, and vacation pay; health,
life, and disability insurance; and pension plans. Formal
analysis was limited to these benefit areas since they ac
count for a significant portion of personnel costs.8
Starting in 1980, the survey was expanded to include
data on the incidence—but not plan provisions—of ap
proximately 25 other benefit and paid leave items as a
means of providing more information on the wide spec
trum of employee benefits provided in private industry.
The survey is now conducted annually and has replaced
the benefit surveys conducted in the 1960’s and 1970’s.
The remaining discussion is limited to this survey.
8 The last Bureau survey o f employer expenditures for employee compensa
tion, in 1977, showed that the cost o f paid leave, insurance, and pension plans
accounted for 15 percent o f total compensation. Other surveyed private sup
plements to wages and salaries amounted to less than 1.5 percent o f compensa
tion.
89
Description ©f the Surwey
The Bureau’s annual survey of the incidence and
characteristics of employee benefit plans covers private
sector establishments in the United States, excluding
Alaska and Hawaii, employing at least 50, 100, or 250
workers, depending on the industry. Industrial coverage
includes: Mining; construction; manufacturing;
transportation, communications, electric, gas, and
sanitary services; wholesale trade; retail trade; finance,
insurance, and real estate; and selected services.
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.
Sampled establishments are requested to provide data
on work schedules and details of plans in each of the
following benefit areas: Paid lunch periods; paid rest
periods; paid holidays; paid vacation; personal leave;
sick leave; accident and sickness insurance; long-term
disability insurance; health insurance; private retire
ment pension; and life insurance. These 11 benefit
areas, and the specific characteristics studied, were
selected by opm for use in the Total Compensation
Comparability process.
Bls also collects for opm limited data on the in
cidence of the following additional benefits: Funeral
leave; military leave; profit sharing plans; saving and
thrift plans; stock bonus plans; stock purchase plans;
other stock plans; severance pay; employee discounts;
gifts; relocation allowances; recreation facilities; sub
sidized meals; educational assistance; automobile park
ing; personal use o f company-owned car; and in-house
infirmary.
Detailed study of retirement benefits is restricted to
pension plans providing monthly cash income for life to
eligible workers. The analysis of employment-related
health plans yields separate information on such
benefits as hospitalization, surgical, medical, major
medical, dental and vision care, and out-of-hospital
diagnostic and laboratory services.
Data Sources and Collection fVlethods
Data for the survey are collected primarily by visits of
Bureau field representatives to the sampled
establishments. To reduce the reporting burden,
respondents are asked to provide documents describing
their private pension plans and plans covering the four
insured benefit areas within the scope of the survey.
These are analyzed by BLS staff in Washington to obtain
the required data on plan provisions. Whenever possi
ble, the field representative also obtains the identifica
tion number for each plan filed with the Department of
Labor under the reporting requirements of the
Employee Retirement Income Security Act ( e r i s a ) . If
plan documents are not available at the establishment or
are incomplete, the Bureau attempts to obtain the
necessary information from the e r i s a filings. Because
of the time period given plan officials to submit updated
plan summaries, e r i s a material is usually not as current
as descriptions received from the establishment. Plans
which are fully employee paid are not reported. Data on
paid leave and other paid time off generally are obtain
ed directly from the employer at the time of the visit.
Since o p m ’s evaluation of employee benefits is based
largely on plan provisions, respondents are seldom ask
ed to provide data on employer costs. However,
employer contribution rates are requested for certain
collectively bargained multiemployer health, welfare,
and pension plans where benefit amounts are tied direct
ly to the negotiated contribution level, such as in the
construction and trucking industries.
Information obtained from respondents and plan
documents is entered on computer files. These files con
tain the data required for both o p m evaluation of in
dividual benefit plans and b l s estimation of the number
of workers covered by specified plan provisions.9
The data resulting from the analysis and coding of
plan documents are not directly linked to a particular
establishment. Instead, two computer data files are
created—a control file and a plan data file. The control
file contains information on the establishment surveyed,
including: Number of employees, number of plan par
ticipants, industry, geographic location, and sampling
weight.
The plan data file contains the provisions of each plan
for which information was obtained. Plan identification
codes are such that a plan, once analyzed, need not be
analyzed again regardless of how many establishments
report it (e.g., a corporate-wide health insurance pro
gram or a multiemployer pension plan).
Surrey Design
The scope of this survey is the same as 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 the same as that developed for the p a t c . This sam
pling 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 procedures in
clude an effort to ensure that sampling frame units cor
respond to the definition of an establishment adopted
for this survey.
9 BLS provides OPM with the data collected in individual establishments in a
way that does not reveal the identity o f respondents.
90
The sample for this survey is a subsample of the p a t c
sample to reduce the costs and resources required for
data collection. The sample of about 1,500
establishments is selected by first stratifying the sam
pling 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 us
ing 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
establishments in the sample.
Each of the 36 combinations of occupational groups
and work schedule or benefit areas (e.g., health in
surance for production employees) is treated as an in
dividual survey, and separate estimates are developed
for each. This treatment facilitates the use of partially
completed establishment reports in the survey.
Therefore, the actual number of responses for the
survey varies for each of the 36 combinations.
Two procedures are used to adjust for missing data
from partial reports and total refusals. First, imputa
tions are made for the number of plan participants when
the number is not reported. Each of these participant
values is imputed by randomly selecting a similar plan
from another establishment in a similar industry and
size class. The participation rate from this plan is used
to approximate the number of participants for the plan
which is missing a participation value but is otherwise
usable. For other forms of missing data (or
nonresponse), an adjustment is made using a weight ad
justment 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 respondent is multiplied by a factor
equal to the total employment in the cell divided by the
employment of the responding units. The weight ad
justments 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 com
pleted 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 developed during the adjust
ment for nonresponse. The estimator, after modification
to account for the weight adjustment factor, fi} devel
oped during the adjustment for nonresponse, is:
Y=
I
i= l
—
-Pi
where:
nj
fi
Yi
Pi
=
=
=
=
Uses and limitations
number of responding units
weight adjustment factor for the ith unit
value for the characteristic of the ith unit
the probability of including the ith unit in the
sample
Sampling and estimating procedures are designed to
meet the specific needs of the Office of Personnel Man
agement for national data for all studied industries com
bined. Survey findings do not yield reliable estimates for
individual industries or geographic regions. Data are,
however, reported separately for three occupational
groups—professional-administrative, technical-clerical,
and production workers.
Presentation
Annual b l s bulletins summarize major survey find
ings. Estimates show the percent of employees that are
covered by paid leave plans, participate in insurance or
pension plans, or are eligible for other benefits. Counts
of workers covered by benefit plans include those who
have not met possible minimum length-of-service re
quirements 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. 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.
A 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
percent of all employees with health insurance receive
dental coverage?
The third approach provides a closeup look at an im
portant feature of the plan; for example, what percent
of all employees with dental coverage in their health in
surance are covered for orthodontic work?
The Bureau develops and publishes selected estimates
from the data of general interest to the public. There is a
wide range of detailed information on the employee
benefits for which estimates and tables are not
developed by b l s . Although these items are too narrow
in scope to warrant publication, researchers may pur
chase computer tapes of the survey results (with
establishment identifying information removed).
91
The survey of the incidence and characteristics of
employee benefits was designed for o p m as a major part
of the Total Compensation Comparability system
described previously. The survey scope and design and
the selection of employee benefit characteristics were
dictated by the requirements of o p m ’s actuarial models.
The extensive body of information on employee benefits
generated in this survey provides a unique data
resource.
Results of the survey provide a major source of data
on the extent to which workers receive paid leave and
are protected by job-related insurance and pension
plans. The survey data provide information for labor
and management representatives involved in contract
negotiations, State and Federal conciliators and
mediators, public and private arbitrators, Members of
Congress and Congressional staff considering legisla
tion affecting the welfare of workers, and government
officials responsible for recommending legislation and
reviewing proposed legislation, b l s tabulations and
analyses of employee benefits can be of use to teachers,
students, and others in the academic field; private con
sultants; researchers; writers; and others not directly in
volved in legislation or collective bargaining but con
cerned with the development, status, and trends in
employee benefits.
Users of the employee benefits survey data should
keep in mind that the scope of the survey excludes small
firms—those with up to 50, 100, or 250 employees,
depending on the industry. Studies of employee benefits
that include all firms typically report lower participation
rates for most benefits. The survey also excludes ex
ecutive management and traveling operating employees
(such as airline pilots), as well as part-time, temporary,
and seasonal employees. Alaska and Hawaii are not
surveyed; neither are the public sector and some in
dustries such as agriculture, education, and health serv
ices. The data, therefore, do not statistically represent
all employees in the United States, or even all employees
in private industry.
Technical References
Employee benefits—general
Bureau of Labor Statistics. Employee Benefits in Industry:
A Pilot Survey, Report 615, 1980.
Blostin, Allan P. “Noninsured Death Benefits Under Union
and Company Programs,” Monthly Labor Review,
October 1977.
Bureau of Labor Statistics. Employee Benefits in Industry,
1980, Bulletin 2107, September 1981.
Bureau of Labor Statistics. Digest of Selected Health and
Insurance Plans, 1977-79 Edition, Vol. I: Health
Benefits; Vol. II: Insurance Benefits, 1978 and sup
plements.
U.S. Office of Personnel Management. Total Compensation
Comparability: Background, Method, Preliminary
Results, July 1981.
Kittner, Dorothy R. “Changes in Health Plans Reflect
Broader Benefit Coverage,” Monthly Labor Review,
September 1978.
Pension plans
Bureau of Labor Statistics. Digest of Selected Pension Plans,
1976-78 Edition, 1977 and supplement.
Kittner, Dorothy R. “Maternity Benefits Available to Most
Health Plan Participants,” Monthly Labor Review,
May 1978.
Frumkin, Robert, and Schmitt, Donald. “Pension Improve
ments Since 1974 Reflect Inflation, New U.S. Laws,”
Monthly Labor Review, April 1979.
U.S. Department of Health, Education, and Welfare, Social
Security Administration. “Private Industry Health In
surance Plans: Employment Requirements for Cover
age in 1974,” Social Security Bulletin, March 1977.
Kittner, Dorothy R. “Forced Retirement, How Common Is
It?” Monthly Labor Review, December 1977.
Schulz, James H. “Private Pensions Fall Far Short of Pre
retirement Income Levels,” Monthly Labor Review,
February 1979.
U.S. Department of Health, Education, and Welfare, Social
Security Administration. “Private Industry Health In
surance Plans: Type of Administration and Insurer in
1974,” Social Security Bulletin, March 1977.
Srasyram
xsts plans
Bell, Donald R. “Dental and Vision Care Benefits in Health
Insurance Plans,” Monthly Labor Review, June 1980.
Blostin, Allan P. “Is Employer-Sponsored Life Insurance
Declining Relative to Other Benefits?” Monthly Labor
Review, September 1981.
Directory
Bureau of Labor Statistics. A Directory of b l s Studies
in Employee Compensation, 1960-75, 1975.
92
Chapter 13. Produetiwity
Measures: Business
Economy and Major
Sectors
time in these major indexes reflect efficiency in the use
of labor, and indirectly, the effect of other input factors
in the domestic production of goods and services. The
changes in the productivity and related measures
through the business cycle typically show patterns which
differ substantially from those found in long-term
movements, and, therefore, are the objects of special
analytic studies.
Labor input measures are based primarily on b l s
establishment payroll data on employment and hours
and reflect hours at work and paid time off for vaca
tions, holidays, and sick leave as well. 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. (See b l s form 2000M at end of chapter.)
Indexes of compensation per hour measure the hourly
cost to employers of wages and salaries, as well as sup
plemental payments, which include employers’ con
tributions 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 AH Urban
Consumers (CPi-U).
Unit labor cost measures the cost of labor input re
quired to produced one unit of output and is 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, in
terest, and indirect business taxes, in addition to cor
porate profit and profit-type income of proprietorships
and partnerships.
The Bureau has begun a program of multifactor pro
ductivity measurement to supplement the labor produc
tivity measures and to provide additional insights into
productivity growth and economic changes. This pro
gram is an outgrowth of analytic studies undertaken by
the Bureau investigating some of the factors con
tributing to productivity growth.3 The multifactor pro
ductivity measures for the business and nonfarm
business sectors will be published based upon capital
and labor inputs. Subsequently, measures based on
Indexes of labor productivity and compensation per
hour, unit labor cost, and related measures for broad
economic sectors are published by the Bureau of Labor
Statistics. These measures provide information about
the relationship between productivity, prices, wages,
employment, and economic growth. Measures of output
per hour have been developed for the business sector,
and nonfarm and farm subsectors, from 1909 to the pre
sent. Since 1947, these data have been supplemented
with comparable measures of compensation and costs
and corresponding series for manufacturing (total,
durable, and nondurable) and nonfinancial corpora
tions. For the latter period, indexes are available
quarterly as well as annually. These productivity
measures, first published in 1959, represent the culmina
tion of a long series of developments in productivity
measurement in the Bureau.1
Productivity and related measures are prepared for
the following sectors of the U.S. economy:
Quarterly and annual measures
Business sector
Nonfarm business sector
Nonfinancial corporations
Manufacturing, total, durable, nondurable
Annual measures only
Agriculture
Mining
Transportation
Communications
Utilities
Wholesale and retail trade
Finance, insurance, and real estate
Government enterprises
Description of treasures
The Bureau’s output per hour measures are con
structed as the ratio between gross domestic prod
uct—g d p —originating in the private business economy
and its subsectors, and the corresponding hours of all
persons engaged in each sector.1 The changes through
2
1 Trends in Output per Man-Hour in the Private Economy, 1909-58, Bulletin
1249 (Bureau o f Labor Statistics, 1959).
2 Gross domestic product is gross national product less the net return on
foreign investments. Net return on foreign investments is considered as
originating in the “ rest-of-world’’ sector:
3
J.R. Norsworthy, Michael Harper, and Kent Kunze, “ The Slowdown in
Productivity Growth: Analysis o f Some Contributing Factors,” Brookings
Papers on Economic Activity, Fall, 1979.
93
gross product of nonprofit institutions and private
household workers.
Output data for the manufacturing sector based on
gross product are derived 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.
capital, labor, energy, and materials4 will be introduced
for the manufacturing sector and 2-digit Standard In
dustrial Classification (Sic) manufacturing industry
groups. For these new measures, labor input is
measured according to the same principles as in the pre
sent labor productivity measures. For the private
business and nonfarm business sectors, output is also
measured according to the same principles as in the pre
sent measures. When energy and materials are introduc
ed, new measures of output for the manufacturing sec
tor will also be developed.
Data Sources and Estimating
Procedures
Labor input
The primary source of hours and employment data is
the BLS Current Employment Statistics (CES) program,
which provides monthly survey data on total employ
ment and average weekly hours of production and nonsupervisory workers in nonagricultural establishments.
Jobs rather than persons are counted, so that multiple
jobholders are counted more than once. Weekly hours
are measured as hours paid rather than hours at work.
These statistics 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 CES methods are described in
chapter 2.) Establishment data are published monthly in
Employment and Earnings.
Output
Real gross domestic product originating in the
business sector and subsectors is the basis of the output
component of the productivity estimates. Thus, the out
put components of the entire set of measures are based
upon and consistent with the National Income and
Product Accounts ( n i p a ) prepared by the U.S. Depart
ment of Commerce. Gross product is the market value
of final goods and services produced within a given
period, and includes purchases of goods and services by
consumers, gross private domestic investment, net
foreign investment, and purchases by government.
Gross national product (GNP) is equal to income
received by labor and property for services rendered in
the current production of goods and services, in addi
tion to capital consumption allowances, indirect
business taxes, and several other minor items. Gross
domestic product (GDP) is simply gross national product
less “ rest-of-world” output, and excludes net factor
payments to domestic owners of factors of production
located outside the United States.
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 (BEA) in the U.S. Depart
ment of Commerce prepares estimates of constantdollar g d p for the business economy and its major sec
tors; these estimates exclude changes in the value of pro
duction resulting from price change. Therefore, they
reflect only changes in real product, which is the basis
for output-per-hour measures.5
Output for the business economy equals g d p less
gross housing product of owner-occupied dwellings and
Compensation and iabor costs
4 The term “ materials” is commonly used in this context to describe all in
termediate goods and services exclusive o f energy.
s A detailed description o f the methods and procedures for estimating GNP
and GDP in current and constant dollars is given in the 1954 National Income
Supplement to the Survey o f Current Business, U .S. Department o f Commerce.
Further information on estimates for major industry sectors is presented in the
October 1962 issue o f the Survey o f Current Business.
94
B e a develops employee compensation data as part of
the national income accounts. These quarterly data in
clude direct payments to labor—wages and salaries (in
cluding executive compensation), commissions, tips,
bonuses, and payments in kind representing income to
the recipients—and supplements to these direct
payments. Supplements consist of employer contribu
tions to funds for social insurance, private pension and
health and welfare plans, compensation for injuries,
etc.
The compensation measures taken from establish
ment payrolls refer exclusively to wage and salary
workers. Labor cost would be seriously underestimated
by this measure of employee compensation alone in sec
tors such as farm and retail trade, where hours worked
by proprietors represent a substantial portion of total
labor input. B l s , therefore, imputes compensation 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.
Unit labor and nonlafoor <e@
sts
The Bureau also prepares data on labor and nonlabor
costs per unit of output for the business sector and its
major components. Unit labor cost relates hourly com
pensation of all persons to output per hour and is defin
ed as compensation per unit of constant-dollar output.6
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, and nonlabor
cost: Interests, depreciation, rent, indirect business
taxes, etc.
Since ces data include only nonfarm wage and salary
workers, data from other sources (National Income and
Product Accounts or the Current Population Survey)
are used for farm employment and, in the nonfarm sec
tor, proprietors, unpaid family workers, and private
household workers.
Separate estimates for employment and hours paid
are developed for each major sector and are aggregated
to private 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.
In the manufacturing sector, separate estimates for
production and nonproduction worker hours are deriv
ed and aggregated to the manufacturing total. Employ
ment and average weekly hours for production workers
are taken directly from CES data. Average weekly hours
for nonproduction workers are developed from bls
studies of wages and supplements in the manufacturing
sector which provide data on the regularly scheduled
workweek of white-collar employees.
For nonmanufacturing sectors, employment and
average weekly hours are taken from the ces survey.
Although c e s weekly hours data refer only to nonsupervisory workers, it is assumed for hours computation
that the length of the workweek in each nonmanufactur
ing industry is the same for all wage and salary workers.
MyStifaetor prodyetiwity nn®asyr@s
The major new elements of multifactor productivity
measurement, as distinct from labor productivity
measurement, for the private business sector are the
measurement of capital and the method of aggregating
dissimilar inputs. The capital measures will be
developed from investment data in the national ac
6 Unit labor cost is explained and discussed in detail in J.R. Norsworthy and
L.J. Fulco, “ Productivity and Costs in the Third Quarter,” Monthly Labor
Review, February 1976.
counts using the same methodology for both the private
business and nonfarm business sectors.
The multifactor productivity measures for the
business and nonfarm business sectors will be developed
from 1948 to the present. The subsequent measures for
the manufacturing sectors and the 2-digit sic manufac
turing industry groups will be developed from 1958 for
ward.
amd Presentation
Indexes of output per hour show changes in the ratio
of output to hours of labor input; however, these in
dexes should not be interpreted as representing solely
labor’s contribution to production. Rather, they reflect
the interaction of many factors working in cooperation
with the hours of labor input, including technology,
capital investment, human capital (education and skill),
energy, and raw materials.
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
productivity and multifactor productivity are influenced
by the relative shift of inputs (labor and capital) from
low- to high-productivity industries and by productivity
changes In the component sector.7 Within industries,
other shifts occur which are not accounted for ade
quately—changes in income and tastes, for example,
may contribute to shifts in consumption patterns to
higher quality goods, or to services rather than goods.
Short-term movements in productivity and unit labor
costs often result from cyclical variation in output; this
tends to distort the long-term relationship between out
put and labor input, as noted below, or output and
multifactor input. A number of studies are being con
ducted to separate cyclical from long-term productivity
movements.
Indexes of output per hour, compensation per hour,
and related cost data are published quarterly in two
series of b ls news releases: ‘‘Productivity and Costs in
the Business Sector,” and “ Productivity and Costs in
Nonfinancial Corporations.” In addition, quarterly and
annual analyses are published regularly in the Monthly
Labor Review. Historical indexes of these and related
data are available on request, as are detailed descrip
tions of data sources and computational procedures.
Indexes of output per hour and related cost data are
published monthly in Employment and Earnings and
the Monthly Labor Review, and in each edition of the
Handbook o f Labor Statistics.
7
The farm-nonfarm shift is examined in some detail in J.R. Norsworthy and
L.J. Fulco, “ Productivity and Costs in the Private Economy, ” Monthly Labor
Review, June 1974, pp. 3-9.
95
Calculation Procedures
UNLC = (OC-C-PR) / O
where:
Labor productivity
Labor productivity or output per hour, is computed as:
OC is current-dollar gross product originating
C is current-dollar compensation
O is constant-dollar output
PR is current-dollar profits
T ,
, . .
Constant-dollar output
Labor productivity = -------------------------- ------Hours of labor input
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:
or
P = O/H
In instances where several sectors are involved, labor
productivity can be computed equivalently as
LS = C / OC
P = ( 2 ^ / 2 ^
and, analogously, the nonlabor or capital share is defined
as
or as
CS = (OC-C) / OC = 1-LS
P = I jWj (Oj/H j)
where:
is constant-dollar output in sector i
Hj is hours of labor input in sector i
Wj = H j/ 1 j Hj is the hours-based weighting factor
for sector i
P is average labor productivity for the aggregate sector
Most of the measures noted above are prepared quar
terly in index form for the major sectors of the private
economy. In addition, quarterly percent changes at a
compound annual rate and percent changes from the
same quarter in the previous year are computed:9
Qt = lO O O yV t.j)4- 100
Yt = 100 (Vt/Vt_4) - 100
The computation of labor compensation per hour is
equivalent to the computation of output per hour. Unit
labor cost ( u l c ) is computed as labor compensation (c)
per unit of (constant dollar) output, but is often repre
sented as:
where:
ULC = (C/H) - (O/H)
This form highlights the relationships among unit labor
cost, hourly compensation, and labor productivity.
Real compensation per hour ( r c ) is computed as hourly
compensation deflated by the seasonally adjusted Con
sumer Price Index for All Urban Consumers (CPI-U):8
RC = (C/H) -s- CPI-U
Unit nonlabor payments ( u n l p ) include all nonlabor
components of gross product originating in a given sec
tor—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 = (OC-C) / O
t is a time subscript denoting the quarter
V is a series described above
Qt is the quarterly percent change in series V from
quarter t-i to quarter t, measured at a compound
annual rate
Yt is the percent change in series V from quarter t-4
(the same quarter 1 year before) to quarter t
In order to achieve greater precision in reported meas
ures, all computations are made from the measures
themselves rather than from their corresponding indexes.
Multifactor productivity
The computational method used by b l s for its multi
factor productivity measure is defined as a Tornquist in
dex. (A Tornquist index is the discrete approximation to
the continuous Divisia index.) Some of the basic proper-
9
The handling o f quarterly (or subannual) changes at compound annual rates
involves approximations. For changes in the neighborhood o f 1 or 2 percent,
these approximations are good; however, the inexactness o f these approxima
tions is amplified by relatively large changes in the economic measures such as
those caused by the recent periods o f inflation, sharp recession, and rapid
recovery.
Since most o f the productivity and cost measures are reported as percentages
8
Changes in real hourly compensation are analyzed in J.R. Norsworthy and to one decimal place, e.g., 2.6 percent, questions sometimes arise because the
greater precision carried in the automated computation results in differences in
L.J. Fulco, “ Productivity and Costs, First Quarter 1976,” Monthly Labor
the final decimal place.
Review, July 1976.
and
96
ties 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 pro
ductivity measures, the weights are defined as the
weighted average of the relative compensation shares of
the components in the two adjacent time periods.
Hence, the growth rate of the index (i/I) is the propor
tional change over time (the dot notation refers to
change with respect to time), such that:
1/1
l iwit<i it/xit)
-
where Xjt/xjt is the growth rate of the ith input, calulated as
V xit = ln x it -
ln x i n
The weights (wjt) are defined as the average of the
relative compensation shares of all the inputs;
wit =<si t + sit - i ) /2
_ p it*it
s >t —
—
z ip it xit
Pjt _ price or wage of input xj in period t.
Multifactor productivity growth is defined as the growth
rate in output (O/O) less the growth rate in aggregated
inputs:
MFP = 0 / 0 - I/I
where:
= wk K/K + WjL/L
wjc = relative compensation share of capital
wj = relative compensation share of labor
K/K = growth in capital services
L/L = growth in hours at work
I/I
Us@s and Limitations
Measures of output per hour, output per unit of input
(multifactor productivity), and related costs are designed
for use in economic analysis and public and private
policy planning. The data are used in forecasting and
analysis of price, wage, and technological change.
The labor productivity, multifactor productivity, and
related cost measures are useful in understanding and
investigating the relationships among productivity,
wages, price, profits, and costs of production. As noted
above, gross domestic product represents the sum of all
production costs: Compensation, profits, depreciation,
interest, rent, indirect business taxes, etc. Unit labor
cost, or compensation per unit of output, represents a
major portion of total unit costs and so reflects the com
bined effect of changes in output per hour and compen
sation per hour; thus, an increase in compensation per
hour tends to increase unit labor cost while an increase
in output per hour tends to reduce it, other things being
equal, Therefore, through its impact on unit labor cost,
output per hour is an important element in the wageprice 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 in various constituent
industries as well as shifts in the relative importance of
these industries: A significant portion of labor produc
tivity growth from 1947 to the present is attributable to
the relative shift of workers from the farm to the non
farm sector. 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 ef
fects of short-term swings in product demand. Third,
data and other resources available for estimation
somewhat limit the productivity, output, compensation,
and employment measures which can be constructed. In
several sectors where output is difficult to define in a
satisfactory way, productivity measures are correspon
dingly weak. Examples are the construction industry
and financial services sector where output is an imputed
value of labor and other inputs. In consequence, the
productivity and cost measures for these sectors should
be interpreted with caution.
Technical References
Bureau of Labor Statistics
Harper, Michael J. “The Measurement of Productive Capital
Stock, Capital Wealth and Capital Services,” Working
Paper No. 128, 1982.
Analysis of the formulation of capital depreciation for
productivity measurement.
Mark, Jerome A. Current Developments in Productivity,
1973-74, bls Report 436, 1975.
Summarizes recent developments in productivity in
industry and aggregate sectors with attention to possible
sources of the productivity slowdown since 1966.
97
Ts©hni©al Ref©r©ness-=C©ritSnu@d]
Denison, Edward F. Accounting for Slower Economic
Growth, The United States in the 1970’s. Washington,
D.C., The Brookings Institution, 1979.
Mark, Jerome A. “Productivity and Costs in the Private
Economy, 1974,” Monthly Labor Review, June 1975.
Annual review article for 1974. Recent rates of capital
formation are examined.
Denison, Edward F. Accounting for United States Economic
Growth, 1929-1969. Washington, D.C., The Brookings
Institution, 1974.
Meaning and Measurement of Productivity, b l s Bulletin 1714,
1971, prepared for the National Commission on Produc
tivity.
An integrated discussion of methods for measuring
labor productivity and the interpretations consistent with
the methods.
Denison, Edward F. Why Growth Rates Differ, Sources
of Economic Growth. Washington, D.C., The Brookings
Institution, 1967.
A study of output and productivity growth in nine
Western countries. Includes a discussion of factors
affecting productivity growth and the effects of these
factors in contributing to differential growth rates
between countries.
Norsworthy, J.R., and Fulco, L.J. “Productivity and Costs
in the Private Economy, 1976,” Monthly Labor Review,
September 1977.
Annual review article for 1976. Examines relations
among growth rates in labor productivity and hourly
compensation. Updates the shift analysis between farm
and nonfarm sectors.
Greenberg, Leon, and Mark, Jerome A. “Sector Changes in
Unit Labor Costs,” The Industrial Composition of
Income and Product. New York, National Bureau
of Economic Research, 1968.
Norsworthy, J.R., and Fulco, L.J. “Productivity and Costs
in the Private Economy, 1975,” Monthly Labor
Review, May 1976.
Annual review article for 1975. Examines relations
among growth rates in labor productivity, capital
productivity, and the capital/labor ratio. Shift of hours
from the farm to the nonfarm sector since 1909 is
presented.
Kendrick, John W. Postwar Productivity Trends in the
United States, 1948-1969. New York, National Bureau
of Economic Research, 1973.
Presents historical measures of output, factor input,
and productivity for the U.S. economy and industry
groups, including descriptions of concepts and methods
of measurement. Also includes discussion of implications
of productivity change for economic growth, prices, in
comes, and resource allocation.
Norsworthy, J.R., and Fulco, L.J. “Productivity and Costs
in the Private Economy, 1973,” Monthly Labor Review,
June 1974.
Annual review article for 1973. Examines productivity
effects of farm-to-nonfarm shift since 1947.
Mark, Jerome A. Wage-Price Guidepost Statistics: Prob
lems of Measurement. American Statistical Association,
Proceedings of the Business and Economic Statistics
Section, 1968.
Describes some of the problems of developing the
measures which were used in the specification of the
guideposts.
Norsworthy, J.R., and Harper, Michael J. “The Role of
Capital Formation in the Recent Slowdown in Produc
tivity Growth,” Working Paper No. 87, 1979.
Analysis of capital growth and the productivity slow
down using Tomquist index growth rates.
Productivity: A Selected Annotated Bibliography 19761978, b l s Bulletin 2051, 1980.
Nearly 800 references concerning productivity and pro
ductivity measurement. Each reference includes a brief
annotation.
National Bureau of Economic Research. New Developments
in Productivity Measurement and Analysis, Studies in
Income and Wealth, Vol. 44. Chicago, The University
of Chicago Press, 1980.
Other publications
Caves, Douglas W.; Christensen, Laurits R.; and Diewert, W.
Erwin. “A New Approach to Index Number Theory and
the Measurement of Input, Output, and Productivity,”
SSRI Workship Series No. 8112. Madison, Wisconsin,
University of Wisconsin-Madison, Social Systems
Research Institute, May 1981.
National Bureau of Economic Research, The Measurement of
Capital, Studies in Income and Wealth, Yol. 45. Chicago,
The University of Chicago Press, 1980.
National Research Council Panel to Review Productivity
Statistics. Measurement and Interpretation of Produc
tivity. Washington, D.C., National Academy of Sciences,
1979.
Christensen, L.R., and Jorgenson, E.W. “U.S. Real Product
and Real Factor Input, 1929-1967,” Review of Income
and Wealth, Series 16, March 1970.
Outline of the theoretical framework and empirical
application of an integrated approach to measuring
labor and capital input.
Norsworthy, J.R.; Harper, Michael J.; and Kunze, Kent.
“The Slowdown in Productivity Growth: Analysis of
Some Contributing Factors,” Brookings Papers on
Economic Activity No. 2S 1979. Washington, D.C. The
,
Brookings Institution, 1979.
98
Bureau of Labor Statistics
Hours A t Work
Mining, Manufacturing, and Construction
T h is r e p o r t is a u t h o r iz e d b y la w 2 9 U .S .C . 2 .
c o m p r e h e n s iv e , a c c u r a t e , a n d t i m e l y .
U.S. Department o f Labor
Y o u r v o l u n t a r y c o o p e r a t io n is n e e d e d t o m a k e t h e r e s u lt s o t t h is s u r v e y
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 is f o r m
F o rm A p p ro v e d
b y t h e B u re a u o f L a b o r S t a t is t ic s w i l l b e
h e ld in c o n f id e n c e a n d w i l l b e u s e d f o r s t a t is t ic a l p u r p o s e s o n l y .
O .M .B . N o . 1 2 2 0 - 0 0 7 6
A p p r o v a l E x p ir e s 1 2 - 3 1 - 8 2
BLS Use Only
n
r
jicfluw h tttm
2nd m )
RETURN TO:
L
B U R E A U OF LA B O R S T A T IS T IC S
OSO, F/SMS, HW
ROOM 2068, M A IL CODE 04
441 G STR EE T, N.W.
W A S H IN G T O N , D.C. 20212
J
(C h a n g e n a m e & m a ilin g a d d re s s i f in c o r r e c t )
Telephone No. 202-523-1049
P L E A S E R E A D IN S T R U C T IO N S O N R E V E R S E B E F O R E E N T E R IN G D A TA
BLS Use Only
Production Workers
All Employees
Number of Employees During Payroll
Period which Includes March 12, 1981
j; 18-32
[ 13-17
| 23 -24
Production Workers Only
Quarterly Period
Hours Paid
Hours at Work
( o m i t fr a c tio n s )
( o m it fr a c tio n s )
t 23-32
f 33*40
41 42
fla S o ”
| ' M *5*
i
I 614»
j '69-76
[ 7SW
S6
f ii iiil
| 37-105
F irst Q uarter 1981
1
100*114 j
January—March
Second Quarter 1981
-
-u
A p r il—June
T hird Quarter 1981
;. 7 7 - 7 8
J u ly —September
F ourth Q uarter 1981
O ctober—December
Annual T ota l 1981
'«
05-95
116-115
January—December
•
1a. W h a t r e c o r d s d id y o u u se t o c o m p il e t h e a b o v e i n f o r m a t i o n ?
1n
P a y r o li
2 IT
P e rs o n n e l
j* w
3 I- ! O t h e r
110
b . A r e th e s e r e c o r d s c o m p u t e r iz e d ?
2.
□
Yes
2 D
No
|
O n w h a t b a s is a re th e s e r e c o r d s k e p t ?
1D
3.
1
W e e k ly
2 D
B i w e e k ly
3 D
M o n t h ly
4 (1
Q u a r t e r ly
6 f1
A n n u a l ly
6 IT
t ie .
O th e r
I f y o u r r e c o r d s c a n n o t b e e a s ily t r a n s c r ib e d t o r e f l e c t q u a r t e r ly t o t a ls , p le a s e a t t a c h a d d it io n a l s h e e ts r e p o r t i n g t h e d a t a r e q u e s te d a b o v e in t h e f o r m a t
in w h ic h y o u r r e c o r d s a re k e p t .
4.
E n t e r b e lo w a n y u n u s u a l f a c t o r s r e s p o n s ib le f o r s ig n if ic a n t d if f e r e n c e s f r o m
th e s e f a c t o r s o c c u r r e d .
E x a m p le s a re :
n o r m a l h o u r s w o r k e d d u r in g a n y q u a r t e r .
P lease in d ic a t e w h ic h q u a r t e r
M o r e b u s in e s s , la y o f f s , s t r ik e s , f ir e , w e a t h e r , e t c .
I f q u e s tio n s a ris e c o n c e r n in g t h is r e p o r t , w h o m s h o u ld w e c o n t a c t ?
Nam e
B L S - 2 0 0 0 M (J a n . 1 9 8 2 )
T itle
D a te
A re a C o d e
T e le p h o n e
IN S T R U C T IO N S FOR C O M P L E T IN G R EP O R T (BLS 2 0 0 0 M )
ALL EMPLOYEES
Enter the to ta l num ber o f persons on the payroll(s) w h o w o rked fu ll- o r p a rt-tim e o r received pay fo r any part o f the period w h ich
includes March 12, 1981.
PRODUCTION WORKERS
Enter the to ta l num ber o f p ro d u ctio n workers, both fu ll and part-tim e, on yo u r payroll(s), w hether wage or salaried, w h o w o rked during
o r received pay fo r any part o f the pay period reported w h ich includes March 12, 1981.
The term "p ro d u c tio n w o rk e r" refers to all occupational groups whose w o rk is not p rim a rily adm inistrative or managerial, regardless o f
skill level w ith in the fo llo w in g industries: M ining and Q uarrying, Crude Petroleum , Natural Gas and Natural Gasoline P ro d u ction , and
the C onstruction and M anufacturing industries. These occupational groups include: W orking supervisors and all nonsupervisory w orkers,
(including group leaders a n d trainees) engaged in excavation, hauling, tru c k in g , hoisting, v e n tila tio n , drainage, pum ping, d rillin g , blasting,
loading, crushing, processing, inspection, storage, handling, warehousing, shipping, maintenance, repair, ja n ito ria l, record keeping, fa b ri
cating and assembly, as w ell as c ra ft w orkers, mechanics, apprentices, helpers, laborers, plum bers, painters, plasterers, carpenters, masons,
welders o r any o f th e special trades. A lso include all o th er nonsupervisory employees whose services are closely associated w ith those
employees above.
The term "p ro d u c tio n w o rk e r" excludes employees engaged in the fo llo w in g activities: Executive, purchasing, finance, accounting, legal,
personnel, cafeterias, m edical, professional and technical a ctivities, sales, advertising, cre d it co lle ctio n , and in the insta lla tio n and servic
ing o f own products, 'ro u tin e o ffic e fu n ctio n s and fa c to ry supervision (above w orking supervisor's level). (Em ployees in the above
activities, however, should be in clu d ed in the A L L E M P L O Y E E S figure.)
PERIOD. N o rm a lly, data w ill refer to calendar quarter, i.e., fro m January 1 through March 31; A p ril 1 thro u g h June 30; J u ly 1 thro u g h
September 30; and O ctober 1 through December 31. If y o u r records relate to a period oth er than the calendar quarter, please indicate
beginning and closing dates.
HOURS PAID. Include all hours fo r w h ich pay is received d ire c tly fro m the em ployer. Include paid vacation tim e, paid sick leave, paid
holidays and o ther paid personal o r adm inistrative leave. If payments are made in lieu o f tim e o ff, re p o rt the hours equivalent to the
payments made. For example, three hours leave tim e at tw o -th ird s the regular rate should be reported as tw o hours paid. Exclude hours
associated w ith unpaid leave, norm al travel tim e fro m home to w o rk , unpaid washup tim e, and unpaid meal tim e.
HOURS AJ WORK. Include all tim e an em ployee is required to be on the em ployer's premises, on d u ty , o r at a prescribed w o rk
place. Include, besides norm al w o rkin g hours, rest periods, stand-by tim e, do w n tim e , travel tim e fro m jo b site during w o rkin g day
and travel tim e away fro m home if it cuts across w o rkin g day. Do n o t convert overtim e o r prem ium paid hours to straight-tim e equiva
lent hours.
NOTE: For survey purposes HOURS A T W O RK equals HOURS PA ID less paid leave tim e (vacation, sick leave, holidays, and oth er
paid personal o r adm inistrative leave).
ANNUAL TOTAL. The sum o f HOURS PA ID fo r each quarter should equal the A N N U A L T O T A L , HOURS P A ID figure. Likewise,
the sum o f HOURS A T W O RK fo r each quarter should equal the A N N U A L T O T A L , HOURS A T W O RK figure.
100
Chapter 14. Productivity
Measures: industries and
th@ Federal Government
Background
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 an
nual basis and are provided for most years between 1947
or 1958 and the most recent year for which data are
available.
Coverage has been expanded to include industries in
trade and services, and with the increasing importance
of the public sector, to various functional areas in the
Federal Government. Productivity measurement in the
Federal Government was initiated by a request from the
Joint Economic Committee in the fall of 1970 to the
General Accounting Office in conjunction with the Of
fice of Management and Budget and the Civil Service
Commission (now Office of Personnel Management). A
joint Federal productivity measurement task force con
sisting of these agencies, with technical assistance and
support from b l s , was established. This task force col
lected data and constructed indexes for fiscal years
1967-71. In July 1973, the Office of Management and
Budget endorsed the continuation of the project to
measure Federal productivity, and BLS assumed full
responsibility for collecting input, output, and related
information, in addition to the development of produc
tivity measures.
Since July 1973, the Bureau has been expanding
coverage to include organizational units not previously
covered, improving the quality of some of the input and
output data, and refining the methodological pro
cedures used to construct productivity indexes. The
measurement program is part of a multifaceted effort
sponsored by the Office of Personnel Management and
includes analysis, enhancement, and diffusion of pro
ductivity improvement ideas.
The Bureau is also expanding its productivity
measurement program by explicitly accounting for
other inputs besides labor in the industry measures and
developing a supplementary set of productivity
measures. The new measures are referred to as multifac
tor productivity measures. (The conceptual basis for
Studies of output per employee hour in individual in
dustries 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 re
mained an important focus of b l s throughout the
1920’s and 1930’s. Also during this period, the Bureau
began the preparation and publication of industry in
dexes of output per employee hour, which were based
on available production data from the periodic Census
of Manufactures and employment statistics collected by
BLS.
In 1940, Congress authorized the Bureau of Labor
Statistics to undertake continuing studies of productivi
ty 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 in
dustries. 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
unemployment to concern with the most efficient
utilization of scarce labor resources, b l s undertook a
number of studies of labor requirements for defense in
dustries, 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 readily
available data can be used to construct measures.
In recent years, public interest in productivity has
grown, and increases in output per employee hour have
101
these is discussed in chapter 13 with reference to the ma
jor sectors of the economy.) The industry multifactor
series are designed to measure changes in productivity
by relating changes in an industry’s output not only to
changes in labor input but also to changes in capital and
intermediate purchases. In addition to providing in
dicators of productivity change useful for analysis in
their own right, such measures also are helpful in
analyzing the causes of change in output per employee
hour or labor productivity.
Concepts
Indexes of output per employee hour measure
changes in the relationship between the physical volume
of an organization’s output and the employee hours ex
pended in that output. Although traditionally output
per employee hour has been the most frequently used
measure, the expression of physical output per
employee hour often is simplified if stated in terms of its
reciprocal: Employee hour requirements per unit of out
put (unit employee hours). This form of index is used in
the following description.
For an industry producing a single uniform output,
the unit employee hours index is simply the ratio of the
employee hours expended to produce a unit of output
over two periods of time. This ratio may be expressed as
follows:
b. Using a base period composite
r _
*Qo li
iu “ ----------^o
An index constructed according to (a) compares the
employee hours expended in the production of the cur
rent composite with the employee hours which would
have been required to produce the current composite in
the base period. An index constructed according to (b)
compares the employee hours required in both periods
to produce the base period composite. Thus, these in
dexes eliminate the effects of variations over time in the
relative importance of products or services on unit
employee hours.
In either form, an index of unit employee hours also
can be viewed as the quotient of an index of employee
hours and an index of output:
Employee hours Output index Unit employee hours
index
+ (Laspeyres) = index (Paasche)
=
u
/ - I - —L .
i
—
fV ? j = ^ jg j
XlQq 0
2 /0<Zi
Employee hours Output index Unit employee hours
index
“ (Paasche) = index (Laspeyres)
j
ill
^
Zl0 q 0 '
_
U
_ ^ i ffp
2 /0 <
?0
2 /i^ o
2 /0 4 0
The employee hours index measures the change in ag
gregate employee hours between the base and current
periods. The employee hours data are the total hours ex
pended 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 unit employee hours index uses a base period
weighted output index divided into the employee hours
index. Conversely, a base period weighted unit em
ployee hours index is consistent with an output index
which utilizes current period weights.
where:
/u represents the unit employee hour index
7r represents the output per employee hour index
Lj and L Q denote unit employee hours expended in
the current and base periods, respectively.
For an industry producing a number of products or
services (the more typical case), the unit employee hours
index is the ratio for two periods of the total hours re
quired for the output of a given composite of products
or services. Indexes of such industries vary with the
composite specified and can take many forms. Letting
Q0 and Q\ represent base period and current period
quantities of a given product respectively, two of these
forms are:
a. Using a current period composite
6¥ietlh©ds and Sources
Industries
,
Output per employee hour
2?i K
Bls computes an index of output per employee hour
by dividing an output index by an index of aggregate
7U “
1Q[ Iq
102
employee hours. For most industries, measures are
prepared separately relating output to (a) all employee
hours, (b) production worker hours, and (c) nonproduc
tion worker hours. (The standard definitions of produc
tion workers and nonproduction workers are used.)
Three corresponding measures also are computed
relating output to the number of employees. For in
dustries in trade and services, measures are prepared
relating output to the hours of all persons involved in
producing that output, including self-employed and un
paid family workers.
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. This procedure is used almost
exclusively for the nongoods-producing industries. For
manufacturing and mining industries, quantity data on
physical output are usually most comprehensive for
years covered by a census. To make maximum use of the
comprehensive census data, output indexes are derived
from data for two consecutive censuses; these indexes
are referred to as benchmark indexes. For intercensal
years, annual indexes are based on either physical out
put data (generally in less detail than for census years)
or, if such data are not available, value of output ad
justed 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. The mathematical form of the output index
calls for the use of unit employee hour weights. Such
weights, used whenever possible, are derived from
special surveys or from data for specialized
establishments published in the Census o f Manufac
tures. In some industries, however, unit employee hour
information is not available for individual products.
Consequently, 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
substitute 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 hour1. 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
products 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 individual products are first aggregated into primary
product group indexes with unit value weights. These in
dexes 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 cen
sus data on production work hours and nonproduction
worker employment, are supplemented by unpublished
bls estimates of nonproduction worker hours. (See sec
tion on employee hours later in this chapter for the pro
cedures 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 specializ
ing in the product group. This procedure is used only
when the “ specialization” 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 1947-58, 1958 weights for 1959-63, 1963
weights for 1964-67, 1967 weights for 1968-72, 1972
weights for 1973-77, and 1977 weights for years after
1977. The Bureau revises the weights as more current
data become available from the periodic censuses.
Benchmark indexes. For most manufacturing, trade and
services, and all mining industries, indexes reflecting
changes in output between census years are constructed.
These are called benchmark indexes.
1 Irving H. Siegel, “ Further Notes on the Difference Between Index Number
Formulas,” Journal o f the American Statistical Association, December 1941,
pp. 519-24.
2 The “ specialization ratio” is the value o f shipments of primary products o f
plants in the industry expressed as a percent o f total shipments o f all products
(primary plus secondary) made by these same establishments. The “ coverage
ratio” is the value o f shipments o f the primary products made by plants
classified in the industry as a percent o f the total shipments o f the industry’s
primary products made by all producers, both in and out o f the specified
industry.
103
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 pro
duct within the class, whether made in the industry or
elsewhere. Producer price indexes are used wherever
possible to convert the product values to constant dollar
estimates. If a producer price index is not available, a
price index is developed 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 constant-dollar values.
These “ wherever made” primary product class price
indexes are used to deflate the value of primary prod
ucts produced only by the industry. This procedure
assumes that the price movements of the primary prod
ucts within the industry are the same as the price
movements for all primary products wherever made.
These constant-dollar values are related to correspond
ing base year values in 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 in
dustry 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 propor
tions 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 ad
justed, if necesary, to the levels of the benchmark in
dexes 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
104
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 quan
tities of products combined with unit value weights.
This index is derived by dividing the value of the in
dustry’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 in
dustry 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 ser
vices, 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. Depart
ment 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 Govern
ment sources include the the U.S. Department of
Energy, the U.S. Department of Agriculture, the Fish
3For example:
Value index -s- Price index (Paasche) = Output index (Laspr/res)
2p0 q{
2Pi<7j
X pQ q Q
2p0 q -
£p0 qQ
Where p^ and pQ represent prices of products in the industry in the
current 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 -s Price index (Laspeyres) = Output index (Paasche)
ZPj Qj
XPoVo
^ ^PjQ 0
ZPoVo
=
SPjgj
* P iq 0
and Wildlife Service, U.S. Department of the Interior,
the Interstate Commerce Commission, the Internal
Revenue Service, and the Civil Aeronautics Board. Im
portant sources of trade association data include the
Textile Economics Bureau, Inc., National Association
of Hosiery Manufacturers, 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. Employee hours are treated as
homogeneous and additive with no distinction made
between hours of different groups of employees. Data
on changes in qualitative aspects of employee hours,
such as skill, efficiency, health, experience, age, and sex
of persons comprising the aggregate are not used and
generally are not available. For mining and manufactur
ing industries, employee hour indexes are constructed
for employees, production workers, and nonproduction
workers. For service and trade industries, indexes are
constructed for the hours of all persons, which includes
paid employees, partners, proprietors, and unpaid
family workers.
Sources. Industry employment and employee hour in
dexes are developed from basic data compiled by the
Bureau of Labor Statistics or the Bureau of the Census.
For trade and service industries, these data are sup
plemented with data from the Internal Revenue Service.
For most private nonagricultural industries (including
manufacturing), b ls publishes employment and average
weekly hours data for production or nonsupervisory
workers and employment data for all employees. For
manufacturing industries, 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
definition of employee hours. Census data include all
hours at the plant, whether worked or paid for, and ex
clude paid time for vacations, holidays, or sick leave
when the employee is not at the plant. Overtime and
other premium pay hours are included on the basis of
actual time at the plant. In contrast, b ls data include
time for paid vacations, holidays, and sick leave, as well
as plant employee hours. Differences in the data from
the two sources for the same industry, however, also
stem from the difference in sampling and reporting
methods.
Generally, whenever employment and hours data are
available from both the Bureau of the Census and the
105
Bureau of Labor Statistics, the labor input data which
are used are those consistent with the data on output.
Thus, when output data from the Bureau of the Census
are used, employment and hours data from the same
source usually are preferred.
Nonproduction worker hours. While both the Bureau of
the Census and b ls provide data on production worker
employee hours, neither source provides annual data by
industry on nonproduction worker or all-employee
hours. Therefore, these measures are 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 nonproduc
tion worker.
For years prior to 1968, the estimates of average an
nual hours worked are calculated by multiplying the
number of workweeks in the year times the scheduled
weekly hours. This produces an estimate of average an
nual hours paid. Estimated hours for vacations,
holidays, disability, and personal time off are sub
tracted from average annual hours paid to obtain an
estimate for average annual hours worked.
Estimated hours for vacations, holidays, and
disabilities are based on data from various b l s surveys
and studies of the Department of Health and Human
Services. Personal time off has been estimated as a con
stant 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 be
ing 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 Mining 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 of Population.
All-employee hours estimates for manufacturing in
dustries are derived by summing the aggregate hours for
production workers and the estimated aggregate hours
for nonproduction workers. For trade and service in
dustries, all-person hours estimates are derived by sum
ming 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 industries other than in trade and services,
employee hours data are based on total employee hours
of establishments classified in an industry, whether the
employee hours are applied to production of primary or
secondary products. Annual physical output data, on
the other hand, usually include only primary products
of an industry and are usually 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
variation from year to year in the proportion of primary
products to total products of an industry, or if there is
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 specialization 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 opera
tions performed by establishments of an industry, while
output usually is measured in terms of the final product.
If establishments undertake additional operations (such
as the manufacture of components which had pre
viously 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 ex
amines 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.
F®d©ra! Government
Indexes of output per employee year, output, and em
ployee years for selected functional areas of Govern
ment activity4 and for the more than 400 participating
organizations are constructed in a manner similar to
that described for industries. At the present time, these
measures cover about 67 percent (1.9 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 organiza
tion may be consumed wholly or partially by another
4 The 28 functions are:
Audit o f 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; Per
sonnel investigations; 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.
Federal organization in the production of its final out
puts, all output indicators in the Federal sample may
not be final from the perspective of a higher level
organization; for example, the entire Federal Govern
ment. 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.
In the Federal sample, more than 3,000 products and
services are aggregated into output indexes by combin
ing the quantities of each type of output by their respec
tive base-year labor requirements. These unit employeeyear weights are constructed from the detailed output
and input data provided by each organization. For fiscal
years 1967-1972, fiscal year 1967 weights are used; for
years 1972-1977, fiscal year 1972 weights are used; for
years after 1977, fiscal year 1977 weights are used. The
three output segments are combined and referenced to a
fiscal year 1977 base.
The organizational indexes are grouped into 28 func
tional categories, based on type of Government activity.
Some of these categories, such as standard printing and
electric power production, are more homogeneous than
others, such as specialized manufacturing and informa
tion services. Nonetheless, these categories provide in
sight into the trends for the major functional areas
underlying the overall sample. 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 ex
ample, to stimulate further examination of the causes of
productivity change within each organization). This is
one method used by b l s to validate the basic data (that
is, by examining the reasonableness of the derived
trends).
Employee year indexes are developed from agency
data submissions. As in all labor input measures used by
the Bureau to develop productivity indexes, employee
years are considered homogeneous 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 40 hours a week, 52 weeks a
year.
Presentation
Bls industry indexes are published annually in the
bulletin, Productivity Measures fo r Selected Industries.
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 pro
ductivity, output, and employment trends in the in
dustry. Technical notes describing the methodology
106
of the work force, managerial skill, and labormanagement relations. Also, indexes which relate out
put to one group of employees represent the total output
of the industry resulting from all employees; they are
not representative of the specific contribution of that
group of employees.
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 consist
ency 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 subject 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
for nonproduction workers, however, have a relatively
insignificant effect on the estimates of hours for all
employees. 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.
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. Federal Government indexes are
published annually in the Monthly Labor Review.
Indexes of output per employee hour also are publish
ed in the Statistical Abstract o f the United States and in
the Handbook o f Labor Statistics. Some indexes for
earlier years are published in Historical Statistics o f the
United States.
Us®s and Limitations
Measures of output per employee hour are particular
ly useful for studying changes in labor utilization, pro
jecting 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 employ
ment are useful in studying technological effects; to
study trends in labor costs, data on earnings and other
labor expenditures are necessary.
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 fac
tor of production. 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
Technical References
Bureau ©f Labor Statistics
Mark, Jerome A. Measuring Productivity in Government—
Federal, State, and Local. A speech before the Con
ference on Productivity Research, American Produc
tivity Center. Houston, Texas, April 24, 1980.
Differentiates between measures o f efficiency, in
termediate work activity, and effectiveness; describes
concepts, methods, and problems relevant to the meas
urement of government productivity; stresses the need
for detailed product data and discusses examples of
current efforts in the collection of, and the improve
ment in, pertinent data and surveys problems o f con
cepts, methods, and data; adequacy in measuring pro
ductivity at the State and local government levels.
Ardolini, Charles, and Hohenstein, Jeffrey. “ Measuring Pro
ductivity in the Federal Government,” Monthly Labor
Review, November 1974.
Presents the results of b l s efforts of measuring pro
ductivity in the Federal sector and discusses concepts,
methods, trends, and measurement problems.
Mark, Jerome A. “ Industry Indexes of Output Per ManHour,” Monthly Labor Review, November 1962.
Describes the methods used in constructing b l s in
dexes of output per employee hour. Covers methods and
sources, construction o f production and employee hour
indexes, and limitations.
Mark, Jerome A. “ The Significance of Measures of Pro
ductivity.” A speech before the Conference on
Productivity and Work Quality. New York City, May
1975.
Discusses concepts of productivity and their inter
pretation, available measures, and recent trends in
productivity.
Mark, Jerome A ., and Stein, Herbert. Meaning and Meas
urement o f Productivity, b l s Bulletin 1714, 1971.
Prepared for the National Commission on Productivity.
Explains why productivity increase is important to
the economy, how it is measured, and why it is difficult
to measure.
Productivity: A Bibliography,
107
bls
Bulletin 1226, 1958;
T@©teii©fil References—Continued
Columbia University Press, 1973.
Presents trends in productivity by industry groupings
from 1948-66, with preliminary estimates through
1969. Long-term trends, patterns o f productivity
growth, and interrelations among variables are ana
lyzed. Includes descriptions o f concepts, methods, and
sources.
Bulletin 1514, 1966; b l s Bulletin 1976, 1973; and
Bulletin 2051, April 1980.
Collections of annotated references concerning pro
ductivity and productivity measurement.
bls
BLS
Other pyblieats®iiis
Denison, Edward F. Accounting fo r Slower Economic
Growth. The United States in the 1970’s. Washington,
The Brookings Institution, 1979.
An updated version o f Accounting fo r United States
Economic Growth, 1929-1969, The Brookings Institu
tion, 1974. Emphasizes factors likely to have retarded
more recent productivity growth, such as compara
tively unfavorable changes in labor force character
istics, a less favorable legal and “ human” environ
ment, and a decline in advances in knowledge.
Kendrick, John W. Understanding Productivity. An Intro
duction to the Dynamics o f Productivity Change.
Baltimore, The Johns Hopkins University Press, 1977.
Discusses concepts, meaning, and measurement of
productivity, as well as national and sectoral pro
ductivity trends, and the forces underlying them.
Presents international comparisons, and analyzes the
relation between productivity and costs and prices.
Kendrick,
Fabricant, Solomon. A Primer on Productivity. New York,
Random House, 1969.
Introduces basic ideas about productivity. Discusses
the sources of productivity and relates productivity to
business cycles, inflation, and economic policy. Also
discusses productivity abroad.
and
Vaccara,
Beatrice N .,
eds.
1980.
Collection o f papers on such subjects as labor
and multifactor productivity by industry; produc
tivity in selected service sectors; and international
comparisons o f productivity. Includes a study o f high
and low productivity establishments; current efforts
to measure productivity in the public sector; effects
o f research and development on industry productivity
growth; energy and pollution effects on productivity;
and international comparisons of economic growth.
National Bureau of Economic Research, Conference on
Research in Income and Wealth. Output, Input, and
Productivity Measurement. Studies in Income and
Wealth, Vol. 25. Princeton, Princeton University
Press, 1961.
Collection of papers on such topics as the design of
consistent output measures; employment and output
in the natural resource industries; the estimation of
real product and factor inputs; concepts o f real
capital stock.
Greenberg, Leon. A Practical Guide to Productivity Meas
urement. Washington, D .C ., Bureau of National
Affairs, 1973.
Discusses concepts o f productivity measurement with
emphasis on measurement at the company level.
Kendrick, John W. Postwar Productivity Trends in the
United States, 1948-1969. National Bureau of
Economic Research, General Series 98. New York,
W .,
Chicago, The University o f Chicago Press,
Fuchs, Victor R., ed. Production and Productivity in the
Service Industries. Studies in Income and Wealth, Vol.
34. New York, National Bureau of Economic Research,
1969.
Collection of essays dealing with conceptual and
measurement problems of output and productivity in
service industries, including medical care, commercial
banks, and retail trade. Also deals with service in
dustries in Canada and with the development of serv
ice industries in the 19th century.
International Labour Office. Measuring Labor Productivity.
Geneva, 1969.
Discusses methods and problems in the measurement
of productivity, analysis of national series, and inter
national comparisons.
John
New Developments in Productivity Measurement and
Analysis. Studies in Income and Wealth, Vol. 44.
National Academy o f Sciences. Measurement and Inter
pretation o f Productivity. Washington, 1979.
Collection o f papers on such topics as the concepts
and measurement o f productivity; the limitations of
productivity statistics; the measurement o f outputs
and inputs; the sources o f economic growth; measures
o f company productivity; and international compar
isons o f productivity.
108
Clhapt@ir 15= T@cSiiirii©l©gi©al
©hainig®
Baelkgroyndl
Studies of technological changes and their labor im
plications have been undertaken by b ls over the years
for a variety of purposes. During the 1930’s, public in
terest focused on the unemployed, and reports were
prepared on technological changes and displacement of
workers in various industries. During World War II,
emerging technologies were studied for purposes of im
proving manpower utilization.
Beginning in the mid-195Q’s, nationwide attention
was focused on the implications of new developments
classified under the general term “ automation.” BLS
made a series of studies on a plant basis, in the in
surance, petroleum refining, bakery, air lines, and elec
tronics industries, to explore the manpower implications
of various changes. Later, broader studies were under
taken, including a survey of the manpower impact of
changeover to electronic computers in 20 large com
panies 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 future
changes. Research now underway pinpoints
technologies which will become increasingly important
over the next decade in key industries and attempts to
provide advance information about their manpower im
plications.
Description of Studies
The Bureau’s research program on technological
change involves a variety of reports and studies of dif
ferent degrees of detail and approach. The current pro
gram thus provides: Summary reports surveying trends
in major industries; detailed industry studies; and
studies of major technological innovations, such as
computers, that affect workers in different industries.
Summary reports in 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
109
have an important effect over the next 5 to 10 years. The
industries covered comprise a cross-section of the
economy and include those where the pace may be slow
as well as those where change is rapid. The first version
of this report, entitled Technological Trends in 36 Ma
jor American Industries, was issued by the President’s
Advisory Committee on Labor-Management Policy in
1964. A revised edition covering 40 industries was
published in 1966. Bulletins containing more recent
reports for 30 industries were published in 1974-82.
The emphasis of the report is on technological
developments within each industry in an early stage of
the innovation’s commercial use; i.e., the period after
introduction on the market but before widespread adop
tion. 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
developments, indicating insofar as practicable some
economic advantages of various types of new equip
ment, processes, or products; their importance in terms
of the employee hours engaged in the operations af
fected; 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 advantages described include not on
ly labor savings per unit, but also quality improvements,
fuel and material economies, greater accuracy, new
markets, etc.
In assessing the employment implications of
technological 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 that are taking place.
Detailed industry studies
Intensive studies are made of selected major in
dustries where far-reaching changes, on a large scale,
are taking place, such as coal and printing. These
studies involve detailed analysis of the economic im
plications of major technological developments within
individual industries. Factors analyzed include invest
ment trends and factors affecting the prospects for the
diffusion of recent technological advances, such as the
structure of the industry. Estimates are developed of the
displacement of present by new methods over the next
10 years. Unit labor requirements under new and old
technologies are compared, wherever possible. Since the
focus of the study is on the industry as a whole, data on
recent industry trends in output per employee hour, pro
duction, and employment are examined in relation to
long-term trends, and projections of future trends are
developed.
Technological innovation 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 relations. In analyzing their impact in dif
ferent industries, differences as well as similarities are
revealed.
Data Sources and Collection Methods
A variety of data sources and collection methods are
utilized in making studies of technological change and
its impact.
Personal interviews
In making studies, analysts personally conduct inten
sive interviews with plant managers, personnel direc
tors, 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 infor
mal interviews 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 pro
duction.
Personal interviews also are utilized 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
judgment about the nature, pace of introduction, and
possible impact of developments with which few plants
have had any experience. The emphasis in these inter
views is on the technological change rather than on ex
periences in adjusting.
Trade and technical publications
Important sources of information concerning
technological trends are trade journals, technical
magazines and books, conference proceedings, govern
ment hearings, and company reports. Annual reports of
leading corporations and company house organs often
contain useful information 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; Bureau of In
dustrial Economics, U.S. Industrial Outlook (annual);
International Data Corporation, EDP 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 equipment; and the National Science
Foundation’s estimates of research and development.
Plant records
In making detailed studies of the impact of
technological 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 in
dividual 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 future 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 ade
110
quacy of projected trends. Over 450 persons were con
tacted in this way in the preparation of a report on
technological trends in major industries. Some experts
are visited personally to review draft statements in
detail. Reports on technological prospects are designed
to reflect, as much as possible, the authoritative 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 in
terpretation and analysis in this type of research are,
therefore, set forth briefly.
Definition of technological change
Technological change is defined broadly in the b l s
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
thought processes,” and “ a concern with production
processes as a system.”
While b l s studies have been concerned with
developments in automation, particularly in an
ticipating long-term trends, they are not the only
technological changes taking place that affect labor re
quirements and industrial relations. For example, new
ways of generating power, piggybacking in transporta
tion, use of synthetic materials in manufacturing,
mechanized methods of material handling, and faster
steelmaking processes are important technological
developments, not usually covered by technical defini
tions of automation, but having significant manpower
implications.
Impact on productivity
Since one of the principal consequences of
technological change, so far as manpower utilization is
concerned, is an increase in productivity (output per
employee hour), special attention is given in b l s studies
to analyzing changes in industrial productivity. Such
trend analysis is a useful method of measuring the pace
of technological change. Changes in productivity,
however, 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,
b l s studies try to indicate the reduction in unit labor re
Ill
quirements 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
technology 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 addi
tional labor in engineering and maintenance work. The
impact on plant productivity, therefore, would be con
siderably less than the effect on productivity of any
department or operation directly affected.
Impact on employment
In assessing the impact of technological change on
employment, it is necessary to consider the implications
of plant manpower 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 op
portunities in operations directly affected. If efforts are
made, however, to eliminate these jobs by not filling
vacancies created by quits, deaths, and retirement of
employees, or by transfer of affected workers to other
positions 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.
Since many changes exert their effects on employment
through the competitive market, 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 competing plants, which
may be forced to shut down, displacing workers far
from the location of the change.
Because of the whole complex of economic factors
that operate through the market, including changes in
demand, location, foreign competition, corporate
organization, and consumer taste, it is very difficult to
isolate the expanding and displacing 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 within the plant or office by transfer or pro
motion, the setting of wage rates, and selection of per
sons for new jobs. Often the adjustment proceeds
according to rules established in advance through collec
tive bargaining. Provisions to assist workers whose jobs
are eliminated include severance pay, retraining, and
early retirement. Besides analyzing the operation of for
mal provisions under collective bargaining, Bureau
studies describe informal efforts to provide training, to
utilize attrition, and to obtain jobs for displaced
workers elsewhere. The limitations of these measures as
well as their advantages are important matters studied.
Uses and limitations
Bls studies of technological change are prepared as
part of the U.S. Department of Labor’s program for
carrying out the objectives and responsibilities of the
Manpower Development and Training Act of 1962 and
the subsequent Comprehensive Employment and Train
ing Act ( c e t a ) of 1973. Under these acts, the Secretary
of Labor is required “ to establish techniques and
methods for detecting in advance the potential man
power impact of automation, technological progress,
and other changes in the structure of production.” As
part of such an early warning system, b l s studies and
reports of technological change are useful to managers,
union leaders, educators, economists, government of
ficials, and others in planning policies to cushion the im
pact of change. The study of emerging technological
trends and possible implications, moreover, provides a
basis for more valid projections of productivity and
economic growth. They also are useful in pinpointing
manpower problems and determining the most produc
tive direction of future research to obtain possible
solutions.
Some limitations of the Bureau’s studies of
technological 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 im
plications 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 pro
spects, 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, new developments are constantly ap
pearing so that assessments of the outlook must be reap
praised from time to time in the light of new informa
tion.
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 in
dustries and could, therefore, have important implica
tions for production and employment in industries sup
plying inputs. The accumulation of information on in
terindustry relationships, through the Bureau’s
economic growth studies, provides a quantitative basis
for analyzing this aspect of technological change.
Technical [References
Bureau of Labor Statistics
Computer Manpower Outlook,
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. Dis
cusses outlook for future and implications for produc
tion and productivity.
Bulletin 1826, 1974.
Presents information on the current employment
and education and training characteristics o f com
puter occupations; explores the impact of advancing
computer technology on computer manpower and
education; and projects computer occupational re
quirements and their implications for training.
bls
112
Technical References—Continued
Outlook fo r Technology and Manpower in Printing and
Publishing, b l s Bulletin 1774, 1973.
Describes changes in technology in the printing and
publishing industry and its impact on productivity,
employment, occupational requirements, and methods
of adjustment.
and Labor in Five Industries (bakery
products, concrete, air transportation, telephone com
munication, insurance), b l s Bulletin 2003, 1979.
Appraises major technological changes emerging in
five key industries and discusses their impact on pro
ductivity and occupations over the next 5 to 10 years.
R ailroad Technology and M anpower in the 1970’s,
Technology and Labor in Four Industries (meat prod
Bulletin 1717, 1972.
Describes changes in technology and their impact on
productivity, employment, occupational requirements,
and methods o f adjustment.
ucts, foundries, metalworking machinery, electrical
and electronic equipment), b l s Bulletin 2104, 1982.
Appraises major technological changes emerging in
four key industries and discusses their impact on
productivity and occupations over the next 5 to 10
years.
BLS
Technological Change and Its Labor Impact in Five
Energy Industries (coal mining, oil and gas extraction,
petroleum refining, petroleum pipeline transportation,
electric and gas utilities), b l s Bulletin 2005, 1979.
Appraises major technological changes emerging in
five key energy industries and discusses their impact
on productivity and occupations over the next 5 to 10
years.
Technology, Productivity, and Labor in the Bituminous
Coal Industry, 1950-79, b l s Bulletin 2072, 1981.
Appraises some o f the major structural and tech
nological changes in the bituminous coal industry and
their impact on labor in the industry.
The Revised Workweek: Results o f a Pilot Study o f
16 Firms, b l s Bulletin 1846, 1975.
Technological Change and Its Labor Impact in Five
Industries (apparel, footwear, motor vehicles, rail
Explores the impact of changes in the workweek
schedule to determine objectives and methods for in
troducing workweek changes and to assess the avail
ability of data for further research on the implications
for productivity and manpower.
road, retail trade), b l s Bulletin 1961, 1977.
Appraises major technological changes emerging in
five key industries and discusses their impact on pro
ductivity and occupations over the next 5 to 10 years.
Technological Change and Manpower Trends in Five
Industries (pulp and paper, hydraulic cement, steel,
Vickery, Mary L. “ New Technology in Laundry and
Dry Cleaning Services,” Monthly Labor Review,
February 1972. Reprint No. 2792.
Discusses the impact of technology on manpower
and productivity in laundry and dry cleaning esta
blishments. Discusses outlook to 1980.
aircraft and missiles, wholesale trade), b l s Bulletin
1856, 1975.
Appraises major technological changes emerging in
five key industries and discusses their impact on pro
ductivity and occupations over the next 5 to 10 years.
Technological Change and Manpower Trends in Six
Industries (textiles, lumber and wood products, tires
Z eisel, Rose N . “ M odernization and M anpower in
Textile Mills,” Monthly Labor Review, June 1973.
Reprint No. 2893.
Analyzes the impact of technology on manpower in
textile mills. Includes data on capital expenditures
and research and development.
and tubes, aluminum, banking, health services), b l s
Bulletin 1817, 1974.
Appraises major technological changes emerging in
six key industries and discusses their impact on pro
ductivity and occupations over the next 5 to 10 years.
Technology
113
Chapter 18. Foreign Labor
S ta tis tic s and Trade
M onitoring
F©reigin Labor Statistics
Background
From its inception, the Bureau has collected and
published statistical information on labor conditions
and developments abroad. Foreign labor research and
statistical analyses have been undertaken because (1) in
formation 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 ©f ileasures
The BLS foreign labor statistical reports cover a varie
ty of international comparative measures, mainly for
the Western industrial countries. The principal
measures cover the labor force, employment, and
unemployment; productivity and labor costs; hourly
compensation of manufacturing production workers;
and trends in consumer prices.
Labor force, employment, and unemployment. Com
parative measures of the labor force, employment, and
unemployment 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.
concepts, are prepared monthly for most of the coun
tries; the other measures are calculated annually.
114
Productivity and labor costs. Comparative trends in
manufacturing productivity (output per hour), hourly
compensation, unit labor costs (labor compensation per
unit of output), and related measures are compiled on
an annual-average basis for the United States, Canada,
Japan, Belgium, Denmark, France, Germany, Italy, the
Netherlands, Sweden, and the United Kingdom. Trends
are expressed in index form (1977 = 100) and as percen
tage changes at annual rates. For most countries, the
series begin with 1950. Indexes of unit labor costs for
foreign countries are calculated in national currency and
in U.S. dollars converted at prevailing commercial ex
change rates.
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 (1977 = 100) and at annual rates of
change.
Comparative levels (United States = 100) and trends
(1977 = 100) in gross domestic product (GDP), g d p per
capita, and g d p per employed person are calculated on
an average-annual basis for the United States, Canada,
Japan, Belgium, France, Germany, Italy, the
Netherlands, and the United Kingdom beginning with
1950. The g d p level comparisons, which are based on
estimated purchasing-power-parity exchange rates, are
benchmarked to data from the United Nations Interna
tional 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 market exchange rates. However, market ex
change rates seldom reflect the relative purchasing
power of different currencies.
Hourly compensation. Measures of total compensation
per hour worked for production workers in all manufac
turing and in over 30 selected manufacturing industries
are prepared annually for about 30 countries. The
measures are developed from data on average earnings,
as published by each country, plus information on other
direct payments to the worker and employer expen
ditures for legally required insurance programs and con
tractual and private benefit plans. They are expressed in
national currency and in U.S. dollars at prevailing com
mercial exchange rates. Hourly compensation, when
converted to U.S. dollars at commercial exchange rates,
indicates comparative levels of employer labor costs. It
does not indicate relative living standards of workers or
the purchasing power of their income. Prices of goods
and services vary greatly among countries and commer
cial exchange rates are not reliable indicators of relative
differences in prices.
Estimating Procedures
Because statistical concepts and methods vary from
country to country, international comparisons of
statistical data can be misleading. The Bureau attempts
to derive meaningful comparisons by selecting a concep
tual 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 intercountry comparability.
Data Soyrees
Labor force, employment, and unemployment. For
labor force and unemployment comparisons, the
Bureau adjusts each country’s published data, if
necessary, to provide measures approximately consis
tent with U.S. definitions and standards. The labor
force and unemployment statistics for 6 of the 10 coun
tries regularly studied—the United States, Canada,
Australia, Japan, Italy, and Sweden—are obtained
from monthly or quarterly household surveys. No ad
justments are made to the published data for Canada
and Australia, since their concepts and methods are vir
tually identical to those in the United States. Slight ad
justments 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
(France, biannually; Germany and the United
Kingdom, annually; the Netherlands, biennially) which
contain benchmark data that are used to adjust the level
of the labor force and total unemployment for greater
comparability with U.S. concepts. Measures of current
labor force and unemployment are obtained by applying
adjustment factors from the most recent year’s labor
force surveys to published labor force and unemploy
ment registration figures.
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 Na
tions, International Labour Office, Organization for
Economic Cooperation and Development, and the
European Community; 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.
Productivity and labor costs. Indexes of manufacturing
productivity, hourly compensation, and unit labor costs
are constructed from three basic aggregative measures;
i.e., 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. However,
the Italian figures for 1970 forward relate to mining and
manufacturing less energy-related products.
The long-term output measures are gross product
originating in manufacturing (value added) in constant
Consumer prices. Indexes for consumer prices are com
piled 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 11 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 14 coun
tries; selected producer price indexes for 8 countries; im
plicit price deflators for gdp and consumer expenditures
for 9 to 11 countries; and ratios of capital investment,
excluding residential construction, to output for 11
countries.
115
prices from the national accounts of each country—ex
cept those for Japan prior to 1970 and the Netherlands
for 1969 forward, which are indexes of industrial pro
duction. Methods of deriving national accounts
measures of manufacturing output differ substantially
from country to country, however, and the Canadian
and British national accounts measures are identical to
their indexes of industrial production. For current
measures, indexes of industrial production are used un
til 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 of
ficial series from their statistical agencies. For the other
countries, 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. For
Belgium, France, and the Netherlands, the hoursworked measures may not reflect all random hours
changes, such as time lost because of industrial disputes.
The compensation measures are from national ac
counts—except those for France, from 1967 forward,
and Belgium, which are developed by the Bureau using
statistics of employment, average hours, and hourly
compensation. 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 current indicators of manufac
turing employment, average hours, and hourly compen
sation until national accounts and other statistics used
for the long-term measures become available.
The Bureau’s 1964 and 1972 measures of comparative
productivity and labor costs in the iron and steel in
dustry, with the exception of the exclusion of wire pro
ducts 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 pro
ducts), and the labor input data have been carefully
matched with the output figures. Measures for years
subsequent to the latest benchmark are obtained by ap
plying trend indexes to the benchmark measures. Except
for the United States, the trend indexes are based on dif
ferent 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 comparative
levels of g d p extrapolated or interpolated to other
years, and on annual population and employment
estimates. The GDP level comparisons are based on
estimated purchasing-power-parity exchange rates. The
employment figures for some countries have been ad
justed for greater comparability with U.S. concepts.
The benchmark (1970 and 1973) level comparisons of
GDP for Japan, Belgium, France, Germany, Italy, the
Netherlands, and the United Kingdom are from Irving
B. Kravis, Alan Heston, and Robert Summers, Interna
tional Comparisons o f Real Product and Purchasing
Power (United Nations International Comparison Pro
ject: Phase II). The benchmark (1965) level comparisons
for Canada were derived from E.C. West, “ Real Out
put Comparison, Canada and the United States,” ap
pendix to Dorothy Walters, Canadian Income Levels
and Growth, An International Perspective (Ottawa,
Staff Study No. 23, prepared for the Economic Council
of Canada, 1968). The benchmark figures were derived
by combining output quantities, at detailed levels of ex
penditure, according to a common set of price weights.
Because of differences in price structures, however, no
single set of price weights is ideal for combining the out
puts of different countries. Therefore, the GDP com
parisons are shown using “ international” price weights
(except for Canada), U.S. price weights, own (foreign
country) price weights, and the geometric mean of U.S.
and own price weights. Output quantities at the detailed
levels of expenditure were obtained by converting na
tional accounts expenditures data into a common cur
rency (U.S. dollars) using purchasing-power-parity ex
change rates for each detailed category. The base-year
comparisons of real g d p are extrapolated or inter
polated to other years using relative changes in g d p at
constant market prices, as measured by each country.
Hourly compensation. Measures of hourly compensa
tion for production workers in all manufacturing and in
selected manufacturing industries are prepared because
hourly compensation provides a better basis for interna
tional comparisons of labor costs than the earnings
statistics which are regularly published by most coun
tries. 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 expenditures for legally required in
surance programs and contractual and private plans for
the benefit of employees. In addition, taxes on payrolls
or employment are included even if they are not for the
direct benefit of employees, because such taxes are
116
regarded as labor costs. However, some items of labor
cost—the costs of recruitment, training, and plant
facilities and services (e.g., cafeterias, medical clinics,
and employee parking)—are excluded because data are
not available for all countries. For consistency, compen
sation is measured on an hours-worked basis for every
country.
The total compensation measures are derived by ad
justing 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.
For the United States and other countries that measure
earnings on an hours-paid basis, the figures are also ad
justed in order to approximate compensation per hour
worked. Adjustment factors are obtained primarily
from periodic labor cost surveys and interpolated or
projected to nonsurvey years on the basis of other
available information, or they are obtained from cen
suses 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, sam
ple benchmarks, or frequency of surveys. Compensa
tion is converted to U.S. dollars using average daily ex
change 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 and the national accounts implicit price
deflator for consumer expenditures. Definitions of
worker stoppages usually refer to strikes and lockouts,
but the exact definition differs from country to country.
The statistics are not adjusted for comparability. No ad
justments are made to country producer price indexes or
implicit price deflators for gdp and consumer expen
ditures except to link indexes published on different
base years and to convert them to a common index base.
Ratios of fixed capital investment to output include
government, exclusive of government outlays for
military use, as well as private investment. They are ad
justed for definitional consistency where possible; e.g.,
by the exclusion of breeding stocks, which are classified
as inventories in the United States.
117
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 change 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. Also, an an
nual news release is issued on comparative trends in
manufacturing productivity and labor costs. The b l s
Handbook o f Labor Statistics and the Bureau of the
Census’ Statistical Abstract o f the United States publish
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 position
of the United States in international trade; (d) to review
foreign experience for possible application domestical
ly; and (e) to provide labor statistics and related infor
mation to individuals, corporations, labor unions, and
others concerned with foreign investment and develop
ment.
Although considerable progress has been made in
making international economic statistics more uniform
among countries, e.g., through the work of interna
tional agencies such as the United Nations and the Inter
national Labour Office, international statistical com
parisons should be used cautiously. Nevertheless,
through careful analysis of each country’s data, valid
statistical comparisons can be made. Whenever possi
ble, BLS adjusts foreign data, if necessary, for greater
consistency with U.S. measures; in some cases, data are
sufficiently similar in definition and concept for valid
comparisions without adjustment. Moreover, when
conceptual differences are substantial, the Bureau at
tempts to describe the differences in sufficient detail to
provide guidance in the interpretation of the data.
Trade fi@&iit©ring
Background
Section 282 of the Trade Act of 1974 provides that the
Secretary of Commerce and the Secretary of Labor shall
establish and maintain a trade monitoring system,
which will reflect (a) changes in the volume of imports
into the United States; (b) the relation of such imports
to changes in domestic production; (c) changes in
employment within domestic industries producing ar
ticles like or directly competitive with such imports; and
(d) the extent to which such changes in production and
employment are concentrated in specific geographic
regions of the United States. The primary responsibility
for the trade monitoring program has been delegated to
the b l s .
Description of Measures
The trade monitoring program is intended to measure
changes in U.S. imports, output, and employment in
sufficient detail to permit analysis of the relationships
between trade and employment. The principal measures
developed thus far by the Bureau cover U.S. merchan
dise trade, comparable domestic production, and
related industry employment.
Merchandise imports. Quarterly measures (with annual
totals) of the customs value and quantity of U.S. im
ports for consumption are prepared regularly. Quarterly
series for all commodities begin with 1975; annual series
begin with 1972. Value and quantity data are available
on a Tariff Schedule o f the United States Annotated
(TSUSA) basis; only value data are available on a Stan
dard Industrial Classification (SIC) basis. At present,
country-of-origin, duty, and other information are not
included. Data are not seasonally adjusted; value data
represent current value measures.
Industry employment. Quarterly measures of national
mining and manufacturing industry employment are
prepared from the Bureau’s monthly establishment
survey on employment and earnings, beginning with
1972. State and area employment tabulations are
prepared from the same survey for selected manufactur
ing industries.
Data Sources
The principal data sources for trade monitoring are
(a) for imports, the Bureau of the Census’ monthly data
tape on U.S. imports (IM-145), supplemented by annual
concordance tapes which relate import data to the sic
classification system; (b) for exports, annual issues of
the Bureau of the Census’ EA-675, U.S. Exports o f
Domestic and Foreign Merchandise, s i c Division by SICbased 2-digit, 3-digit, and 4-digit Product Code; (c) for
domestic output, the quinquennial industrial censuses
and annual surveys of manufacturing conducted by the
Bureau of the Census; and (d) for employment, the b l s
monthly establishment survey of employment and earn
ings.
Estimating Procedures
Several adjustments are made in the trade and
domestic data used in the trade monitoring system
because U.S. imports, U.S. exports, and domestic pro
duction and employment are reported on different
classification bases. U.S. imports are reported on the
basis of (over 10,000) legal tariff commodity classifica
tions designed for the collection of duties in the Tariff
Schedules o f the United States Annotated. U.S. exports
are reported on the basis of the U.S. Department of
Commerce’s (over 4,000) Statistical Classification o f
U.S. Exports (Schedule B). Domestic production and
employment are reported on a classification of domestic
Merchandise trade as related to output. Annual valuebased measures of U.S. merchandise imports as a per
cent of new supply (domestic shipments plus imports)
and U.S. merchandise exports as a percent of domestic
shipments are calculated each year for over 300
manufacturing sic-based product groups, beginning
with 1972.
118
economic activity of establishments based upon the
Standard Industrial Classification Manual, 1972 edition
and its 1977 Supplement. Furthermore, while manufac
turers’ shipments (used as a proxy for output) are
available on both an industry basis—total shipments of
firms classified in a given industry, including shipments
of secondary products—and on a product-class
basis—total shipments of the primary products of an in
dustry, including shipments of the same products made
by firms classified in other industries—employment
data are only available on an industry basis. Since inter
national trade classifications are commodity-based, im
ports and exports have been matched with domestic
product-class shipments on a wherever-made basis.
For the purposes of relating imports to domestic out
put and employment, imports classified by t s u s a
number are assigned by the Bureau of the Census to
5-digit sic-based product classes. An sic-based product
class is a group of individual products primary to a
4-digit sic industry and is designated by a 5-digit code,
the first four digits indicating the sic industry and the
fifth digit, the specific group of products. In cases
where the t s u s a numbers include items which should be
classified in two or more sic-based product classes, an
assignment is made to the sic-based product class to
which the principal content of the t s u s a number
belongs, if this will not result in signficant classification
distortions. In cases where distortions will result from
such an assignment, the principal sic-based product
classes are combined to form an sic-based import code
and the TSUSA number assigned to the combination. For
the 1972 edition of the sic manual, and its 1977 supple
ment, there are 555 (452 in manufacturing) 4-digit sicbased industry codes and 414 (347 in manufacturing)
4-digit sic based import codes. Of the 414 sic-based im
port codes, 266 coincide entirely with a corresponding
4-digit sic-based industry code and 31 represent com
binations of 72 4-digit sic-based industry codes; the re
maining 117 correspond to only a part of an sic-based
industry code, to an sic-based industry code plus parts
of other sic-based industry codes, or are residual import
categories that are not comparable to domestically pro
duced products. Of the 555 sic-based industry codes, 41
are not covered by import classifications.
The current sic-based import code assignment for
each t s u s a number is obtained from an annual
Bureau of the Census import edit master tape. Since
these assignments may change over time, b l s maintains
a master file of current and past assignments and ad
justs for historical consistency, sic-based export data
are obtained directly from the Bureau of the Census and
are not adjusted except for historical consistency.
Some adjustments of output (shipments) data are also
made for the purpose of relating domestic output to the
sic-based import and export groups. The value of
manufactures product shipments at the 4-digit industry
119
level often includes a small amount which is not
distributed among the individual 5-digit product classes
(manufactures shipments not specified by kind). When
sic-based import (export) codes represent combina
tions of product classes from different 4-digit industry
groups, a share of the undistributed output is allocated
to each 4-digit sic-based import (export) code accor
ding to the 5-digit product class share of the total. Since
this allocation is an approximation, the value of
shipments for the 4-digit sic-based import (export)
codes might be slightly over or under stated.
Analysis and Presentation
Analyses of U.S. imports and related domestic pro
duction and employment focus both on long-term and
short-term changes. Merchandise imports are
monitored quarterly for significant increases in current
dollar value over the year-ago quarter and the prior 12
months, and industry employment is monitored
quarterly to note significant declines over the same time
spans. In addition, measures of import penetration,
which are updated annually, with about a 2-year lag, are
monitored for substantial long-term increases or sus
tained high levels.
Since the trade monitoring program is still in the early
stages of development, the form of presentation thus far
has been confined almost entirely to statistical data
tabulations and summary rankings. However, several
papers have been prepared describing the measures, and
staff papers have been developed on the concordance
problems.
Uses arid Limitations
The purpose of the Bureau’s trade monitoring system
is to provide information, for policy analysts and those
engaged in the administration of trade adjustment
assistance programs, that might help in identifying
domestic industries which possibly are experiencing
adverse employment effects because of changing inter
national trade patterns. However, the Bureau does not
make estimates of jobs potentially lost as a result of ris
ing imports—because of the limitations associated with
such estimates.
There are several conceptual and measurement dif
ficulties in comparing trade, output, and employment
data. These problems vary by the product or industry
considered and the scope of the measure. Perhaps the
most critical problem is the incongruity among
classifications used for reporting imports, exports, and
domestic output and employment. In addition to the
differences in the basic classification systems, there are
other problems affecting comparability between trade
and domestic economic data. Differences in methods of
valuation present the principal such problem in com
paring imports with domestic output. Aggregation of
5-digit product-class shipments to a 4-digit sic level will
result in duplication to the extent that these com
modities are used as materials in other commodities pro
duced within the industry considered, whereas there are
no similar problems for imports since only final pro
ducts are recorded; low-value transactions are excluded
from the data for individual import commodity
classifications; and a small portion of manufacturing
shipments that are not allocated to detailed commodity
lines has been distributed over constituent product
classes. All these factors affect comparability to some
degree. For this reason, the measures of import penetra
tion should be considered only as approximations. Fur
thermore, since the output data used for comparison
with the trade data are product-based, they are not
directly comparable to industry-based measures such as
employment.
Technical References
Foreign labor statistics
U .S.
Labor force, employment, and unemployment
Labour O ffice. International Recom
mendations on Labour Statistics. Geneva, 1976.
International
Presents recom m endations standardizing labor
statistics, including recommendations on employment
and unem ploym ent statistics and statistics o f
labor costs.
Jusenuis,
Carol
L .,
and
von
Rabenau,
Burkhard.
Unemployment Statistics in the United States and the
Republic o f Germany: Problems o f International
Comparisons. National Commission on Employment and
o f Labor S tatistics. Youth Unemployment:
An International Perspective, Bulletin 2098, Septem
Bureau
ber 1981. A lso, Sorrentino, Constance. “ Youth
U nem ploym ent: A n In tern ational P er sp e ctiv e ,”
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.
Unemployment Statistics, Background Paper No. 30,
April 1979.
R eview s b l s p roced u re for a d ju stin g the
unemployment rate of the Federal Republic of Ger
many to U.S. definitions and examines the extent to
which each country has slack labor in categories
other than the unemployed.
Productivity and labor costs
C.P. The Measurement o f Real Product.
Hill,
M oy,
Jo y a n n a ,
and
S o rre n tin o ,
C o n sta n ce .
“ Unemployment, Labor Force Trends, and Layoff
Practices in 10 Countries,” Monthly Labor Review,
December 1981.
P resident’s Comm ittee to Appraise Em ployment and
Unemployment Statistics. “ Comparative Levels of
Unemployment in Industrial Countries,” in Measuring
Employment and Unemployment, Appendix A, 1962.
Also, Myers, Robert J., and Chandler, John H. “ Inter
national Comparisons of Unemployment” and “ Toward
Explaining International U nem ploym ent R a tes,”
Monthly Labor Review, August 1962 and September
1962.
The Bureau’s original study comparing unemploy
ment rates in eight countries.
Paris,
O rganization for E conom ic C o-op eration and
Development, February 1971.
A theoretical and empirical analysis of the growth
rates for different industries and countries.
Kravis, Irving B. “ A Survey of International Com
parisons 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.
Kravis, Irving B.; H esto n , A lan; and Sum m ers,
Robert. International Comparisons o f Real Product
and Purchasing Power, United Nations International
Comparison Project. Phase II, produced by the Statistical
Office of the United Nations and the World Bank and
published for the World Bank by the Johns Hopkins
University Press, Baltimore and London, 1978.
Phase II of the United Nations International Com
parison Project to provide internationally comparable
data on real product and the purchasing power of curren
cies. Provides the benchmark gross domestic product
(GDP) comparisons used in the Bureau’s comparisons of
real g d p per capita and per employed person.
S o rre n tin o , C o n sta n ce . “ C om p aring E m p loy m en t
Shifts in Ten Industrialized Countries,” Monthly
Labor Review, October 1971.
S orrentin o, C onstan ce. “ U nem ploym ent C om p en sa
tion in Eight Industrial Nations,” Monthly Labor
Review, July 1976.
Department of Labor, Bureau of Labor Statis
tics. International Comparisions o f Unemployment,
Bulletin 1979, August 1978.
Provides the conceptual framework and a compre
hensive description of the Bureau’s work on interna
tional unem ploym ent comparisons, describes in
detail the methods o f adjusting foreign unemploy
ment rates to U.S. concepts, and analyzes various
factors contributing to differences in unemployment
levels.
120
Technical References—Continued
Kravis, Irving B.; Heston, Alan; Summers, Robert. “ New
Insights into the Structure of the World Economy,” The
Review o f Income and Wealth, Series 27, No. 4,
December 1981.
Presents summary results for the 34 countries covered
by Phase III of the United Nations International
Comparison Project.
Provides the conceptual framework and 1964 results
of the Bureau’s comparisons of absolute levels o f pro
ductivity and labor costs in a major industry.
Bureau o f Labor Statistics. Comparative Growth in Manu
facturing Productivity and Labor Costs in Selected Indus
trialized Countries, Bulletin 1958, 1977.
Describes trends and analyzes the effect of relative
shifts in industry shares o f output and input on pro
ductivity trends for total manufacturing. Sources and
methods for the Bureau’s international comparisons o f
manufacturing productivity and labor costs are sum
marized in a technical appendix.
Nelson, Richard R. “ Research on Productivity Growth and
P rod uctivity D ifferences: Dead Ends and New
Departures,” 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:
Concepts and Methods,” Monthly Labor Review, May
1963.
Summers, Robert; Kravis, Irving B.; and Heston, Alan.
“ International Comparison of Real Product and its Com
position: 1950-77,” The Review o f Income and Wealth,
Series 26, No. 1, March 1980.
Develops comparisons o f real gross domestic product
(GDP) and g d p per capita for over 100 countries on the
basis of structural relationships estimated from data for
the 16 countries covered in Phase II of the United Na
tions International Comparison Project.
Bureau of Labor Statistics. An International Comparison o f
Unit Labor Cost in the Iron and Steel Industry, 1964:
United States, France, Germany, United Kingdom,
Bulletin 1580, 1968.
Trade monitoring
Biles, Elmer S.; Chandler, John H.; Mark, Jerome A.;
and Schoepfle, Gregory K. “ Impact o f the Trade Act of
1974 on Industrial and Foreign Trade Statistics,” Pro
ceedings o f the Business and Economic Statistics Section,
American Statistical Association, 1978.
Schoepfle, Gregory K. “ Imports and Domestic Employment;
Identifying Affected Industries,” Monthly Labor Review,
August 1982.
Considers the problems of constructing indicators of
import market share at the industrial (4-digit sic) level,
presents some leading results relating imports to new
supply, and discusses the usefulness and limitations of the
measures.
11
2
Chapter 1?„ ©eeypstGonal Safety
and H ealth S ta tistie s
Part L Amual Bmmy ®f Oeeupati®oal
Itmjyries and Illnesses
Background
January 1975, involved the classification of lost
workdays as either days away from work or days of
restricted work activity. The second change, in January
1978, represented an effort to reduce the burden on the
employer and to streamline o s h a recordkeeping and the
reporting system. A new recordkeeping form—o s h a
No. 200, Log and Summary of Occupational Injuries
and Illnesses—made it easier for employers, employees,
and safety and health officers to identify the major in
jury and illness problems. (An example of o s h a 200 is
included at the end of the chapter.)
The cases which must be recorded include all workrelated deaths, illnesses, and those injuries which result
in: Loss of consciousness, restriction of work or mo
tion, 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 oneline entry on the form. A case is recorded as a lost work
day 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 the duties of
his or her regular job. 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 temporary basis, works at a permanent
job less than full time, or works at a permanently
assigned job but cannot perform 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 necessary information. A copy of the summary totals
of the injuries and illnesses must be posted at each
establishment where notices to employees are posted no
later than February 1 and remain in place until March 1.
Virtually all employers are covered by the act.
However, to ease the recordkeeping burden on em
ployers, Federal regulations exempt groups of
employers from mandatory keeping of o s h a records
of occupational injuries and illnesses. This exemption
The Bureau of Labor Statistics has long been in
terested in statistics on safety and health conditions for
workers on the job and issued its first report on work in
juries as early as 1893. Subsequent b l s publications
reflected a growing concern for the worker disabled on
the job and were helpful in the development of the pre
sent 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 developing and maintaining an effec
tive program of collection, compilation, analysis and
publication of occupational safety and health
statistics.” The Secretary further directed the Commis
sioner to coordinate the above functions with the Assis
tant 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
operations, to provide b l s and State agencies with
uniform and reliable safety and health statistics, to pro
vide employers and employees with information about
conditions 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 manage
ment and employees pinpoint problem areas.
R@(S©rdlc®®piog and Reporting
Reqyirements
Two major changes in the recordkeeping system have
taken place since it began in July 1971. The first, in
122
does not affect the obligation of employers to observe
all safety and health standards, to report within 48
hours any accident which results in one or more deaths
or the hospitalization of five or more employees, and to
participate in the annual survey when notified of their
selection for the survey sample. Farm employers with 10
or fewer employees are totally exempt from any o s h a
regulation or activity involving Federal funds and are
omitted from the survey sample.
Concepts
Definitions used in the annual survey are the same as
those used in the o s h a recordkeeping system. Reports
for all injuries and illnesses occurring during the year in
clude information on the number of fatalities, injuries
and illnesses with workdays lost, the number of
workdays 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
varying sizes. Therefore, incidence rates are produced
for each type of case reported under o s h a definitions.
Incidence rates express various measures of injuries and
illnesses in terms of a constant, i.e., exposure hours in
the work environment (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 estab
lishments. 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 sec
tion.)
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
establishments in the same industry but in different
geographic areas. Further comparisons are possible
using the different types of rates computed for each in
dustry—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.
S<s®p® ®f th@ Surwey
The survey sample selected by b l s consists of
approximately 280,000 units in private industry. Survey
123
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. Department of Labor and
the Federal Railroad Administration of the U.S.
Department of Transportation. The Occupational Safe
ty and Health Administration collects and compiles
comparable data for Federal agencies. Although State
and local government agencies are not surveyed for na
tional estimates, several States have legislation which
enables them to collect these data. Self-employed per
sons 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 oc
cupational injuries and illnesses and to provide the data
from which b l s produces national results. National
data for selected States which do not have operational
grants are collected directly by b l s and by the State
agencies under contract. The participating State agen
cies collect and process the data and prepare estimates
using standardized procedures established by b l s to in
sure uniformity 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 re
sults and provides technical assistance to the State agen
cies on a continuing basis.
Data ColSection
State agencies mail report forms to selected employers
in February to cover the previous calendar year’s ex
perience. For those States not participating in the pro
gram, 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.
Information for the injury and illness portion of the
report form is copied directly from the Log and Sum
mary of Occupational Injuries and Illnesses. The form
also contains 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 incon
sistencies through phone calls, correspondence, or
visits. The data are keypunched and mechanically
edited. Reports which do not meet the computer screen
ing 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 in
juries and illnesses begins.
(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 sic’s.
Estimating Procedures
W eighting
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:
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 In
dustrial Classification (sic) code and employment.
Since these characteristics are highly correlated with an
establishment’s number and rate of recorded injuries
and illnesses, stratified sampling provides greater preci
sion 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 certainty strata are usually
the size groups with more than 100 employees. The
precision of the sample is further improved, hence per
mitting a reduction in sample size, by usfng the ratio
estimator which utilizes available auxiliary information
Xj = S WyXy
j=l
where:
X ; = weighted estimate of characteristics, e.g.,
number of cases reported, in size class i
Wjj = weight of sample unit (establishment) j in
size class i, adjusted for nonresponse
Xy = 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 cur
rent employment 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 benchmarking procedure re
quires 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, benchmark factors are calculated
using both industry level employment data and size class
level employment data. The benchmark factors are ap
plied to the size class “ blow up” estimates.
124
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:
quality assurance program is conducted periodically to
evaluate the extent of nonsampling errors in the
estimates. A sample of the participating establishments
is visited by survey personnel. The entries on the log and
summary are compared with supplementary records
( o s h A No. 101) and other available information to
evaluate the reliability of the log entries which provide
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 a series of industry guides. The in
dustry guides provide an explanation of how to compute
an incidence rate for a firm and how to compare this
rate with the national rate for each industry and
employment-size group. The data also 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.
Department of Commerce, Springfield, Va. 22151.
(Sum of characteristic reported) x 200,000
Incidence rate = -----------------------------------------------------------Sum of number o f hours worked
Incidence rates for higher levels of industry detail are
produced using aggregated weighted and benchmarked
totals. Rates may be computed by industry, employ
ment 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 sam
ple, not on the entire population. Estimates based on
the different possible samples of the same size and sam
ple 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 estima
tion process. For the all industry 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 sampling 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 In
juries and Illnesses in the United States by Industry.
Nonsampling errors in the estimates can be
attributed to many sources; e.g., inability to obtain in
formation 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
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 in
surance carriers involved in workers’ compensation, in
dustrial hygienists, manufacturers of safety equipment,
researchers, and others concerned with job safety and
health.
In terms of the recording and reporting of occupa
tional illnesses, the statistics generated through the an
nual survey 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 ex
cluded, 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 occupational link apply to workers’
compensation cases.
125
Fart II. Suppl@mantary Data System
surance carriers submit to State workers’ compensation
agencies. All jurisdictions, with the exception of
Louisiana, 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 additional burden on employers.
The prescribed data elements which must be uni
formly defined and submitted by all participating States
are:
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
comparability. The s d s 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 in
formation through contracts with States.
Records routinely generated by State workers’ com
pensation programs—employee and employer reports,
medical reports, compensation award records, etc.
—were long recognized as potentially valuable sources
of information about occupational injuries and ill
nesses. However, most workers’ compensation agencies
were primarily concerned with administering claims
systems, and were not particularly concerned with
availability and accuracy of industry, occupation, or in
jury and illness data. Additionally, States processing
such data had different coding systems, sometimes with
identical terms being defined differently.
States were urged to supply the desired information in
machine-readable form. However, the different classi
fication systems and record formats resulted in noncom
parabilities and processing difficulties. The Bureau
revised the program to require participating States to
use comparable 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 s d s does
not affect the variations in coverage and reporting re
quirements among States, it requires that participating
States provide prescribed data elements, and use specific
classification systems, standard record formats, and
uniform procedures.
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
elements, such as length of service, extent of disability,
indemnity compensation, and medical costs, some of
which may be defined differently from State to State.
For example, “ length of service” may refer to time with
an employer, in a particular occupation, or in a par
ticular job. The following optional items as of 1980 may
be submitted by participating States. (The number in
parentheses indicates the number of States providing
that information.)
D<as©fipts©n of SDS
Day of occurrence (31)
Time of accident (15)
Time workday began (7)
Lapsed time (4)
The primary source of information for the SDS is a
first report of injury or illness, which employers and in
126
Associated object or substance (19)
Age of employee (34)
Length of service (26)
Weekly wages (25)
Extent of disability (18)
Kind of insurance (13)
Indemnity compensation (13)
Medical payments (6)
Rehabilitation costs (2)
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
disabilities, 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 pay
ments data are useful economic and social indi
cators, 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 s d s in
clude: 1) The 1972 Standard Industrial Classification
Manual to code industry; 2) the 1970 Bureau of the Cen
sus Alphabetical Index of Industries and Occupations to
code the occupation of the injured or ill employee; 3)
the American National Standards Institute Z16.2—1962
Method of Recording Basic Facts Relating to the Nature
and Occurrence of 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 ex
posure; and 4) a newly developed classification, the
associated object or substance, which provides addi
tional information about the factors associated with the
injury or illness.
The SDS 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 im
proving 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 s d s Inter
pretations Committee, which resolves differences in
coding difficult cases, and the State task force on
coding revisions, which is composed of nine State
members and reviews classification systems and coding
practices with the view to changing current procedures
when necessary.
127
The national office screens and edits all State SDS data
tapes. Standard tabulations produced for each State are
also reviewed for further validation of the data.
By applying the percent distributions of the s d s data
to annual survey data, it is possible to arrive at
estimated numbers of injuries and illnesses on a national
basis by various characteristics. Despite differences in
the number of cases among States, the percentage
distributipns of injuries and illnesses are relatively con
sistent across the States—patterns that have been
observed in several consecutive years of data from ap
proximately 30 States. Observations, statistical tests,
and the geographic and industrial diversity of the States
support the hypothesis that these data are representative
of the national experience. An example of how s d s data
can be used in conjunction with the annual survey data
follows: In 1979, disabling injuries or illnesses affecting
the back accounted for 22.5 percent of all cases in the
SDS program; the total number of lost workday injuries
or illnesses collected through the annual survey was ap
proximately 2.8 million, resulting in a national estimate
of 630,000 disabling back injuries for that year (22.5
percent x 2.8 million).
Presentation
Sds data are available from the National Technical
Information Service ( n t i s ). 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 avail
able from each State can be obtained from the Office of
Occupational Safety and Health Statistics, Bureau of
Labor Statistics, Department of Labor.
Us@s 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
Report o f the National Commission on State Work
men's Compensation Laws suggested that systema
tic collection and exchange of data would be a valuable
source of information for both compensation and safety
agencies.
The s d s is a step in this direction. Because s d 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, in
cluding the types of cases handled and the predominant
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 sd s suffer
from the same low degree of identification as that
experienced in the annual survey of occupational in
juries and illnesses. Recognition of occupational illness
depends on the “ state of the art.” As medical
knowledge increases, illness identification will improve
in both data collection systems.
types of affected workers and work situations. The data
direct attention to problem areas which can be most ef
fectively handled by safety and health standards, train
ing, or compliance programs.
Although the Supplementary Data System standard
izes the classification, processing, and tabulations of
data, it is not a complete census of occupational injuries
and illnesses; as of 1980, 35 States were participating. In
addition, coverage and reporting requirements varia
tions reflect differences in State workers’ compensation
laws. Differences also exist because of statutory and ad
Technical References
Hilaski, Harvey J. “ Understanding Statistics on Occupational
Illnesses,” Monthly Labor Review, March 1981.
Root, Norman. “ Injuries at Work Are Fewer Among Older
Employees,” Monthly Labor Review, March 1981.
Hilaski, Harvey J., and Wang, Chao Ling. “ How Valid are
Estimates of Occupational Illnesses?,” Monthly Labor
Review, August 1982.
Root, Norman, and Sebastian, Deborah. “ B l s Develops
Measure of Job Risk by Occupation,” Monthly Labor
Review, October 1981.
McCaffrey, David. “ Work-Related Amputations by Type and
Prevalence,” Monthly Labor Review, March 1981.
Schauer, Lyle, and Ryder, Thomas. “ New Approach to
Occupational Safety and Health Statistics,” Monthly
Labor Review, March 1972.
Root, Norman, and McCaffrey, David. “ Providing More
Information on Work Injury and Illness,” Monthly
Labor Review, April 1978.
U .S. Department o f Labor, Bureau o f Labor Statistics.
Occupational Injuries and Illnesses in the United States by
Industry (annual).
Root, Norman, and Hoefer, Michael. “ The First Work-Injury
Data Available From New b l s Study,” Monthly Labor
Review, January 1979.
B l s annual statistical bulletin analyzing occupational
injuries and illnesses in the United States.
Bureau o f Labor Statistics. What Every Employer Needs To
Know About o s h a Recordkeeping, Report No. 412-3,
1978.
Provides answers to questions employers most fre
quently asked about the keeping of records of oc
cupational injuries and illnesses under the Occupational
Safety and Health Act of 1970.
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 Data,” Monthly Labor
Review, September 1980.
128
Chart 1
Guide to recordability of cases under the Occupational Safety and Health Act
129
U„§. Department of Lalbor
1981 OSH A No. 200-S
Annual Occupational Injuries and Illnesses Survey Covering Calendar Year 1981
The inform ation collected on this form will be used for statistical purpooeo only by the BLS, OSH A , and the cooperating State Agencies.
Sch. No.
St.
Ck.
Suf.
Bureau of Labor Statistics for the Occupational Safety and Health Adm inistration__________
O.M .B. No. 1220-0045
TH IS R E PO R T IS M A N D A T O R Y U N D E R PU BLIC LAW 91-596. F A IL U R E TO R EPORT
Approval expires 9 /3 0 /8 2
CAN R E S U L T IN TH E ISSUANCE OF C IT A T IO N S A N D ASSESSM ENT OF P EN A L TIE S .
SIC
E D IT
I.
A N N U AL AVERAGE
E M P L O Y M E N T IN 1981
IS. T O T A L H O U R S
W O R K E D ON 1981
Enter the average number
of employees who worked
during calendar year 1981
in the establishment(s)
covered by this report. In
clude all classes of em ploy
ees: full-tim e, part-tim e,
seasonal, tem porary, etc.
See the instructions for an
example of an annual aver
age em ploym ent calcula
tion. (R o u n d to the
nearest w hole num ber.)
Enter the total number
of hours actually worked
during 1981 by all em
ployees covered by this
report. DO N O T include
any non-worktime even
though paid such as vaca
tions, sick leave, etc. If
employees worked low
hours in 1981 due to lay
offs, strikes, fires, etc.,
explain under Comments
(section V II) . (R ound to
th e nearest w hole num ber)
Complete this report whether or not there were
recordable occupational injuries or illnesses.
PLEASE KEAD THE ENCLOSED INSTRUCTIONS
t il. N A T U R E O F B U SINESS IN 1981
A Check the box which
best describes the general
type of activity performed
by the establishment(s)
included in this report.
□ Agriculture
□ Forestry
□ Fishing
□ Mining
□ Construction
□ Manufacturing
□ Transportation
□ Communication
□ Public Utilities
□ Wholesale Trade
□ Retail Trade
□ Finance
□ Insurance
□ Real Estate
□ Services
□ Public Administration
R E P O R T L O C A T IO N A N D ID E N T IF IC A T IO N
Complete this report for the establishment(s) covered by the description below:
B. Enter in order of im
portance the principal
products, lines of trade,
services or other activi
ties. For each entry also
include the approximate
percent of total 1981
annual value of produc
tion, sales or receipts.
C. If this report in
cludes any establish
m en ts) which per
form services for
other units of your
company, indicate
the primary type of
service or support
provided. (Check
as m an y as apply.)
%
%
1. □ Central
administration
2. □ Research, develop
ment and testing
3. □ Storage
(warehouse)
4. □ Other (specify)
%
Please indicate any address changes below.
Complete and return OWLY
THIS FORM within 3 weeks
IV . M O N T H O F OSHA
IN S P E C TIO N
If the establishment(s)
covered by this report
had either a Federal or
State O SH A compliance
inspection during cal
endar year 1981,
please enter the name
of the month in which
the first inspection
occurred.
V. R E C O R D A B LE
IN J U R IE S A N D
ILLN ESSES
Did the estab
lishm ents) have
any recordable
injuries or ill
nesses during
calendar year
1981?
1 .D No (Please
complete
section VII.)
2 . 0 Yes (Please
complete
sections V I
and V II.)
(Leave this
box blank.)
SEE R E VE R S E C >
R E T U R N R E PO R T TO :
F o information Call:
o*
OSHA No. 200-S (Rev. Oct, 1981)
V I. OCCUPATIONAL IN JU RY AND ILLNESS SUMMARY (CoveringCalendar Year 1981)
© Complete this section by copying totals from the annual summary o f your 1981 OSHA No. 200.
O
O
Remember to reverse the carbon insert before completing this side.
Leave section V I blank if there were no OSHA recordable injuries or illnesses d u r in g 1981,
O Note: First aid even when administered by a doctor or nurse is not recordable.
O C C U P A T I O N A L I N J U R Y CASES
IN J U R Y
RELATED
FA TA L
IT IE S 00
(D E A T H S )
O C C U P A T IO N A L
IN JU R IE S
W IT H O U T
LOST
W ORK
DAYS0
IN J U R IE S W IT H LO ST W O R K D A Y S
Injury cane®
w ith days
sway from
w ork and/or
restricted
workdays
Injury
cases
w ith days
away
from
work
Please check your figures to be certain that the sum of entries in columns (7a) + (7b)
+ (7c) + (7d) + (7e) + (7f) + (7g) = the sum of entries in columns (8) + (9) + (13).
If you listed fatalities in columns (1) and/or (8 ), please give a brief description of
the object or event which caused each fatality in the "Comments" section.
Total
days
away
from
work
IL L N E S S C A S E S
ILLN E SS
ILLN E SS E S
RELATED
FA TA L
IT IE S 00
(D E A TH S )
Illness cacao
with days
away from
work and/or
restricted
workdays
Total
days of
restricted
activity
W IT H LOST W O R K D A Y S
lllnsoo
cases
with days
away
from
work
Total
days
away
from
work
ILLNESSES
W IT H O U T
LOST
W ORK
DAYS0
Total
days of
restricted
activity
Num ber of
DEATHS
in col. 1
of the log
(OSHA
No. 200)
Num ber of * Num ber of
CHECKS
CHEC K S
in col. 2
in col. 3
of the log
of the log
(OSH A
(OSHA
No. 200)
No. 200)
Sum of
the D A Y S ,
in col. 4
of the log
(OSHA
No. 200)
Sum of
the DAYS
in col. 5
of the log
(OSHA
No. 200)
Number of
CHECKS
in col. 6
of the log
(OSHA
No. 200)
Number of
D E A TH S
in col. 8
of the log
(OSHA
No. 200)
Number of
CHECKS
in col. 9
of the log
(OSHA
No. 200)
Number of
CHECKS
in col. 10
of the log
(OSHA
No. 200)
Sum of
the D A Y S
in col. 11
of the log
(OSHA
No. 200)
Sum of
the DAYS
in col. 12
of the log
(OSHA
No. 200)
Number of
CHECKS
in col. 13
of the log
(OSHA
No. 200)
(1)
(2)
(4)
(5)
(6)
(8)
(9)
(10)
(11)
(12)
(13)
(3)
DEATHS
D EATH S
WITHO UT LOST.WORKDAYS-CASES (WITH NO DAYS LOST) RESULTING IN EITHER: DIAGNOSIS OF OCCUPATIONAL ILLNESS, LOSS OF CONSCIOUSNESS, RESTRICTION
OF WORK OR MOTION, TRANSFER TO ANOTHER JOB. OR MEDICAL TREATMENT BEYOND FIRST AID.
V I8. REPORT PREPARED BY (Please type or print)
M M _______________________________
AE
T I T L E ____________________________________________
S IG N A T U R E
A R E A CODE
D A T E _______
PHONE
‘ IF YOU LISTED FAT ALLTI ES IN COLUMNS (1) AND/OR (8), PLEASE GIVE A BRIEF
DESCRIPTION OF THE OBJECT OR EVENT WHICH CAUSED EACH FATALITY IN
THE "COMMENTS" SECTION BELOW.
COM M ENTS.
S U R V E Y R EPORTING REGULATIONS
T itle 29, Part 1904.2 0-2 2 o f the Code of Federal Regulations requires that:
each employer shall return the completed survey form, OSHA No. 200-S, within
3 weeks o f receipt in accordance with the instructions shown below.
INSTRUCT IQMS FOR COMPLETING TOE OSHA WO. 200-S FO TO
1981 OCCUPATIONAL INJURIES AMD ILLNESSES SURVEY
{Cowering GaSendsr Year 1981)
Change o f Ownership—When there has been a change of ownership during the report period,
only the records of the current owner are to be entered in the report. Explain fully under
Comments (Section V II) , and include the date of the ownership change and the time period
this report covers.
Partial-Year Reporting—For any establishment(s) which was not in existence for the entire
report year, the report should cover the portion of the period during which the establish
m e n ts ) was in existence. Explain fully under Comments (Section V II) , including the time
period this report covers.
ESTABLISHMENTS IN C LU D E D ON THE REPORT
This report should include only those establishments located in, or identified by, the Report
Location and Identification designation which appears next to your mailing address. This
designation may be a geographical area, usually a county or city, or it could be a brief de
scription of your operation w ithin a geographical area. If you have any questions concerning
the coverage of this report, please contact the agency identified on the O SH A No. 200-S
report form.
D E F IN IT IO N OF ESTABLISHMENT
A n E S T A B L IS H M E N T is d e fin e d as a single physical lo catio n w h e re business is co n d u c te d
or w h ere services o r in d u s tria l o p e ra tio n s are p e rfo rm e d . (F o r exa m p le: a fa c to r y , m ill,
store, h o te l, re s ta u ra n t, m o v ie th e a tre , fa rm , ranch , b a n k , sales office, w arehouse, or
cen tral a d m in is tra tiv e o ffic e .)
F o r firm s engaged in a c tiv itie s such as c o n s tru c tio n , tra n s p o rta tio n , c o m m u n ic a tio n , or
electric, gas an d s a n ita ry services, w h ic h m a y be p h y sica lly dispersed, reports should cover
th e place 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 each day.
R e p o rts fo r personnel w h o do n o t p r im a rily re p o rt or w o rk a t a single establishm ent, such
as tra veling salespersons, te c h n ic ia n s , engineers, e tc ., should cover th e lo cation fro m w h ich
th e y are p aid o r th e base f r o m w h ic h perso nnel o p e ra te to carry o u t th e ir activities.
SECTION I.
A N N U A L AVER A G E EMPLOYMENT IN 1981
Enter in Section I the average (not the total) number of full and part-time employees who
worked during calendar year 1981 in the establishment(s) included in this report. If more
than one establishment is included in this report, add together the annual average employ
ment for each establishment and enter the sum. Include all classes of employees— seasonal,
temporary, administrative, supervisory, clerical, professional, technical, sales, delivery, in
stallation, construction and service personnel, as well as operators and related workers.
Annual Average employment should be computed by summing the em ploym ent from all
pay periods during 1981 and then dividing that sum by the total number of such pay periods
throughout the entire year, including periods with no employment. For example, if you had
the following m onthly em ploym ent— Jan.-10; Feb.-10; Mar.-10; A p r.-5; May-5; June 5;
July-5; Aug.-O; Sept.-O; Oct.-O; Nov.-5; Dec.-5— you would sum the number of employees
for each m onthly pay period (in this case, 60) and then divide that total by 12 (the number
of pay periods during the year) to derive an annual average employment of 5.
SECTION II.
TO TA L HOURS W ORKED IN 1981
Enter in Section II the total number of hours actually worked by all classes of employees
during 1981. Be sure to include O N L Y time on duty. DO N O T include any non-work time
even though paid, such as vacations, sick leave, holidays, etc. The hours worked figure should
be obtained from payroll or other time records wherever possible; if hours worked are not
maintained separately from hours paid, please enter your best estimate. If actual hours
worked are not available for employees paid on commission, salary, by the mile, etc,, hours
worked may be estimated on the basis of scheduled hours or 8 hours per workday.
For example, if a group of 10 ,salaried employees worked an average of 8 hours per day, 5
days a week, for 50 weeks of the report period, the total hours worked for this group would
b e 1 0 x 8 x 5 x 50 = 20 ,000 hours for the report period.
SECTION III.
NATURE OF BUSINESS IN 1981
In order to verify the nature of business code, we must have information about the specific
economic activity carried on by the establishment(s) included in your report during calendar
year 1981,
Complete Parts A , B and C as indicated in Section 1 1 on the O SHA No. 200-S form. Complete
1
Part C only if supporting services are provided to other establishments of your company.
Leave Part C'blank if a) supporting services are not the primary function of any establish
m ents) included in this report or b) supporting services are provided but only on a contract osfee basis for the general public or for other business firms. (Instructions continued on page 2.)
M OTE: If more than one establishment is included, information in Section III should reflect
the combined activities o f all such establishments. One code will be assigned which best
indicates the nature of business of the group of establishments as a whole.
SECTION IV .
M O NTH OF OSHA INSPECTION
Enter the name of the first month in 1981 during which your establishment(s) had an
OSHA compliance inspection. Include inspections under the Federal or State equivalents of
the Occupational Safety and Health Act by Federal or State inspectors and other inspections
which may result in penalties for violations of safety and health standards. Do not include
inspections lim ited to elevators, boilers, fire safety or those which are consultative in nature.
SECTION V.
RECORDABLE INJURIES OR ILLNESSES
Check the appropriate box. If you checked "Yes," complete Sections V I and V II on the
back of the form. If you checked "N o ," complete only Section V-ll.
SECTION V I.
OCCUPATIONAL INJURY AND ILLNESS SUMMARY
This section can be completed easily by copying the totals from the annual summary of
your 1981 O S H A No. 200 form (Log and Summary of Occupational Injuries and Illnesses).
Please note that if this report covers moir® than one establishment, the final totals on the
"Log " for each must be added and the sums entered in Section V I.
Leave Section V I blank if the employees covered in this report experienced no recordable
injuries or illnesses during 1981.
If there were recordable injuries or illnesses during the year, please review your OSHA
No. 200 form for each establishment to be included in this report to make sure that all
entries are correct and complete before completing Section V I. Each recordable case should
be included on the "L o g " in only one of the six main categories of injuries or illnesses:1
1.
2.
3.
4.
5.
6.
IN J U R Y —related deaths (Log column 1)
IN J U R IE S w ith days away from work and/or restricted days (Log column 2)
IN J U R IE S w ith o u t lost workdays (Log column 6)
I L L N E S S -re la te d deaths (Log column 8)
IL LN E S S E S w ith days away from work and/or restricted days (Log column 9)
IL LN E S S E S w ithou t lost workdays (Log column 13)
Also review each case to ensure that the appropriate entries have been made for the other
columns if applicable. For example, if the case is an Injury with Lost Workdays, be sure that
the check for an injury involving days away from work (Log column 3) is entered if necessary.
Also verify that tiie correct number of days away from work ( Log column 4) and/or days of
restricted work activity (Log column 5) are recorded. A similar review should be made for a
case which is an Illness with Lost Workdays (including Log columns 10, 11 and 12). Please
remember that if your employees' loss of workdays is still continuing at the time the annual
summary for the year is completed, you should estimate the number of future workdays
they will lose and add this estimate to the actual workdays already lost. Each partial day
away from w ork, other than the day of the occurrence of the injury or onset o f illness,
should be entered as one full restricted workday.
Also, for each case which is an Illness, make sure that the appropriate column indicating
Type of Illness (Log columns 7a-7g) is checked.
A fte r completing your review of the individual case entries on the "Log," please make sure
that the "Totals" line has been completed by summarizing Columns 1 through 13 according
to the instructions on the back of the "Log" form. Then, copy these "Totals" onto Section
V I of the O SH A No. 200-S form. If you entered fatalities in columns (1) and/or (8), please
include in the 'Comments" section a brief description of the object or event which caused
each fatality.
FIRST A ID
Finally, please remember that all injuries which, in your judgement, required only First Aid
Treatment, even when administered by a doctor or nurse, should not be included in this re
port. First Aid Treatment is Hefined as one-time treatm ent and subsequent observation of
m inor scratches.cuts, burns, splinters, etc., which do not ordinarily require medical care.
SECTION V II.
COMMENTS AND ID E N T IFIC A TIO N
Please complete all parts including your area code and telephone number. Then return the
O SHA No. 200-S form in the pre-addressed envelope. KEEP your file copy.
EDsair Em ployer:
The Occupations! Safety and Health Act o f 1970 requires the Secretary of Labor to collect, compile, and analyze statis
tics on occupational injuries end illnesses. This is accomplished through a jo int Federal/Stafe survey program w ith
States that have received Federal grants for collecting and compiling statistics. Establishments are selected fo r this sur
vey on a sample basis with varying probabilities depending upon size. Certain establishments may be included in each
year's sample because o f their importance to the statistics for their industry.
You have been selected to participate in the nationwide Occupational Injuries and Illnesses Survey for 1981. Under the
Occupational Safety and Health A ct, your report is mandatory:
The following items are enclosed fo r your use: (1) Instructions fo r completing the form ; (2) The O SH A 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 w ithin 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 im portant survey.
T H O R N E G. A U C H TE R
Assistant Secretary for
Occupational Safety and Health
Ohapt@r 18. Labor Fore©
Fr©|@©!5®n8
Bls develops and publishes long-term projections of
the labor force—estimates of its future size and com
position—as part of a comprehensive and integrated
framework for analyzing the implications of growth for
the national economy and for employment by industry
and occupation. Projections, based on specified
assumptions, are made for about 15 years ahead. Seven
sets of labor force projections have been prepared since
1959. The most recent projections, for the labor force as
a whole and for 54 separate age-sex-race groups, were
published in December 1980 for 1985, 1990, and 1995.1
The basic assumptions that underlie all the labor
force projections are: (1) Work patterns will not change
significantly over the projection period; for example,
the average workweek will not be sharply reduced; (2)
social and educational trends will continue, such as the
trend toward increased schooling beyond high school;
and (3) there will be no major war or significant change
in the size of the Armed Forces.
S 3® ©
H th d3s
Projections of the labor force require, first, projec
tions of the population. These are prepared by the
Bureau of the Census by age, sex, and race, based on
trends in birth rates, death rates, and net migration.
Since birth rates pose the most uncertainty in projecting
the population, the Bureau of the Census prepares
several series of projections based on differing assump
tions with respect to birth rates. The most recent bls
labor force projections incorporated the Bureau of the
Census middle (Series II) birth rate projections, which
have the total fertility stabilizing at 2.1 births per
woman by the year 2050.1
2
Once population projections have been prepared, bls
projects labor force participation rates—the proportion
of various groups in the population who will be working
or seeking work. Projections are made for 54 separate
demographic groups since both the level and trends of
participation vary considerably by age, sex, and race.
1Howard N Fullerton, Jr., “ The 1995 Labor Force: A First Look,”
Monthly Labor Review, December 1980.
2 Bureau of The Census, Projections o f the Population o f the United States:
1977 to 2050, Current Population Reports, series P-25, No. 704, 1977.
The labor force participation projection for each age,
sex, and race group is developed by : (1) Analyzing past
rates of growth, (2) selecting a time period deemed most
appropriate for each group, and (3) modifying that rate
if past trends are not likely to continue throughout the
entire projection period.
The projected participation rate for each group is
then multiplied by the corresponding population projec
tion to obtain the labor force projection for that group.
These are summed to obtain the total labor force. At
each stage of projection, the results for specific age, sex,
and race groups are reviewed and modified if not consis
tent with other demographic groups.
In recent years, three alternative sets of assumptions
(scenarios) regarding labor force participation have
been developed for each set of projections. In the latest
projections, for example, one scenario, a high-growth
scenario, assumes a rapid growth in the labor force par
ticipation of women in the 1980’s and the convergence
of participation rates of black and white men under the
age of 65. (These rates have been diverging since 1955.)
A second scenario, the middle-growth scenario, assumes
only the rapid growth of women’s participation. The
low-growth scenario assumes a moderate rather than a
rapid increase in women’s participation and a continued
divergence in the participation rates of black and white
men.
New approaches
Since the last set of projections was prepared, the
Bureau has been examining alternative ways to prepare
labor force projections. The alternative under con
sideration is an economic model of labor force par
ticipation rates. The model just described is an ex
trapolation of past trends in participation rates to some
target year with no explicit consideration of economic
influences on participation rates.
Us®s and limitations
Labor force projections are a basic factor in es
timating 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 addition, along with other factors, they are
used by planners in business and industry to estimate de
mand for their products, develop marketing plans, and
evaluate expansion programs.
As is the case for all projections, users of labor force
projections need to be aware of the underlying assump
tions and should consider projections as likely outcomes
in the light of current and expected trends, not as
forecasts of the future.
TsdMtgafl References
Bur©au © Labor Statistics
f
Bl
s
Economic Growth Model System Usedfor Projections to
1990. Bls Bulletin 2112, 1982.
Flaim, Paul O., and Fullerton, Howard N, Jr. “ Labor Force
Projections to 1990: Three Possible Paths,” Monthly
Labor Review, December 1978. Originally presented at
the August 1978 meeting of the American Statistical
Association. Reprinted in Richard L. Rowan (ed.),
Readings in Labor Economics and Labor Relations,
Chicago, Richard D. Irwin, 1980; also in Employment
Projections for the 1980’ bls Bulletin 2030, 1979.
s,
Fullerton, Howard N, Jr. “ The 1995 Labor Force: A First
L o o k ,” Monthly Labor Review, December 1980.
Reprinted in b l s Projections to 1990, bls Bulletin 2121,
1982.
Fullerton, H. N, Jr., and Flaim, P. O. “ New Labor Force
Projections to 1990,” Monthly Labor Review, December
1976. Reprinted as Special Labor Force Report 197.
Ryscavage, Paul M. “ Bls Labor Force Projections: A Review
of Methods and Results,” Monthly Labor Review, April
1979.
Bureau © the Census
f
Fullerton, Howard N, Jr. “ How Accurate Were Projections
o f the 1980 Labor Force?” Monthly Labor Review, July
1982.
Projections of the Population of the United States: 1977 to
2050, Current Population Reports, series P-25, No. 704,
1977.
136
Chapter 19. Economic Growth
Studies
The primary objective of the Bureau’s studies of
economic growth is to develop projections of industry
employment opportunities under alternative assump
tions in order to analyze various economic problems
such as the future utilization of available labor
resources. A system of models serves as a basis for mak
ing the economic and employment projections.
Methods
Macroeconomic model
A macroeconomic model is used to project gross na
tional product (gnp ) and its major demand components
under different sets of assumptions. These assumptions
involve such factors as demographic trends, the
unemployment rate, inflation, government tax and ex
penditure policies, and long-run productivity trends.
The macroeconomic model provides estimates of
growth in the major sectors of the economy that are
consistent with all assumptions and conditions of a par
ticular projection scenario. The purpose of the ag
gregate projections is to provide consistent and in
tegrated control totals for the projected industry pur
chases that are developed later in the system.
A macro model used recently was a relatively smallscale model (approximately 50 equations) whose pur
p o se w as to capture the im pact o f th ose factors w hich
affect aggregate demand and supply over the medium to
long term. The model was structured around a
framework in which the output produced is balanced
with output demanded via income flows. To bring
about this balance between supply and demand g np , the
model was structured to respond to fiscal policy
changes, which affect the level and distribution of
spendable income in the personal and corporate sectors.
Assumptions made in developing the macro projections.
There were 51 variables in this bls macroeconomic
model that were exogenous, or that had to be estimated
externally in various ways for the projected periods.
From a solution point of view, all exogenous variables
are considered assumptions. From a structural ap
proach, however, the exogenous variables were grouped
in three ways. First were those items projected with
sophisticated techniques outside the Office of Economic
Growth and Employment Projections such as the
population projections. Second were items which
represented either policy instruments or policy goals.
The policy instruments, such as Federal tax rates or
Federal employment levels, represent the Federal
Government’s position at any particular time. The
policy goals, such as the unemployment rate or the
Federal deficit, were the result of such measures.
Finally, there were those exogenous variables which
were assumptions in the narrowest sense; i.e., a judg
ment as to the probable course of a particular item. An
example of this category would be interest rates.
Balancing the macro model. Summation of the derived
real components of demand yields the demand-side
estimate of g n p . The demand- and supply-side estimates
of gnp ordinarily will not agree, and the magnitude of
such an imbalance is calculated. A positive sign for this
gap represents a situation of excess supply, while a
negative sign indicates excess demand. Although the
sum of disposable incomes for all of the sectors
necessarily equals the estimate of the g n p , demand for
gnp will fall short of or exceed the supply of gnp unless
the total purchases of the various sectors happen to
equal their combined incomes.
The gap between supply and demand GNP depends in
part on the government policies incorporated in the
model. If there is a gap, this implies that the target rate
of unemployment cannot be achieved with the existing
fiscal assumptions. Thus, the various policy instruments
in the model are modified to effect a balance between
supply and demand. Many combinations are possible,
and a final choice is made on the basis of many con
siderations that are outside the model.
Final demand projections
Gross national product is the final output of the
economy measured from the demand side, or the output
of the economy distributed among its final users. Final
users are broadly categorized as persons, businesses.
137
governments, and foreign. Final demand consists of the
purchases made by these groups, or the purchases of the
demand sectors of g np . Final demand projections in
volve estimating the future purchases of each demand
sector, by industry of origin. For recent projections, the
economy was disaggregated into 156 different in
dustries. These industries defined the bills of goods, or
lists of purchases, prepared for each final demand sec
tor. The output of the macro model provides control
totals for each final demand sector. The first step in
projecting distributions of purchases for each sector is
to develop data series for the purchases each made in
past years. The years studied are primarily years for
which the U.S. Department of Commerce has published
input-output studies (1958, 1963, 1967, and 1972);1
1972 became a base year for the recent projections. In
addition, many other data series are available. These
historical data are used with a variety of techniques and
submodels to project purchases.
Personal consumption. The distribution of total per
sonal consumption expenditures (pce ) to producing in
dustries is accomplished in two major steps. After total
consumption is determined by the macroeconomic
model, the first step is to project consumption by type
of expenditure for 12 major product groups, which the
Department of Commerce has defined and for which it
maintains data series. These 12 major product groups
are summed and then scaled to the projected total con
sumption provided by the macro model. Next, using
these 12 product groups, a set of 82 lower level product
categories, also maintained by the Department of Com
merce, are projected. These 82 detailed product
categories are also scaled to sum to the appropriate 12
aggregate con trols. H istorical data for each o f these 82
categories are available from the Department of Com
merce as part of the National Income and Product
Accounts.
The second step is to distribute each of the 82 product
expenditures to the producing industries. This is ac
complished using projected “ bridge tables” or matrices
which distribute each of the 82 categories to its compo
nent industries as well as to the transportation, in
surance, and trade industries. The results are aggregated
to form the PCE bill of goods, the largest component of
final demand.
A consumption submodel is used to project the 12
major product groups as well as the 82 detailed product
categories. This model, which relates consumer expen
ditures primarily to income and prices, was originally
developed by Houthakker-Taylor,2 with the 1958
Investment. The development of historical bills of
goods for producers durable equipment (pde ) involves
two approaches that provide a check on the consistency
of the data base from which the projections are made.
The first approach studies the growth of demand in
equipment over time. Annually, the national income ac
counts show pde distributed among 24 major categories
such as agricultural m achinery, con stru ction m achinery,
1 Input-Output Structure o f the United States 1958; 1963; 1967; and 1972
(U .S. Department o f Commerce, Bureau o f Economic Analysis).
1
H .S. Houthakker and Lester D. Taylor, Consumer Demand in the United
States: Analyses and Projections (Cambridge, M ass., Harvard University Press,
1970).
constant-dollar data from 1929 to 1964 used to estimate
a set of 82 product expenditure categories. Total pce
and the annual change in pce are primary variables used
as a proxy for disposable income. P ce has a high level
of explanatory power in these equations. Relative
prices, which are calculated as the implicit price deflator
for that good or service divided by the implicit price
deflator for total pce , are also used extensively. The lag
structure of the equations allows the effect of changes in
explanatory variables to be distributed over time.
A bridge table distributes the 82 product categories to
their component goods and services and the margin in
dustries, i.e., wholesale and retail trade margin and
transportation costs. The products are expressed in pur
chasers’ values, while the bills of goods or the producing
industries are expressed in producers’ values. The dif
ference is the cost added to a particular industry’s out
put in getting that output from the point of production
to the consumer, including transportation costs
(railroad, truck, water, air, and pipeline costs), in
surance costs (for imports only), and wholesale and
retail trade markups. The bridge table accomplishes two
tasks at once—it allocates each of the 82 product
categories to its producing industries, and removes the
transportation and trade margins from the product and
allocates them accordingly.
138
communication equipment, etc. Each of the 24 cate
gories is in purchaser prices and contains a varying
number of supplying industries. For the years for which
input-output tables were prepared (1-0 years), bridge
tables are available which allocate each of these 24
categories to the margin and the supplying industries.
Bridge tables for non-I-0 years are constructed by inter
polation.
The second approach makes use of the assumption
that an industry’s investment is a function of its output.
The Annual Survey o f Manufacturers and the Census o f
Manufactures are the sources for equipment investment
of the historical period. For 1-0 years, capital flows
tables are available which allocate the total investment
of each industry to the supplying industries, thus, pro
ducing a pde bill of goods. Bills of goods derived by
these two approaches can be compared to show changes
that are occurring in the bridge table and the capital
flows matrix.
To make pde projections, both investment output
ratios and capital flows are projected based on historical
trends. Projected outputs by industry are first derived,
then the projected investment output ratios are applied
to derive the level of investment by each industry. This
level of investment is run through a capital flows table
giving a pbe bill of goods. This total investment is made
to equal total pde as derived from the macro model
runs. Obviously, changes in the distribution of pbe by
industry change the output level of each industry which
*causes a further change in the required investment. Ad
justments are made repeatedly to the pbe column until
pbe demand in each industry equals the level of invest
ment that was actually required by the distribution of
output.
Foreign trade projections. For most industries, the
foreign trade projections rely on an analysis of the
trends of imports and exports as shares of total output.
The ratios for 1963, 1967, 1972, and, for merchandise
trade, in 1977 are compared, and the trend carried out
to future years. Ratios are applied initially to estimate
imports and exports. The industry levels of imports and
exports are added and scaled to the total values of the
macro model.
The results are modified, in some cases, based on a
comparison with previous b l s projections of imports
and exports and special analyses. Where the previous
projections relied on special analyses or special trade
agreements that were still in effect, these are taken into
account. Special studies are conducted for important
import and export goods. For example, studies are
made for automobile and electronic imports. Specific
assumptions are made for the energy industries based,
to the degree possible, on the Department of Energy’s
projected rates of growth for domestic output and im
ports under certain price conditions.
Government. The macroeconomic model estimates of
projected State and local government purchases are con
sistent with all macro assumptions and estimates, in
cluding grants-in-aid. This model provides a purchase
total for each projected year, with subtotals for educa
tion and for all other functions as a group. Both of these
categories are divided into compensation and all other
purchases.
A State and local government model has been recently
used which predicts expenditures and employment in
current dollars for 20 functions. These functions are
projected based upon Census and b e a data by calendar
year. They include: (1) Elementary and secondary
education, (2) higher education, (3) other education, (4)
libraries, (5) highways, (6) health, (7) hospitals, (8)
sewerage, (9) public utilities, (10) natural resources, (11)
corrections, (12) police, (13) fire, (14) sanitation, (15)
public welfare, (16) local parks and recreation, (17)
general government, (18) other government enterprises,
(19) public housing, and (20) water and air terminals.
The model structure is based upon data for the years
1960-73. Equations for each function are first estimated
for expenditures and employment. Expenditures in the
model are in current dollars and apply to all outlays, not
just purchases of goods and services. Another set of
equations is used to convert expenditures to purchases
and compensation. A final set of equations is used to
convert purchases to constant dollars. Employment is
estimated in full-time equivalent units. The model is
driven by four major groups of variables: Growth in
personal income; demographic data; grants-in-aid; and
an “ all other” category that includes interest rates,
prices, and unemployment rates.
The macro model levels of projected Federal pur
chases are established exogenously in the process of
balancing supply and demand g n p . This model provides
values for total purchases, total compensation of
military and civilian employees, as well as the number of
civilian and military employees. The levels are estab
lished to insure consistency with overall projection
assumptions. Assumptions are of major importance in
the Federal sector since, in many cases, past experience
is not useful for projection. For example, the projec
tions have always assumed peaceful conditions without
international tensions. A contrary assumption of war
would result in large Federal purchases and a much
larger defense share.
Regression equations are used to derive the total pur
chases of the six Federal Government subfunctions used
in deriving these projections. These are modified based
upon expected program levels in the case of defense and
space. The six subfunctions are modified until they
come to the established macro totals for Federal
Government expenditures. Real compensation is also
derived for each subfunction using regression equa
tions. Historical data for defense and nondefense new
construction are used to derive regression equations to
project purchases from the six new construction in
dustries for each major component of the Federal sec
tor; these two values are then allocated to the six sub
functions based on historical trends.
Projecting inpmtroytpof ccefficients
The input-output tables used as a base in the
economic growth model are developed by the Bureau of
Economic Analysis, Department of Commerce. How
ever, these input-output tables incorporate the
technology and product mix for a base year, and may
not reflect the technology and product mix which may
prevail during the period for which the projection is be
ing made. Thus, it is necessary to project changes in the
input-output coefficients.
Input-output coefficients are projected to change for
several reasons—technological change is an important
factor, but not the only one. Changes in product mix or
relative prices could also cause significant changes in
139
coefficients. Because the bls industries are aggregates
of the more detailed sectors used by the Bureau of
Economic Analysis in compiling the base year inputoutput table, a simple change in the relative importance
of those sectors could have a large impact on the coeffi
cients. Also, as the relative price or availability of
substitute inputs changes, substitutions might occur.
Several different methods are used in projecting coef
ficients. Energy coefficients, both as inputs to other in
dustries and as inputs from other sectors to energy pro
ducing industries, are based, in part, on projections
available from the Department of Energy. Several in
dustries are studied intensively to pick up structural
changes which have occurred since the most recent
Department of Commerce 1-0 study and are then pro
jected forward (for example, the metals industries). In
other industries, changes in tastes are incorporated (for
example, a decrease in sugar in foods and soft drinks).
For other commodities, the rows of the input-output
tables are evaluated and increases or decreases
throughout the economy are made based upon overall
trends in the industry. In some cases, old coefficients
are reweighted based upon expected changes in the
relative importance of detailed industries. Where
resources are not available to study specific coefficients,
they are left unchanged from their previous level.
CP
Cap
and x, y, and z are distribution parameters which are
greater than zero and sum to unity.
w = substitution parameter
v = economies of scale
u = utilization parameter
The elasticity of substitution is equal to 1/(1 +W).
The factor demand model has been estimated for only
76 industries, due to the limited amount of employee
compensation data which is an input in the estimation
of labor services. In order to expand the results to the
156 industries currently in the Economic Growth
system, least squares time trends of labor productivity
and average hours are computed for each of the 156 in
dustries and combined with the 156-order output pro
jections to calculate hours and employment. Then, these
estimates are scaled to the projections from the factor
demand model.
Further disaggregation is required in order to develop
occupational projections. Occupational forecasts are
estimated at the 3-digit Standard Industrial Classifica
tion (sic) level, totaling about 450 individual industries.
Only wage and salary employment data are prepared at
this level. The estimates of jobs at this level of detail are
based on an equation for each industry expressing
employment as a function of total civilian employment,
the level of the Armed Forces, output of the corre
sponding economic growth sector (156-order), and
employment at the appropriate aggregate 2-digit sic
level.
Since projections contain many complex relationships
among economic variables that are developed through a
lengthy sequence of operations, it is necessary to have a
set of checks and balances to insure that the various
stages of the projections make up an internally consis
tent model. The primary element in this balancing is
analysis of results of each stage, modification where
results seem inconsistent, and rerunning the entire pro
jection system until results are consistent. The economic
growth model is designed to provide a feedback and
balancing procedure with respect to three of its
elements: Imports, investment, and employment. In
practice, all three of these elements must be brought in
to balance simultaneously.
Employment projections
Employment projections by industry are developed
using a set of industry productivity projections. The
model used to project annual industry employment and
productivity is a factor demand model which takes into
account the interdependence of both labor and capital
requirements in each industry. In this model, the de
mand for labor is a function of the industry’s output,
capacity utilization (measured by the unemployment
rate of last industry employed), technical change (as ap
proximated by a time trend), and the stock of capital
measured in efficiency units. The form of the model
utilizes a ces (constant elasticity of substitution) pro
duction function, involving factor-augmenting technical
change. Allowing for economies of scale, the produc
tion function can be written as follows:
(1) Y = Cap
u
[x(AL*L)
-w
+ y(BE*E)
-w
+ z(CP*P)
= efficiency augmenting function for plant stock
— capacity utilization
-w v/w
]
where:
Y = output
Uses and limitations
L = labor services
E = equipment stocks
The projections developed in the economic growth
program serve a number of uses. The employment pro
jections by industry are used in developing occupational
outlook projections. The projections developed by the
Bureau form an important part of the U.S. Govern
P = plant stocks
AL
= efficiency augmenting function for labor
BE
= efficiency augmenting function for equipment
stock
140
ment’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. The projections of gnp and in
dustry growth patterns are also used in private industry
to make diversification studies, market analyses, and
long-term capital plans.
The economic growth model permits analytical uses
in addition to long-term employment projections.
Specifically, the model can be used to generate the
industry-by-industry labor requirements of various
economic sectors or types of demand for recent years.
Estimates of this type have been made for some time for
national defense, consumption, exports, and other de
mand categories as a basis for the long-term projections
and for special projects. The model also has been ap
plied to estimate labor requirements for a variety of
specific Federal Government programs, such as the sales
of military equipment to foreign governments, defense
expenditures, and energy programs.
The projections developed using the bls Economic
Growth system are prepared on a 2-year cycle and
published in the Monthly Labor Review. The special
studies of job requirements are also published in the
Review but not on a regular basis.
The preparation of economic projections is, to a
degree, both a science and an art. Thus, misunderstand
ings may arise between the users, who feel the need for
exact numbers, and producers, who recognize their in
ability to predict with such precision. Such conflicts are
all the more likely because projections analysts generally
employ a framework which develops numerical answers
to specific questions, and users are inevitably tempted to
attribute to those numbers an exactness which they
should not be accorded. The Bureau attempts to address
this dilemma, in at least a small way, by making clear all
of the important assumptions underlying its projec
tions, by developing alternative versions which reflect
some of the uncertainties about the future, by
evaluating past projections to assist users in ap
preciating the unpredictable nature of certain future
events, and by updating the projections on a regular
2-year cycle.
Technical References
Bureau! of Labor Statistics
Economic Projections to 1990, Bulletin 2121, March 1982.
B ls Economic Growth Model System Used fo r Projections
to 1990, Bulletin 2112, 1982.
Time-Series Data fo r Input-Output Industries, Bulletin 2018,
1979.
Capital Stock Estimates fo r Input-Output Industries: Methods
and Data, Bulletin 2034, 1979.
141
Chapter 20. National Industry
Oeoupational Matrix
each industry. Estimates of occupational employment
of self-employed and unpaid family workers are
prepared at the total (all-industry) level only. They are
added to the total of wage and salary workers to derive
total employment by detailed occupation for the entire
economy.
The Bureau develops comprehensive data on employ
ment in detailed occupations cross-classified by industry
in the form of a matrix, or table. The matrix can be
presented in absolute numbers or in ratios which show
the proportion of total employment in each industry ac
counted for by each occupation. The data can also be
transposed to show how total employment in an occupa
tion is distributed by industry. The Bureau develops
matrices for current and future years.
Wage and salary workers in OES survey industries. OES
data on occupational employment of wage and salary
workers cover all industries except agriculture and
private households. To develop current-year occupa
tional employment estimates for o e s industries, staffing
patterns are calculated from the most recent survey for
each industry. (Staffing patterns are the ratios of
employment in each detailed occupation in an industry
to total employment in the industry.) These staffing pat
terns are then applied to the current-year annual
averages of industry employment taken from CES data.
In some industries, employment data for some detail
ed occupations are not collected in the OES surveys
because the numbers are too small to be measured ac
curately and because the survey questionnaire in each
industry is limited to 200 occupations. To develop total
employment estimates for an occupation not included in
a survey questionnaire, but which is known to be pre
sent, detailed occupational employment is disaggregated
from the appropriate survey residual by using ratios
derived from decennial census data. The disaggregation
procedure is used to estimate employment in selected in
dustries for about 100 occupations. The proportion of
total 1978 employment estimated through the procedure
was less than 4 percent.
The preliminary matrix developed through the pro
cedures indicated above is reviewed in detail. The focus
of the review is on the estimates generated through the
disaggregation procedure. These are updated when the
preliminary data are believed to be in error. Analytical
judgment is used to make the updates.
Sourness ©f Data
Data used to develop industry-occupational matrices
come from a variety of sources. Since 1980, the major
source of occupational data has been the Occupational
Employment Statistics (OES) survey.1 (See chapter 3.)
The o e s survey collects data from employers on the oc
cupational distribution of workers in all nonagricultural
industries, except private households. Each industry is
surveyed every 3 years.
The occupational distribution of wage and salary
workers in agriculture and private households, not
covered by the o e s survey, is derived from the Current
Population Survey (CPS) (chapter 1). Data on selfemployed and unpaid family workers in each occupa
tion also come from the c p s . The industry distribution
of wage and salary employment is obtained through the
BLS Current Employment Statistics (CES) program
(chapter 2).
Cyrrent-year matrix
Separate estimates of current employment are
developed for wage and salary workers in o e s survey in
dustries, for wage and salary workers in agriculture and
private households, and for self-employed and unpaid
family workers. Data on wage and salary worker
employment are prepared by detailed occupation for
Wage and salary workers in agriculture and private
households. Total wage and salary worker employment
in agriculture and in private households is developed
from c p s data, although these are not strictly com
parable with c e s and o e s data.2
1 The 1978 and 1980 matrices, developed in 1980 and 1981, were the
first to be based on o e s survey data. Prior to 1980, sufficient o e s
survey data were not available to develop national estimates, and na
tional matrices were based primarily on the decennial census modified
by more current data from the Current Population Survey ( c p s ) . The
primary source was changed from census/cps data to o e s survey data
because o e s data, collected from employers according to specific oc
cupational definitions, are believed to be more accurate than c p s data,
derived from a survey of households.
2
In the c p s , each person is counted only once,in his or her primary
job; in the c e s and o e s , a person is counted in all jobs he or she holds.
Also, c p s and o e s data may include workers younger than 16.
Workers on unpaid absences are counted in the c p s but excluded from
the c e s .
142
These trends are extrapolated to the target year to
develop preliminary projected staffing patterns. When
an occupation is added, deleted, or changed in defini
tion from one o e s survey to the next, extrapolated
trends are not developed; the current-year ratios for
those occupations are held constant in the preliminary
projected matrix.
The projected ratios in each industry are applied to
projected industry employment totals for wage and
salary workers from the Bureau’s economic model (see
chapter 19) to derive preliminary target-year occupa
tional projections. These projections are analyzed in
detail based on studies of occupations and industries
conducted during preparation of the Occupational
Outlook Handbook. Factors considered include likely
changes in production methods, technological changes
which would affect the occupational mix, changes in the
product mix of industries, changes in the average size of
establishments in industries, and other factors affecting
specific occupations.
In addition, some occupations are projected in
dependently of the matrix based on the relationship of
the occupation to more closely associated variables. For
example, projections of elementary and secondary
school teachers are based on estimates of the school-age
population and pupil-teacher ratios. Projections
developed in this manner are placed in the matrix and
adjustments in the staffing patterns for other occupa
tions are made when necessary.
A review of the pattern is then made to assure that the
staffing pattern in each industry adds to 100 percent.
The resulting ratios are applied again to total projected
employment of wage and salary workers in each in
dustry to develop the final occupational projections of
wage and salary workers in o e s survey industries.
The occupational distributions of wage and salary
workers in the agriculture and private household in
dustries are based on data from the last census modified
by subsequent trends based on c p s data for large oc
cupations in these industries. The census/CPS employ
ment data are distributed into the detailed occupations
in the matrix.3 In this procedure, c p s data are generally
used as control totals, and the distribution is based on
established relationships between the census and o e s oc
cupational classifications. Many analytical judgments
are necessary to establish relationships for many oc
cupations because a perfect match between o e s and c p s
occupations is not always possible.
Self-employed and unpaid family workers. Estimates of
self-employed and unpaid family workers by occupation
are based on annual averages from the c p s , since no
alternative data series exist. Similar to the procedure
used for wage and salary workers in agriculture and
private households, the employment data in the detailed
census occupations are distributed to the detailed oc
cupations in the matrix. In general, c p s data are used as
control totals, and the distribution is based largely on
the distribution of wage and salary employment in o e s
data unless other data are available or judgment derived
from analyses indicates that this procedure would result
in incorrect data. For example, certain jobs found only
in government (such as health inspector) often fall into a
broader c p s category (inspector) which contains selfemployed and unpaid family workers. In such cases, the
distribution is not based on the wage and salary worker
distribution.
Data for self-employed and unpaid family workers
are developed only at the all-industry level because of
the unreliability of these data at the detailed industry
level.
Wage and salary workers in agriculture and private
households. For agriculture and private households,
past trends in occupational distribution are developed
based on data in the last decennial census and Current
Population Surveys conducted during subsequent years.
The census-based occupational distribution is converted
to the OES survey distribution in the manner described
above for the current-year matrix.
The projected ratios are then applied to the targetyear industry employment projections. The resulting
employment and ratios are reviewed in detail; changes
in patterns that result from this review are incorporated
into the final matrix.
Projections
The basic procedure for projecting occupational
employment is to develop data on past trends in the staf
fing patterns of industries and to extend these trends to
the target year of the projections. These preliminary
projections of the ratios are then reviewed in detail for
consistency with knowledge about technological change
and other factors likely to affect the occupational com
position of industries. Based on this review, changes are
made in the ratios and each industry is checked to see
that the ratios add to 100 percent. Finally, the projected
ratios are applied to projected industry employment
totals.
Self-employed and unpaid family workers. To develop
the projections, the percent distributions of selfemployed and unpaid family workers by occupation
from census/c p s data are extrapolated to the target year
and adjusted to add to 100 percent. A distribution of
these proportions is made to o e s survey occupations
Wage and salary workers in o e s survey industries. To
project staffing patterns, data are compiled from all
previous surveys of each industry to establish a trend.
3 There were 1,678 detailed occupations in the 1978 matrix.
143
Uses and Limitations
based on the distribution of current-year data. These
distributions are reviewed and changes made where
deemed appropriate. The resulting distribution is ap
plied to projected totals for self-employed and unpaid
family workers developed through the Bureau’s
economic model. The resulting projected employment
totals are reviewed for consistency with information
developed in the course of other occupational research,
and changes are made where necessary.
Presentation
A current-year and projected-year matrix are
developed on a 2-year cycle which coincides with the cy
cle used by the Bureau to develop economic, industry,
and occupational projections. Summary data from the
matrix are published in the Occupational Outlook
Handbook and in other Bureau publications.
Because of the large size of the latest set of
matrices—for 1980 and 1990—which include about
1,600 occupations and 378 industries, they have not
been published as a Bureau bulletin. However, data for
689 detailed occupations and 378 detailed industries are
available on computer tape which may be obtained at
cost from the Bureau.4 In general, only occupations
with 5,000 or more workers are included on the tape.
Hard copy of the data on the tape is available through
the National Technical Information Service.
4
Contact the Division of Occupational Outlook, Bureau of Labor
Statistics, U.S. Department of Labor, Washington, D.C. 20212, for
details on how to purchase this tape.
The industry-occupational matrix provides a com
prehensive set of data on the distribution of occuptional
employment by industry and enables comparison of the
occupational structure of industries. Other uses include
studies of the changing utilization of workers by in
dustry over time, analyses of occupational skill re
quirements in new and emerging industries, and market
research.
The industry-occupational matrix also is used in
studies which measure the occupational effects of
changes in the level of expenditures by the Federal
Government for specific programs. The national matrix
is also used by State employment security agencies to
develop estimates of current and projected employment
for States and areas within States.
Because the matrix is based on information obtained
from the o e s survey and the CPS, it is subject to the
response and sampling limitations typical of surveys.
(See the sections on limitations of these surveys in
chapters 1 and 3.) Further errors result from some of the
necessary analytical adjustments in combining data
from the two surveys and in estimating employment for
detailed occupations not included in the OES survey
questionnaire. The matrix data, therefore, indicate only
the general level and position occupations hold in rela
tion to other occupations within each industry. Conse
quently, the estimates should not be viewed as precise
measurements. In general, the smaller the occupational
estimates, the less the reliability.
Technical References
Bureau Of Labor Statistics
The BLS Economic Growth Model System Used fo r Projections to 1990, Bulletin 2112, 1982.
The N ation al Industry-O ccupational M atrix,
1978, and Projected 1990, Bulletin 2086, 1981.
1970,
144
The National OES-Survey-Based Industry-Occupational
Employment Matrix, 1978 and 1990, scheduled for release in 1982 through U.S. Department of Commerce,
National Technical Information Service.
Chapter 21. © eeupational
O utlook
The major objective of the occupational outlook pro
gram is to provide information on future employment
opportunities by occupation for use by counselors,
educators, and others helping young people choose a
field of work and for local and national officials who
plan education and training programs. Analyses of oc
cupations include information on the nature of the
work, employment, education and training re
quirements, the job outlook for about 10 years ahead,
earnings, and related occupations.
Sources ©f Data
Many sources are used to develop occupational infor
mation. The basic statistics on current employment are
derived from the Bureau’s Occupational Employment
Statistics surveys, which provide data by occupation for
wage and salary workers in nonagricultural industries,
except for private household workers. (See chapter 3.)
Employment data for workers in agriculture and private
households and for self-employed and unpaid family
workers are derived from the Current Population
Survey. (See chapter 1.) Employment data by industry
are derived from the Bureau’s Current Employment
Statistics program (chapter 2). The occupational
distribution of employment within industries—industry
staffing patterns—is available through the Bureau’s
industry-occupational matrix (chapter 20).
Analyses of past and projected changes in employ
ment make use of statistics on output, hours of work,
and output per worker hour from bls studies of produc
tivity and technological developments. Information
from the Office of Personnel Management is used to
study trends in employment of Federal Government
workers, and data compiled by Federal regulatory agen
cies, such as the Federal Aviation Administration and
the Interstate Commerce Commission, are used to study
employment trends in activities associated with those
agencies. Data are also obtained from unions, industry
and trade associations, and professional societies.
Analyses of past and probable future supply of
workers use still other sources of information. The Na
tional Center for Education Statistics provides data on
graduates from high school, junior or community col
leges, vocational education programs, and 4-year col
145
leges and universities. The Bureau of Apprenticeship
and Training of the U.S. Department of Labor supplies
information on apprenticeship completions, and the
Employment and Training Administration of the
department supplies data on enrollments and comple
tions in training programs supported by funds provided
under the Comprehensive Employment and Training
Act ( c e t a ). Also used are studies conducted by a variety
of private organizations on the supply and occupational
mobility of trained workers.
Earnings information is drawn primarily from bls
wage and earnings surveys. These are supplemented
with information from Federal regulatory agencies,
labor organizations, professional societies, and other
groups.
Information also is obtained from: (1) Interviews
with employers, union officials, and others closely
associated with an industry or occupation; (2) reports of
professional and trade associations and licensing agen
cies; and (3) labor publications, trade journals, annual
reports, and related materials.
methods
Projections of occupational employment are
developed as described in chapters 18-20. This broad,
systematic framework of projections develops projec
tions of the population, labor force, and national and
industry output and employment. For many occupa
tions, employment is projected on the basis of its rela
tionship to certain independent variables rather than on
proportional representation in each industry. Projec
tions for these occupations are developed by methods
tailored to fit the available data and the nature of the
occupation under study. For example, employment for
elementary school teachers is projected based on trends
in pupil-teacher ratios applied to projected school attend
ance. Projections developed through these independent
ly conducted analyses are then integrated with other oc
cupational data in the matrix.
Projections of changes in employment by occupation
provide only one part of the information needed on job
openings in the years ahead. In most occupations, the
majority of job opportunities arise either as a result of
the transfer of experienced workers to other occupa
tions or through retirements and deaths. To estimate the
number of such openings likely to arise in an occupa
tion, data have been developed on the proportions of
workers who generally leave an occupation during a
year, those who transfer to another occupation, or those
who leave the labor force to attend school, care for
family responsibilities, retire, or become too ill to work.
Replacement needs are affected by many factors, in
cluding the age and sex distribution of workers in an oc
cupation, the nature of the occupation, the career lad
der pattern, stability of work, factors related to the
desirability of an occupation such as wage rates and
working conditions, and specialized retirement pro
grams.
To appraise the future employment situation in an oc
cupation, estimates also must be made of the supply of
workers. Persons enter the job market from many
sources—schools and other training institutions,
transfers from other occupations, and reentries to the
labor force.
Analysis of supply is limited to those fields where the
supply is identifiable. Statistics on college enrollments
and graduations are the chief source of information on
the potential supply of workers in many professions and
occupations requiring extensive specialized education.
Data on the number of apprentices and graduates of
vocational and technical training programs provide
some information on new entrants into skilled trades.
However, in many occupations, workers learn on the
job through company training programs, and statistics
on such training activities are not available.
Not all persons who complete formal training or
education in a particular field enter that field. As a
result, special surveys are used to provide additional in
formation on the actual supply of workers from a train
ing program or a field of study. These include studies of
job placements of college graduates.
Presentation
The Occupational Outlook Handbook is the major
publication of the occupational outlook program.
Oriented toward career guidance, the Handbook is a
basic reference source, published every other year,
which includes comprehensive and nontechnical job in
formation on approximately 250 occupations covering
the entire spectrum of white-collar, blue-collar, and ser
vice occupations. A reprint series provides individual
statements from the Handbook.
The Occupational Outlook Quarterly provides cur
rent occupational and job information between editions
of the Handbook, together with the most recent infor
mation available on earnings, training requirements,
and other related topics.
Occupational Projections and Training Data,
published every 2 years, presents detailed statistics on
employment, job openings, and education and training
completions for many occupations.
In addition, technical and detailed studies are
published on specific occupations and industries in
order to furnish information to employment experts,
educational planners, personnel departments, and
others interested in the more technical aspects of the Na
tion’s future employment needs. These have covered
such topics as the demand for and supply of scientists
and engineers and workers in computer occupations,
and the employment effects of Government programs
for mass transit, pollution abatement, and highways.
Teoteiea!
Bureau of Labor Statisfies
Occupational Outlook Handbook, 1982-83 edition,
Bulletin 2200, 1982.
The BLS Economic Growth Model System Used fo r
Projections to 1990, Bulletin 2112.
Occupational Projections and Training Data, 1980 edition,
Bulletin 2052, 1980. The 1982 edition, Bulletin 2202,
is scheduled to be released in late 1982.
Measuring Labor Force Movements: A New Approach,
Report 581, 1980.
146
Appendix A. Seasonal Adjustment
Methodology at BLS
An economic time series may be affected by regular
intrayearly (seasonal) movements which result from
climatic conditions, model changeovers, vacation prac
tices, 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 variation from year to year and are subject to
changes in pattern over time. Because these intrayearly
patterns are combined with the underlying growth or
decline and cyclical movements of the series (trendcycle) and also random irregularities, it is difficult to
estimate the pattern 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 per
sonal 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 ma
jor forward step, as it made possible the uniform ap
plication 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 additional data became available and seasonal
factors were generated from the lengthened series, the
new factors sometimes differed markedly from the cor
responding factors based on the shorter series. This dif' Julius Shiskin, Electronic Computers and Business Indicators,
Occasional Paper No. 57 (New York, National Bureau of Economic
Research, 1957).
147
ference could affect any portion of the series. It was dif
ficult to accept a process by which the addition of recent
information could affect significantly the seasonal fac
tors 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 revi
sions 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 Fac
tor Method,” was further refined, with the final version
being produced 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,”
better known as simply X -ll. The X -ll provided some
useful analytical measures along with many more op
tions than the b l s method. Taking advantage of the
X - ll’s additional flexibility, b l s began making increas
ing use of the X-l 1 method in the early 1970’s, especial
ly for seasonal adjustment of the labor force data based
on the household data. Later in the 1970’s, Statistics
Canada, the Canadian national statistical agency,
developed an extension of the X -ll 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 In
tegrated Moving 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 limita
tions of the moving average techniques at the ends of
series. After extensive testing and research showed that
use of X-l 1 a r i m a would help to further minimize revi
sions in factors for recent periods, b l s began using the
X -ll a r i m a procedure in 1980 for most of its official
seasonal adjustment.
The standard practice at b l s for current seasonal ad
justment 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 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 accomplish
ed the objective of stabilizing the factors for the earlier
part of the series, and any further revisions would pro
duce 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 b l s method programs
were used. With the X -ll a r i m a procedure, the pro
jected year-ahead factors 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 fac
tors 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 unemploy
ment rate is adjusted directly. Similarly, the sum of
seasonally adjusted unemployment and seasonally ad
justed employment will not quite match the directly ad
justed 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 different seasonal factors may shift radical
ly. 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 com
ponents difficult.
Finally, it is worth noting that the availability of a
fast, efficient procedure for making seasonal adjust
ment computations 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 recommend
ed. And it should also be remembered that, whenever
one applies seasonal factors and analyzes seasonally ad
justed 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, subject to error. Because of the
complex nature of methods such as X -ll a r i m a , the
precise statistical properties of these errors are not yet
known.
Technical References
Organization for Economic Co-operation and Development.
Seasonal Adjustment on Electronic Computers. Paris,
1961.
The report and proceedings of an international con
ference held in November 1960. Describes experience in
the United States, Canada, and several European coun
ties. Includes theoretical sections relating to calendar
(trading day) variation and general properties o f moving
averages.
Barton, H .C ., Jr. “ Adjustment for Seasonal Variation,”
Federal Reserve Bulletin, June 1941.
The classic account of the f r b ratio-to-moving aver
age method, in which the analyst uses skilled judgment
to draw freehand curves at key stages of the pro
cedure.
Dagum, Estela Bee. The X - ll a r im a Seasonal Adjustment
Method. Ottawa, Statistics Canada, February 1980
(Statistics Canada Catalogue No. 12-564E).
Shiskin, Julius. Electronic Computers and Business In
dicators, 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 com
puter program for making seasonal adjustments.
Macaulay, Frederick R. The Smoothing o f Time Series, n b e r
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.
148
Technical References—Continued
Proceedings of a 1976 conference jointly sponsored
by the National Bureau of Economic Research and
the Bureau of the Census.
U .S. Department o f Commerce, Bureau o f the Census. The
X - ll Variant o f the Census Method II Seasonal Adjust
ment Program. Technical Paper No. 15, (1967 revi
sion).
U .S. Department of Labor, Bureau o f Labor Statistics. The
b l s Seasonal Factor Method, 1966.
U .S. Department of Commerce, Bureau o f the Census. Sea
sonal Analysis o f Economic Time Series, Economic
Research Report, ER-1, issued December 1978.
U .S. Department o f Labor, Bureau o f Labor Statistics. Em
ployment and Earnings, January 1980.
149
A p p end ix B. Industrial
C la ssifie a tio n
Miscellaneous electrical machinery, equipment, and
supplies.
Thousands of products and activities are distin
guished at the 4-digit level. For example, in Group 367,
nine industries are defined: Radio and television receiv
ing type electron tubes, except cathode ray; Cathode ray
television picture tubes; Transmitting, industrial, and
special purpose electron tubes; Semiconductors and
related devices; Electronic capacitors; Resistors, for
electric applications; Electronic coils, transformers and
other inductors; Connectors, for electronic applica
tions; and Electronic components, not elsewhere
classified.
The Bureau classifies reports from survey re
spondents, usually based on an establishment concept,
according to their prim ary product or activ
ity. The sic is used in the same way by the agencies sup
plying the Bureau with universe lists and benchmark
data. Hence, a high degree of orderliness and con
sistency is attained, which benefits not only the users of
all bls establishment statistics, but also the users of all
Government figures.1
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 ac
tivities 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 classification includes such combined activities;
(2) the employment 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
physically 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
B ls 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.
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; (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
employed, 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.
The broadest grouping divides the economy into 11 divi
sions: Agriculture, forestry, and fishing; mining; con
struction; manufacturing; transportation, communica
tions, 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 electronic machinery, equip
ment, and supplies.
The 3-digit level provides several hundred categories.
In the electrical machinery major group, the Sic pro
vides eight groups of industries: Electric transmission
and distribution equipment; Electrical industrial ap
paratus; Household appliances; Electric lighting and
wiring equipment; Radio and television receiving equip
ment, except communication types; Communication
equipment; Electronic components and accessories; and
1
Certain BLS programs may deviate from the SIC due to operational and con
ceptual reasons. These deviations are noted in the chapters on such programs.
150
corporations, partnerships, individual proprietors,
government agencies, joint ventures, etc.
This change from the 1967 edition removes “ Govern
ment” as an industry division, per se, and treats it as an
ownership characteristic. Government establishments,
therefore, are now classified by their primary economic
activity, rather than by type of owner. Because of this
change, it will be necessary to determine if a particular
statistic covers all of the industry or only the private sec
tor, and also if it includes all government or only public
administration.
The 1977 supplement to the 1972 sic manual con
tained new, deleted, and modified industries, titles,
definitions, and index items. This supplement reflected
the experience of Government agencies in using the 1972
sic manual. It serves primarily to make corrections to
the 1972 edition in areas crucial to statistical programs.
(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 ac
tivities are not ordinarily considered to be estab
lishments.
An establishment is not necessarily identical with an
enterprise or company, which may consist of one or
more establishments. Also, it is to be distinguished from
subunits, departments, or divisions. Supplemental inter
pretations of the definition of an establishment are in
cluded in the industry descriptions of the Standard In
dustrial Classification where appropriate.
In 1972, classification 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
151
Appendix C. Geographic
CSassifieationi
bls is represented on this committee along with other
organizations.5
The geographic detail for which bls publishes data
varies with the scope and size of the surveys it under
takes. In addition to national summaries, the Bureau
publishes data for four different geographic classifica
tions; individual States, the District of Columbia, and
outlying areas (Puerto Rico, Guam, and the Virgin
Islands); individual cities; M SA’s (M etropolitan
Statistical Areas); and labor areas and other classifica
tions developed to meet specific survey objectives.
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
standing” and not closely associated with other
metropolitan statistical areas. These areas are typically
surrounded by nonmetropolitan counties. Areas quali
fying for recognition 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
central counties, containing the area’s main population
concentration. A metropolitan statistical area may also
include outlying counties which have close economic
BLS regions
For survey estimates and indexes (including estimates
of the civilian labor force and unemployment, area
wage surveys,1 productivity surveys, and the Consumer
Price Index), bls generally uses a four-region classifica
tion system2 as follows:
Northeast: Connecticut, Maine, Massachusetts,
New Hampshire, New Jersey, New York, Penn
sylvania, Rhode Island, Vermont;
North Central: Illinois, Indiana, Iowa, Kansas,
Michigan, Minnesota, Missouri, Nebraska, North
Dakota, Ohio, South Dakota, Wisconsin;
1A lask a and H a w aii
South: A lab am a, A rk ansas, D elaw are, D istrict o f
West: Alaska, Arizona, California, Colorado,
Hawaii, Idaho, Montana, Nevada, New Mexico,
Oregon, Utah, Washington, Wyoming.
Data for the Producer Price Index and Industry Wage
Surveys are published for nine regions.3 4
Data published by State, e.g., annual employment
estimates, are aggregated by bls in accordance with the
10 Federal Administrative Regions established by the
Office of Management and Budget. A map of these
regions appears on the inside back cover of this Hand
book.
liVi©tr@p®litan Statistical Areas
Metropolitan Statistical Areas are designated by the
Office of Management and Budget through the Federal
Committee on Standard Metropolitan Statistical Areas.
are not covered in the area wage surveys.
2 This classification is the same as the four regions used by the Bureau o f the
Census.
3 The Producer Price Index is published for the nine regions: New England:
Connecticut, Maine, Massachusetts, New Hampshire, Rhode Island, Vermont;
Middle Atlantic: New Jersey, New York, Pennsylvania; South Atlantic:
Delaware, District o f Columbia, Florida, Georgia, Maryland, North Carolina,
South Carolina, Virginia, West Virginia; North East Central: Illinois, Indiana,
Michigan, Ohio, Wisconsin; South West Central: Arkansas, Louisiana,
Oklahoma, Texas; East South Central: Alabama, Kentucky, Mississippi, Ten
nessee; West North Central: Iowa, Kansas, Minnesota, Missouri, Nebraska,
North Dakota, South Dakota; Mountain: Arizona, Colorado, Idaho, Montana,
New Mexico, Nevada, Utah, Wyoming; Pacific: Alaska, California, Hawaii,
Oregon, Washington. These are the same nine divisions used in Bureau o f the
Census publications.
4 Industry wage surveys are published for the nine regions: New England:
Connecticut, Maine, Massachusetts, New Hampshire, Rhode Island, Vermont;
Middle Atlantic: New Jersey, New York, Pennsylvania; Border States:
Delaware, District o f Columbia, Kentucky, Maryland, Virginia, West Virginia;
Southeast: Alabama, Florida, Georgia, Mississippi, North Carolina, South
Carolina, Tennessee; Southwest: Arkansas, Louisiana, Oklahoma, Texas; Great
Lakes: Illinois, Indiana, Michigan, Minnesota, Ohio, Wisconsin; Middle West:
Iowa, Kansas, Missouri, Nebraska, North Dakota, South Dakota; Mountain:
Arizona, Colorado, Idaho, Montana, New Mexico, Utah, Wyoming; Pacific:
Alaska, California, Hawaii, Nevada, Oregon, Washington.
5 The other organizations include the Employment and Training Administra
tion o f the Department o f Labor, the Department o f Housing and Urban
Development, the Bureau of the Census, the Federal Reserve Board, the Depart
ment o f Agriculture, and the Department o f Transportation. The Committee is
chaired by a representative o f the Office o f Management and Budget.
Columbia, Florida, Georgia, Kentucky, Louisiana,
M aryland, M ississippi, N orth C aro lin a,
Oklahoma, South Carolina, Tennessee, Texas,
Virginia, West Virginia;
15:
and social relationships with the central counties. Such
counties must have a specified level of commuting to the
central counties and must meet certain standards regar
ding metropolitan character, such as population densi
ty. In New England, metropolitan statistical areas are
composed of cities and towns, rather than whole coun
ties. Under specified conditions, two adjacent areas may
be consolidated 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 in
clude up to three central city names, as well as the name
of each State into which the metropolitan sta
tistical area extends.
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
153
million population (Level A), primary metropolitan
statistical areas may be identified. These areas consist of
a large urbanized county, or cluster of counties, that
demonstrates very strong internal economic and social
links, in addition to close ties to neighboring areas.
When primary metropolitan statistical areas are defin
ed, the large area of which they are components is
designated a consolidated metropolitan statistical area.6
Labor areas
A labor area consists of a central city or cities and the
surrounding territory within commuting distance. It is
an economically integrated geographical unit within
which workers may readily change jobs without chang
ing their place of residence. Labor areas include one or
more entire counties, except in New England where
towns are considered the major geographical units.
Major labor areas usually have at least one central
city or Bureau of the Census urbanized area with a
population of 50,000 or more. In most instances, boun
daries of major labor areas coincide with those of
metropolitan statistical areas. Geographical boundaries
of all classified areas are listed in an Employment and
Training Administration publication, Directory o f Im
portant Labor Areas.
6 Federal Register, Vol. 45, N o. 2, Jan. 3, 1980, pp. 956-63.
Geographic areas currently used in selected BLS programs
4
Program and major publication
State
Labor areas
Other areas'
X
Region
MSA
X
X
X
X
Nation
X
X
X
Cities
Labor Force Statistics
Labor Force and Unemployment,
Employment and Earnings.............................
Nonagricultural Employment,
Employment and Earnings.............................
Local Area Unemployment Statistics,
Employment and Earnings.............................
Occupational Employment,
Occupational Employment— ......................
Employment and Wages,
Employment and Wages..................................
X
X
X
X
X
X
Prices and Living Conditions
Consumer Expenditures and Income,
Consumer Expenditures and Income............
X
o
International Prices,
U.S. Import Price Indexes;
U. S. Export Price Indexes...............................
X
Consumer Prices,
Consumer Price Index ......................................
X
X
X
o
o
Producer Prices,
Producer Prices and Price Indexes................
Wages and Industrial Relations
The Survey o f Professional, Administrative,
Technical, and Clerical Pay
The National Survey o f Professional,
Administrative, Technical, and
Clerical Pay .........................................................
X
Area Wage Surveys,
Area Wage Surveys, .........................................
X
o
X
o
X
o
Industry Wage Surveys
Industry Wage Surveys.....................................
X
X
Productivity and Technology
Construction Labor Requirements,
Construction Labor Requirement Studies...
Occupational Safety and Health
Occupational Safety
and
Health
Statistics,
Occupational Injuries and Illnesses...............
X
X
3 Nine-region classification designated in footnote 3.
4 Nine-region classification designated in footnote 4.
1 Defined according to survey objectives.
2 Four-region classification designated in “ BLS regions.”
15 4
* U. S. GOVERNMENT PRINTING O FFIC E : 1982 381-608/3876
x
Byreau of Labor Statistics
Regional Offices
Region i
1603 J F K Federal Building
Government Center
Boston, Mass. 02203
Phone: (617) 223-6761
Region IV
1371 Peachtree Street, N.E.
VSi and VSSl
911 Walnut Street
Atlanta, Ga. 30367
Phone: (404) 881-4418
Kansas City, Mo. 64106
Phone: (816) 374-2481
Region V
Region S
i
Suite 3400
1515 Broadway
New York, N.Y. 10036
Phone: (212) 944-3121
Region SS
S
Regions
SX
and
X
Federal Office Building
450 Golden Gate Avenue
Box 36017
230 S. Dearborn Street
San Francisco, Calif. 94102
Chicago, III. 60604
Phone: (415) 556-4678
9th Floor
Phone: (312) 353-1880
Region Vi
3535 Market Street
Second Floor
P.O. Box 13309
Philadelphia, Pa. 19101
555 Griffin Square Building
Dallas, Tex. 75202
Phone: (215) 596-1154
Phone: (214) 767-6971
Regions
@FedFRASER