View original document

The full text on this page is automatically extracted from the file linked above and may contain errors and inconsistencies.

The Consumer Price Index:
History and Techniques




Bulletin No. 1517
UNITED STATES DEPARTMENT OF LABOR
W. Willard Wirtz, Secretary
BUREAU OF LABOR STATISTICS
Arthur M. Ross, Commissioner

The Consumer Price Index:
History and Techniques




Bulletin No. 1517
UNITED STATES DEPARTMENT OF LABOR
W. Willard Wirtz, Secretary
BUREAU OF LABOR STATISTICS
Arthur M. Ross, Commissioner

For sale by the Superintendent of Documents, U.S. Government Printing Office,
Washington, D.C., 20402—Price 60 cents




Preface
This bulletin presents an historical summary covering the scope and
method of compiling the Consumer Price Index since its inception, a
rather detailed explanation of present techniques, and a description of
the 1964 comprehensive revision of the index* A bibliography of pub­
lications on methodology and analysis of price trends is included.
The bulletin was prepared by members of the staff of the Office of
Prices and Living Conditions, under the supervision of Doris P. Rothwell,
Chief, Division of Consumer Prices and Price Indexes and the general
direction of Arnold E. Chase, Assistant Commissioner of Prices and
Living Conditions. Several of the chapters were prepared primarily by
a single individual who is listed as the author, and much of the material
has been issued earlier as separate articles or releases. In some cases,
supplemental material has been added for this bulletin.




iii




Contents
Page

Chapter I.
Chapter II.
Chapter III.

Scope and coverage of U.S. Consumer Price Index prior to 1964
Major features of the revision program, 1959-64
Statistical and conceptual structure of the revised Consumer
Price Index
„
Chapter IV. Sampling aspects of the 1964 revision
Chapter V.
City sample selection, 1964 index
Chapter VI. Housing and expenditure surveys
Chapter VII. Weighting structure of the CPI, 1964
Chapter VIII. Outlet samples, 1964 index
Chapter IX. Pricing procedures
Chapter X.
Calculation procedures
Chapter XL Publication of data
Bibliography
Appendix tables:
Characteristics of wage earners and clerical workers whose expendi­
tures were used in the derivation of weights
II.
General description of U.S. Consumer Price Index coverage
III. Comparison of old and new series Consumer Price Index.
IV. Summary of samples from comprehensive urban housing unit survey
and urban survey of consumer expenditures, 1960-61
V.
Number of items in sampling frame and number of items priced by
expenditure class
VI. List of items priced for the revised Consumer Price Index as of
December 1963
VIL Revised CPI weight derivation procedures for major expenditure cate­
gories, individual expenditure class totals, certainty items, and
probability item totals within expenditure classes, by region and citysize stratum, families of two or more persons, 1960-61 CES
VIII. A. Derivation of population weights for revised CPI
B. Population weights for B cities before and after addition of six cities
IX. Consumer Price Index (new series), relative importance of major
groups, subgroups, and selected individual items, December 1963, and
comparison with old series
X.
Required number of reporters per replicated subsample by commodity
group (excluding food at home and items obtained from secondary
sources)
XL Size of independent food store sample by type of outlet by SMSA or
city
XII. Cities and pricing schedule for the revised Consumer Price Index

1
13
19
25
34
40
45
58
66
71
81
114

I.

84
84
85
86
87
88

92
95
96

97
98
99
100

Exhibits:
A. BLS 2549. Comprehensive housing unit survey listing form
B. Sampling frame for selection of the CPI item sample
C. BLS 2742. Survey of where goods are bought




101
104
112

V

The Consumer Price Index:
History and Techniques
Chapter I. Scope and Coverage of U. S. Consumer Price Index
Prior to 1964
General Review
The present Consumer Price Index (CPI) for
Urban Wage Earners and Clerical Workers,
formerly called the "Cost-of-Living Index," was
initiated at the time of World War I for use in
wage negotiations. There had been even earlier
studies of the cost of living for wage earners.
Statistical studies of prices and living con­
ditions in the United States were an outgrowth
of the tariff discussions of 1887 and 1888.
Studies of family expenditures, covering the
years 1888-90, were included in the Annual
Reports of the Commissioner of Labor for 1890
and 1891.
The Early Food Index
The systematic collection of retail prices be­
gan in conjunction with a study of consumer
expenditures and income in 1901. There had
been only one earlier major price investigation
in the United States—the imposing collection
of wholesale price data included in the "Aldrich
Reports" by the Senate Committee on Finance,
in 1892 and 1893.
Extreme fluctuations in the "purchasing
power" of money, resulting from great in­
creases in the output of precious metals and
from violent inflations during major wars, had
led to many small-scale collections of informa­
tion on price movements during the 18th and
19th centuries. These price data, recorded in
account books, ledgers, and trade journals, con­
tributed much to the understanding of economic
conditions, and stimulated demands from Amer­
ican economists for more organized informa­
tion. The long agricultural depression which
began about 1884, and the disastrous panic of




1893 made unemployment, strikes, and the de­
pletion of the gold reserve subjects of public
policy. The worldwide rise in prices beginning
in 1896 brought concepts of the "living wage"
and "high cost of living" to the public forum.
When the BLS began formal and regular col­
lection of price data, the need for such facts
generally was recognized, and their use in the
arbitration of wage disputes was accepted read­
ily by the public. For example, the report of
the President's Anthracite Coal Strike Commis­
sion used the average change in food prices in
the anthracite region as a basis for its wageincrease award in March 1903.
Initially, prices for about 30 foods were ob­
tained retroactively from account books and
records of about 800 firms in 171 cities through­
out the country. Prices were obtained monthly
from 1890 through 1903. As described in 1903,
the program "had for its objective the collection
of data which would show the extent of increase
or decrease in retail prices of staple articles of
food during the period and thus render it pos­
sible to determine, approximately at least, the
changes in cost of living in the several years
covered."* Indexes of retail food prices for the
United States as a whole and for five geo­
graphic regions were compiled for all years
back to 1890. The weighting pattern for these
indexes was derived from detailed data for
2,567 families in the study of expenditures,
covering 22,000 families, conducted by the Bu­
reau in 1901-02.2 The sample of cities was
t
Coat of Living and Retail Prices in the United States (18901903), (U.S. Bureau of Labor, Bulletin 54, 1904), p. 1129, and Cost
of Living and Retail Prices of Food (18th Annual Report of the
Commissioner of Labor, 1903), pp. 15-17.
2
Retail Prices of Food, 1890 to 1904 (U.S. Bureau of Labor, Bul­
letin 59,1905).

1

reduced to 151 in 1904 and to 68 for the years
1905-07. Food pricing was discontinued tempo­
rarily from 1907-11; when resumed, prices
were obtained retroactively in 1907 for 16 foods
in 39 cities. At the same time, arrangements
were made for a mail collection system, which
was continued until 1933. The list of priced
items gradually grew longer. Indexes for 39
individual cities became available beginning in
1913.
Between 1908 and 1920, much discussion and
difference of opinion arose concerning the sta­
tistical methods used to measure price change
and the commodity content of an index de­
signed to measure changes in the "cost of liv­
ing" or more specifically, the general consumer
price level.8 In the period of rapidly increasing
prices during and immediately following World
War I, it became increasingly clear that a meas­
ure of change in food prices was not an ade­
quate measure of the cost of living nor the
general price level. Arbitration boards and
commissions were considering many aspects of
living costs in rendering rulings and awards,
and their demands for data helped to shape the
scope, concept, and procedures for the index.
Cost-of-Living Index, 1913-35
The first BLS cost-of-living index grew out
of a decision by the Shipbuilding Labor Adjust­
ment Board during World War I. The Board
was a product of an agreement between Gov­
ernment officials and labor chiefs. Its function
was to adjust labor disputes so that national
defense production would not be interrupted.
In arriving at a "fair wage scale," 1 of the 2
factors considered by the Board was "adjusting
wages to the higher cost of living resulting
from the war." 4 In November 1917, the Board
determined that readjustment of wages in the
shipbuilding yards was warranted when there
had been a general and material increase in the
cost of living. Even earlier, by joint resolution
of Congress on December 20, 1916, the Depart­
ment of Labor was directed to inquire into the
cost of living of wage earners in the District of
•The term "cost of living" was used to describe the Bureau's
index until its name was changed following the controversy in the
World War H period over the index's validity as a measure of cost
of living. It has always been merely a measure of changes in prices
for goods and services purchased for family living.
4
"Labor and the War/' Monthly Review of the U.S. Bureau of
Labor Statistics, March 1918, pp. 67-76.

3



Columbia.5 An index for Washington was pub­
lished in mid-1919.
During 1918-19, in cooperation with the
Shipbuilding Labor Adjustment Board, the Bu­
reau investigated the cost of living in a number
of shipbuilding and other industrial centers. De­
tails of the cost of goods in the family market
basket were obtained for a year within the
period July 1917 to February 1919 from each
of 12,000 wage-earner families in 92 cities. In
addition, records of retail establishments pro­
vided prices for a large number of articles for
December of each year from 1914-17 in 19
cities, and for only December 1917 in 13 addi­
tional cities. Regular price collection was initi­
ated after 1917 in all 32 cities; prices were
collected 1 to 4 times a year for about 145 com­
modities and services. In 1919, the Bureau
began the publication of complete "cost-of-liv­
ing" indexes at semiannual intervals for 32
large shipbuilding and industrial centers, using
a weighting structure based on expenditures of
wage-earner and clerical-worker families in
1917-19.6 Indexes were estimated for the
United States back to 1913 based on wholesale
price movements. The reference base period of
the early CPI was 1913=100, later changed to
1923-25=100. In February 1921, regular, peri­
odic publication of the U.S. index in roughly its
present framework was established, although
there have been many changes in scope, cover­
age, frequency, and publication format since
then. Quarterly indexes were initiated in 1935.
Monthly indexes were inaugurated at the re­
quest of the National Defense Advisory Com­
mission beginning October 1940.7
Technological developments of the 1920's con­
tributed to the changing composition of the
goods and services that constituted the "neces­
sities" of life, and the growing body of em­
pirical knowledge of price behavior tended to
emphasize the significance of the "market
basket" used in the construction of consumer
price indexes. Surveys of expenditures by Fed­
eral employees in five cities in 1927-28 and in
Washington, D.C. in 1933, and of Ford Motor
Co. employees in Detroit in 1929, indicated the
extent of the changes which were taking place in
5
"Cost of Living in the District of Columbia," Monthly Labor
Review, June 1919, p. 117.
•Cost of Living in the United States (BLS Bulletin 357, 1924).
* "Changes in Cost of Living from Sept. 15 to Nov. 15, 1940,"
Monthly Labor Review, January 1941, p. 146.

the type of goods purchased and the manner of
living. However, the surveys' information on
family expenditures of wage-earner and cleri­
cal-worker groups in large cities throughout
the country was not complete enough to pro­
vide an adequate systematic basis for revision
of the then existing "market basket."
Consumers' Price Index, 1935-52
On March 15, 1933, the Secretary of Labor
requested that the American Statistical Asso­
ciation (ASA) appoint an Advisory Committee
to advise the Department on its general staistical program. The committee, working closely
with the Committee on Government Statistics
and Information Services of the ASA, paid par­
ticular attention to cost-of-living indexes. Act­
ing on the recommendations of the Advisory
Committee, the Bureau initiated steps leading
to a comprehensive revision of the index.
In 1934-36, the Bureau undertook a compre­
hensive survey of Money Disbursements of
Wage Earners and Clerical Workers, which cov­
ered 14,500 families of two or more persons in
42 cities with population over 50,000. In 1935,
while the surveys were underway, and pending
completion of the comprehensive index revision,
the Bureau incorporated methodological changes
in price collection and index calculation, modi­
fying both the weights used in combining group
indexes to obtain the "all-items" index, and the
population weights for combining cities.8 A
complete revision of the system of food-item
weights was inaugurated,9 with specific weights
based on city food expenditure patterns replac­
ing the regional weights formerly used. Also
the principle of imputation was adopted, i.e.,
ascribing to a sample item the price change for
groups of items presumed to have price move­
ments similar to the sample item. In this way,
imputation provides an estimate of a group's
price change which is not measured directly.
In addition to the weight revision, new com­
modities were added, and food indexes were con­
structed on the new basis back to March 1919.
The comprehensive revision of the index was
completed in 1940.10 New weights were intro8
"Revision of Index of Cost of Goods Purchased by Wage Earners
and Lower Salaried Workers," Monthly Labor Review, September
1935, pp. 819-837.
• Retail Prices of Food, 1923-36 (BLS Bulletin 635, 1938).
w
Changes in Cost of Living in Large Cities in the United States,
1913-41 (BLS Bulletin 699, 1941).




duced, and revised indexes were computed back
to 1935. Indexes were partially revised using
new group weights back to 1925. See appendix
table II. At the same time, the reference base
period was shifted to 1935-39=100 on advice of
the Central Statistical Board (predecessor of
the present Office of Statistical Standards of
the Bureau of the Budget).
During World War II, temporary adjust­
ments in pricing and weights for foods, fuels,
transportation, and other selected items were
made to take account of rationing and wartime
shortages.11 These adjustments were necessar­
ily imperfect. Proper allowance for wartime
quality deterioration could not be made, al­
though wartime specifications replaced prewar
standard qualities; black-market prices could
not be measured except to a limited degree;
comprehensive revision of weights for changed
family expenditure patterns could not be under­
taken. However, weights were reduced on ra­
tioned foods such as meat and sugar. Wartime
product specifications were priced and, follow­
ing institution of rent controls, collection of
rent data directly from tenants instead of from
management was instituted. Pricing of nonavailable items such as automobiles, washing
machines, and radios were dropped temporarily
and the index values assigned to those items
were adjusted each month based on the change
in the prices of all items. In 1946, when war­
time restrictions were eased, the prewar weight
patterns were restored, and long-time compari­
son of prices for these items before and after
the war was made. The difference in price
movement resulting from imputation to other
items and the actual long-time change was re­
flected in the index. This long-term method of
reintroducing the item and grade priced before
the war was adopted to adjust the level of the
index for any errors made in handling wartime
quality changes in the index calculation.
Prior to World War II, the weights and
"market basket" content used in a price index
continued to be a subject of discussion among
statisticians and economists. However, evidence
of the relatively greater influence that price
movements have on the course of the index
shifted the main area of controversy back to
u
"BLS Cost of Living Index in Wartime/' Monthly Labor Re*
view, July 1943, pp. 82-95, and Consumers' Prices in the United
States, mt-te
(BLS Bulletin 966, 1949).

3

pricing problems during World War II. Be­
tween 1925 and 1940 there had been consider­
able development of statistical methodology,
particularly in data collection and sampling
techniques. Statisticians began to center their
attention on the reliability of the basic data
used in compiling the index, the design of
schedules and questionnaires, the phrasing of
interview questions, training of data collection
agents, representativeness of samples, and
errors of response. When, in 1943-45, the index
was undergoing critical appraisal because of
special wartime conditions, it was natural for
economists and statisticians to raise questions
about the basic reliability of observations.
In 1946, a number of important changes were
made in the processing of food prices. A system
was instituted whereby separate average prices
for chain and independent stores were com­
puted, and these averages were combined using
fixed (internal) weights. Also, the food outlet
samples were revised, taking into account type
of store, sales volume, and location. Up to this
time, price changes for food had been computed
by comparing, in successive periods, prices from
a matched sample of outlets. The improvement
in outlet sample design, the system of internal
weights, and the large size of the store sample
made it possible for the Bureau to abandon the
more costly matched sample procedure. Instead,
the average of all price quotations obtained
was compared with the average in the previous
period. This system has been continued up to
the present time for food in large cities. How­
ever, the limited samples for the small cities
added to the index in 1953 necessitated a return
to the matched sample procedure for these
cities. A reduction of one-third in appropria­
tions for fiscal year 1948 necessitated changes
in the frequency of pricing for nonfood items
and in the number of items currently priced.
Postwar changes in consumption patterns of
wage-earner and clerical-worker families, re­
vealed in expenditure surveys conducted in a
few cities in 1947-49, indicated a serious need
for revision of the index weights used and the
market-basket items selected for periodic com­
parison.12 In 1949, the Congress authorized a
large-scale 3-year program for modernization of
the index. By this time, the postwar rise in
^"Revision of the Consumers* Price Index/' Monthly Labor Re~
view, July 1950, pp. 129-132.

4



prices, which followed elimination of price con­
trols in mid-1946, appeared to have run its
course; prices had begun to decline from their
postwar peaks, and the period 1951-52 was
expected to be characterized by relatively stable
economic conditions.
The reliability of data continued to be a cen­
tral statistical issue; however, precision of defi­
nitions and biases in the index also received in­
creasing attention. For instance, critics pointed
out that the BLS method of measuring rent
change was not able to reflect the difference be­
tween rents of newly constructed housing units
and those of comparable units already in the
rental market. Because of the impossibility of
judging comparability of quality for differ­
ent housing units, BLS based its rent index
(and continues to do so) on comparisons of
average rents for identical housing units in two
successive periods. New units can be introduced
into the sample from time to time, but the initial
rent for such units cannot be compared with
those for units already in the sample. Under
ordinary circumstances, new units rent at about
the same levels as do existing units of compara­
ble quality; thus the index can be said to reflect
rents for new units. However, during the period
of rent control in the 1940's when market
forces which tend to equate rents for new and
comparable existing units were not permitted
to function, new units tended to rent at a premi­
um over existing units. The fact that the tech­
nique of pricing identical units did not permit
the index to reflect the higher rents as price
changes, resulted in a downward bias during
this period. This bias was known as the "new
unit bias." 18
Interim Adjustment, 1950
The outbreak of hostilities in Korea was ac­
companied by sharp and diverse price increases
in the United States. These divergent price
changes, coupled with widespread use of the
index in wage escalation contracts, made adjust­
ment of the index weights to post-World War II
patterns extremely urgent. Pending completion
of the comprehensive revision in 1953, an in­
terim revision was carried out.14 Using data
» "Estimate of New Unit Bias in CPI Rent Index," Monthly Labor
Review, July 1949, pp. 44^-49.
** Interim Adjustment of Consumers9 Price Index (BLS Bulletin
1039,1952).

from expenditure surveys in seven cities con­
ducted in 1947-49, group weights were ad­
justed; 25 additional items were selected for
pricing; and 1950 population weights were intro­
duced into the index. To correct for the "new
unit bias," the rent index was raised 6.8 index
points and the all items index by 1.8 points. The
correction was spread over the period back to
1940. Both the "old series" index and the ad­
justed index were published simultaneously
through 1952, when the "old series" was dis­
continued.15

By the late 1950's, it became apparent that
the index weights should not go unrevised for
more than a decade. The Bureau asked for and
received authorization for a 5-year revision pro­
gram, which was begun in 1959. Surveys of
consumer expenditures for 1960 and 1961 were
conducted to provide the basic data for selecting
a new sample of goods and services and for com­
puting new weights for the index. The revised
index was first issued for January 1964. It is
described in detail in subsequent chapters.
Appraisals of the CPI

Consumer Price Index, 1953-63
A comprehensive revision of the index was
begun in 1949 and completed in 1953. Surveys
of consumer expenditures were conducted in
91 cities, the index concepts were reexamined
completely, and the index reference base was
changed from 1935-39 to 1947-49. The general
concept of the index as a measure of price
change for a fixed market basket of goods and
services was retained; but a major change in
CPI scope was made; the purchase of a home
was included in the weighting diagram. The
classification of goods and services into groups
and subgroups was revised, and indexes were
computed retroactively on the new base period
(1947-49) for the new major groups by recombining appropriate data for individual items
from the old index. The revision introduced a
new sample of 46 index cities out of the 91
cities in the CES, including for the first time
small urban places down to 2,500 population as
well as large cities; revised weights reflected
the 1950 spending pattern of wage-earner and
clerical-worker families adjusted to 1952; and
the list of items priced was expanded to include
new products such as television, frozen foods,
and items which had not been priced previously,
such as restaurant meals and owned homes.16
The new index was linked to the adjusted index
at December 1952 to form a continuous series.
In January 1962, the reference base period
was changed to 1957-59=100 (in keeping with
the recommendation of the Office of Statistical
Standards for all Government general-purpose
indexes). The Bureau continued to publish in­
dexes on the 1947-49 base as well.
** "Correction of New Unit Bias in the Rent Component of C M , "
Monthly Labor Review, April 1951, p p . 487-444.
16
Consumer Prices in the United States, Price Trends and Indexes,
1958-58 (BLS Bulletin 1256, 1959).




Throughout its history, the index has been
used extensively in the evaluation and adjust­
ment of wages and, for this reason, has been
subjected to public scrutiny repeatedly. The
Department of Labor's Conciliation Service, or­
ganized in 1913, made use of the Bureau's costof-living data in mediation and conciliation of
labor disputes. Change in the cost of living was
one of the main factors governing the recom­
mendations of the Railway Wage Commission,
the Shipbuilding Labor Adjustment Board, the
Bituminous and Anthracite Coal Commission,
and other boards and commissions active dur­
ing and following the First World War.17 Gen­
eral use of the index in wage negotiations was
expanded gradually, and when World War II
began, an increasing number of union-manage­
ment agreements specified automatic wage ad­
justments based on changes in the index. The
index came into great prominence in 1942 when
the National War Labor Board originated the
"Little Steel Formula" as a guiding policy in
the control of wage rates. This formula stabi­
lized wages at prevailing levels but permitted
increases up to 15 percent above January 1941
levels to compensate for the approximate rise
in the index from January 1941 to May 1942.
This use of the index led to protracted argu­
ments over its nature and accuracy.
At the end of the war, there was a decided
increase in the number of agreements which
related wage adjustments and the reopening of
wage contracts to changes in the CPI. In 1948,
one of the most important agreements using the
index was made between the General Motors
Corp. and the UAW-CIO; a number of simi­
lar agreements were quickly concluded in which
1T
See The Use of Cost-of~Living
(BLS Bulletin 869, 1925).

Figures in Wage

Adjustment

5

wages were tied directly to the CPI by specific
escalator clauses. Use of the index in this man­
ner expanded rapidly until by the end of 1957
the wages of over 4 million workers were ad­
justed automatically in accord with contract
provisions,18 and for millions more, movements
of the index were a consideration in collective
bargaining. Subsequently, with greater price
stability, there was some lessening in the direct
use of the index in escalator clauses, but in
1964, when the revised index was issued, there
were still 2 million workers or more whose
wages were tied to the index.
In addition to this use in wage adjustment,
the index is employed widely in other types of
contract-escalation provisions, such as those
concerning property rentals, service contracts,
annuities and pensions, welfare allowances, ali­
mony payments, etc.
The CPI is used extensively as a guide to pub­
lic economic policy decisions—administration of
wartime price and rent controls, establishment
of income and excise tax rates, and generally as
a measure of inflation in the determination of
various fiscal, public finance, international
trade, and monetary policies. As a measure of
change in purchasing power of the consumer
dollar, it is used in calculating changes in real
earnings and its component indexes are essential
statistical tools for deflation of the national
accounts.
As one of the most used of all statistical
measurements, it has perhaps been the subject
of more analysis and appraisal than any other
series. One of the first comprehensive reviews
of the index was made in 1933-34 by the Advis­
ory Committee appointed by the American Sta­
tistical Association, at the request of the Secre­
tary of Labor. This review led to improvement
of data collection and calculation methods, end­
ing with the comprehensive revision of the
index in 1940.
The increasing significance of the index in
wage and price stabilization during World War
II, and misunderstanding as to its intended
purpose (stemming largely from its title, "The
Cost of Living Index") led to discussions over
the correctness of the index as a measure of the
effect of price change on the wartime cost of
living. In the spring of 1943, the Secretary of
Labor, anticipating controversy over the accuM
Deferred Wage Increases and Escalator Clauses, 1958-53 (BLS
Report 285, 1963).

6



racy of the index arising out of the Little Steel
Formula, asked the ASA to appoint another
committee to review and appraise the index.
This committee, under the chairmanship of
Professor Fredrick C. Mills, concluded "First,
that within the limitations established for it,
the cost-of-living index provides a trustworthy
measure of change in the prices paid by con­
sumers for goods and services. Second, that
many of the difficulties and doubts which have
arisen concerning the index have their origins
in attempts to use it uncritically for purposes
to which it is not adapted."19
However, the intensity of the controversy
over the index increased and caused the Presi­
dent to ask the Chairman of the War Labor
Board, Mr. William H. Davis, to form a tri­
partite committee representing labor, business,
and Government, "to look into the question and
try to answer and make clear how the index
figure is arrived at, whether any changes should
be made in its component parts, or other im­
provements." The two labor members of the
committee, Mr. George Meany of the AFL and
Mr. R. J. Thomas of the CIO, presented their
own report in January 1944 alleging that the
index greatly understated the wartime rise in
the cost of living. The President's Committee
examined materials and testimony prepared by
the BLS, the "Mills Committee," the labor
unions, a technical committee under the chair­
manship of Dr. Wesley C. Mitchell, the Na­
tional Industrial Conference Board, and various
other groups and individuals. After this search­
ing review, the committee concluded that the
BLS index figures "constitute a competent
measure of price changes for goods customarily
purchased by families of wage earners and
lower salaried workers living in large cities;" 20
that much of the public misunderstanding was
created by use of the term "cost of living" when
referring to the index; that under the excep­
tional market conditions which exist in war­
time, allowance should be made for certain
increases in the cost of living due to quality
deterioration, disappearance of cheaper goods,
decrease of special sales, underreporting of
prices actually charged, and other temporary
disadvantages of the buyer in a seller's market
» "An Appraisal of the U.S. Bureau of Labor Statistics Cost-ofLiving Index," Journal of the American Statistical Association,
December 1943, p. 388.
20
Report of the President?* Committee on the Cost of Living,
(Office of Economic Stabilization, 1946), p. 14.

not measurable by the index; and that there
should be an additional upward adjustment if
the index were used to represent workers in
small as well as large cities.
Following this investigation and on the ad­
vice of the committee, the title of the index was
changed in September 1945 to "Consumers'
Price Index for Moderate Income Families in
Large Cities," to make clear that the index was
solely a measure of price change. This title
was later shortened to the "Consumer Price
Index." The committee recommended that an
allowance (estimated later at 5 percentage
points as of September 1945), be used in the
application of the index to wage stabilization
in the war economy, but that the allowance not
be incorporated into the official figures. From
June 1944 through December 1946, each
month's report on the CPI explained that "the
index does not show the full wartime effect on
the cost of living of such factors as lowered
quality, disappearance of low-priced goods, and
forced changes in housing and eating away
from home." By the end of 1946, a number of
these factors had disappeared or decreased in
importance and the statement was dropped from
BLS reports. Committee recommendations that
were adopted for the revised index included
coverage of small cities; pricing of restaurant
meals, children's clothing and owned homes,
and adjustment of the rent index for new unit
bias.
Following the outbreak of hostilities in
Korea, the Wage Stabilization Board promul­
gated a 10 percent wage formula analogous to
the Little Steel Formula of World War II. Thus,
the CPI again became a controlling factor in
wages, and criticism of the index again began
to be expressed. Partly to forestall repetition
of the 1943-44 debates, in the spring of 1951, a
subcommittee of the Committee on Education
and Labor of the House of Representatives un­
der the chairmanship of Representative Tom
Steed initiated an investigation of the Consumer
Price Index, for the following reasons: "(1)
The Consumers' Price Index has become an ex­
tremely important factor in maintaining har­
monious labor-management relations. (2) The
Consumers' Price Index affects, in numerous
ways, all the citizens of the country. (3) Since
the index is a statistic promulgated by a gov­
ernmental agency, it should be the best and
most accurate available. (4) Any governmental




statistics of such paramount importance as the
Consumer Price Index should be understood by
the public so that it will receive proper confi­
dence and respect." 21 With these purposes in
mind, the subcommittee heard many witnesses,
including officials of the BLS, members of the
American Statistical Association, and users of
the index from both labor and management.
These hearings clarified the meaning, construc­
tion, uses, and limitations of the index. On the
basis of the testimony presented, the subcom­
mittee concluded that "the Consumer Price
Index of the Bureau of Labor Statistics is an
excellent index and that it enjoys widespread
confidence among labor and management
groups and the general public."22 The subcom­
mittee expressed approval of the improvements
in index construction and coverage that were
underway at the time.
Wider application of the CPI in evaluating
economic conditions and the use of these evalua­
tions as a basis for vital Government and pri­
vate policy decision have focused attention on
the validity and accuracy of the index for these
purposes, as well as for wage adjustment. In
its report on "Employment, Growth and Price
Levels in 1959,23 the Joint Economic Committee
of the Congress included a brief evaluation of
the nature and limitations of the CPI. In regard
to its use as a measure of inflation, the report
pointed out that the index "does not presume
to represent all consuming units, though there
is no obvious reason to believe it understates or
overstates the movement of consumer prices to
other persons in the economy—self-employed,
nonurban, or extreme income groups." 24 This
report also mentions "several important defi­
ciencies, most of which are extremely difficult
to deal with by precise statistical techniques;" 25
for example, problems are faced in dealing
with changes in the quality of items included
in the index, and in determining when, which,
and by what methods entirely new items are to
be introduced into the index.
In July 1959, a Price Statistics Review Com­
mittee under the chairmanship of Dr. George
» Consumer? Price Index (U.S. House of Representatives, Special
Subcommittee of the Committee on Education and Labor, 80th Cong.,
2d sess., 1951), p. 1.
»Ibid., p. 81.
28
Employment, Growth and Price Levels (Staff Report, Joint
Economic Committee, 86th Cong., 2d sess., December 1959).
a* Ibid., p. 106.
*Ibid.

7

Stigler was created by the National Bureau of
Economic Research under contract with the
Office of Statistical Standards of the Bureau of
the Budget. The committee's function was to
review the basic concepts and procedures under­
lying data collection and calculation procedures
for the various price series published by the
Federal Government and to make suggestions
for their extension and improvement. The com­
mittee met at approximately monthly intervals
beginning in the fall of 1959. A substantial
amount of special work, including tabulation of
data and staff consultation, was performed by
BLS at the request of the committee. The com­
mittee's report, submitted in November I960,26
had an important impact upon the conduct of
the CPI revision project then underway. It
became the subject of hearings by the subcom­
mittee on Economic Statistics of the Joint Eco­
nomic Committee of the Congress in January
and May 1961.
The Review Committee's major recommenda­
tion was that the CPI should be moved toward
becoming an index of welfare or constant utility
rather than remaining a price index. The con­
cept of a constant utility index, which bases
price change upon a comparison of different
market baskets which are judged to be "equiva­
lent" in some objective sense is extremely com­
plex. It has not yet been formulated in opera­
tional terms. The committee made specific
recommendations for different treatment for a
very few components. Therefore, after serious
consideration, the Bureau decided to maintain
the basic historical orientation of the index as
a "constant market basket" index, but many of
the Review Committee's other recommendations
did influence the structure of the revised index
(first published in 1964 and referred to as "new
series"). Among recommendations incorporated
in the new series are more extensive use of
probability sampling, establishment of a re­
search division for developing methods of ad­
justing for quality changes and other purposes,
use of greater flexibility in specification pric­
ing, better documentation of procedures, and
inclusion of single persons in the index popula­
tion. Implementation of certain other recom26
See Government Price Statistics, Ft. I (Hearings before the
Subcommittee on Economic Statistics, Joint Economic Committee,
87th Cons., 1st sess., Jan. 24,1961, Pt. II, May 1, 2, 3, 4, and 5,1961).

8



mendations is dependent upon availability of
additional resources, not yet provided.
Population Representation
Historically, the CPI has been designed to
measure the price experience of a specific group
of the U.S. population—until 1964, families of
wage earners and clerical workers living in
urban places. In the revision completed in 1964,
the index coverage was extended to include
single workers living alone. Thus, beginning
in 1964, the index became more representative
of the total urban wage-earner and clericalworker population.
This representation has been accomplished by
deriving the index weights and item content
from expenditure data obtained from this more
inclusive group. The pattern of expenditures
derived from these data covers all consumer
goods and services purchased by the group in
a specified year. In effect the goods and services
purchased provides weights for the CPI market
basket, although quite often the specific set of
items selected for pricing has been described as
the "market basket."
Variation among families in the quantities
of goods and services purchased and the
amounts spent in any given year is very great.
Some families spend nothing for a specific item
while others have large expenditures for the
item, depending on their income, family com­
position and other characteristics, place of resi­
dence, living conditions, and the choices they
make in the market place. Because of this vari­
ability which is inherent in family spending,
the index has been representative of all urban
wage-earner and clerical-worker families, con­
sidered as a group, but not necessarily of any
one type of family or individual family included
in the group. Various procedures have been
followed at different times in the collection and
compilation of data to reduce or adjust for the
variability of the expenditure data and thus ob­
tain a more stable, balanced, and representa­
tive weight base for the index.
Until the 1953 revision, the desired results
were obtained primarily by eliminating from
the averages data for families who, because of
their composition or economic condition, might
be expected to contribute most to the total vari­
ability. Rather elaborate sets of "eligibility re­
quirements" were developed to eliminate such

families. Appendix table I shows the charac­
teristics of the families whose expenditures
were used in the derivation of index weights.
In addition to this method of controlling vari­
ability, a method of grouping data for cities
within regions was employed to average out
large random variations in expenditures for
clothing and housefurnishing items.
Restrictions, placed on the economic level and
other characteristics of families whose expendi­
tures were included in the basic weight data,
were not meant to restrict the representative­
ness of the index. Orientation of the index
towards the urban worker has always been the
Bureau's objective. Most of the restrictions
were aimed at eliminating aberrations that were
peculiar to the period of the expenditure survey
and which would not be expected to continue
as the economic status of workers advanced; for
example, the presence of boarders and lodgers,
families on relief and underemployed during the
great depression, and low economic status of
Negro workers.
On the other hand, the character of the urban
wage-earner family has changed over the years,
and this fact has been reflected in the index
structure. In the earlier periods, wage earners
and clerical workers could be characterized real­
istically as being of "low income." Clerical and
sales workers were identified as "lower salaried" workers, and the index was referred to
as one for "low and moderate-income" families.
There were renters primarily, living in the
more densely populated city centers, and includ­
ing relatively more of the older established
households and larger families. The large in­
crease in the size of the middle-income group
that took place in the last two decades and popu­
lation movement to the suburbs reflected to a
large degree the improving economic status of
the worker included in the CPI population.
Thus, the occupational classification of the
group as wage earners and clerical workers lost
much of its significance because of the similar­
ity in the manner of living of this group as com­
pared with the total urban population in the
middle-income range. Also, the continuing high
level of American economic prosperity reduced
to some degree the extreme variations in family
spending patterns that were of considerable
concern earlier. The 1953 revision of the index,
therefore, placed no eligibility restrictions on
the population whose expenditures were used




in weight derivation, except those which pre­
served the definitional base of the index—urban
families, whose heads were classified as wage
earners or clerical workers.27 The index, how­
ever, is no longer properly related to low and
middle income workers.
City Coverage
As indicated previously, the work of the Bu­
reau on consumer prices began with collecting
food prices.
The retail food index initiated in 1903 was
based on prices from a varying number of
cities from 1890 to 1911. In 1911,39 cities were
included. After 1912, additional cities were
added from time to time; by 1920 the sample
of cities for the food index had been increased
to 51. In 1943, the number was increased to
56 cities, which were retained for the food index
through 1952.
In 1918, the Bureau began to compile and
publish price indexes of all goods and services
for 48 shipbuilding and industrial centers and
for Washington, D.C. In 1919, an estimated
index for the United States based on a weighted
combination of 32 of these cities was published;
in 1940 and 1941, 2 other cities were added to
complete the list of 34 cities (plus 22 additional
cities for food). The 34 cities were included in
the index sample through 1952, and a separate
index was published for each city.
Up to the 1953 revision, the cities priced for
the index were not chosen by systematic sam­
pling methods to represent the total U.S. urban
population. They were selected primarily be­
cause of their individual importance in wage
negotiations. Some effort was made to obtain
regional representation for the food index, but
not through systematic sampling procedures.
Small cities (under 50,000 population) were not
represented. During World War II, special in­
dexes were calculated for 20 small cities and
12 cities where rapid expansion of war activities
had created emergency situations; and prices of
foods, fuels, and rents were obtained in 7 addi­
tional cities, for which indexes were estimated.
Data for these 39 places were never included
in the national average.
29
In the 1953 revision, an upper limit of $10,000 family income
was imposed as a means of insuring the correct occupational classifi­
cation. In the 1964 revision, because of better coding, by occupational
classification, no income restriction was needed*

9

In the 1953 revision, a new sample of 46
urbanized areas and small cities was selected
systematically28 to represent urban places of
all sizes down to 2,500 persons. The sample in­
cluded all of the 12 cities having populations of
1 million or more in 1950, and a representative
sample of three other strata (other large,
medium-size, and small cities) classified by
climate, population density, income level and,
for small cities, distance from a major market.
Through this sampling procedure, only 20 of
the 34 large cities formerly included in the
U.S. index were retained in the sample, and
city indexes were continued only for these 20.
The 1953 revision placed emphasis on the im­
portance of the national average. City indexes,
except those for very large cities, were con­
sidered to be byproducts of the U.S. index cal­
culation. No indexes have been published for
the 26 smaller cities.
The 50 city sample of the 1964 revision is
described later.
Frequency of Pricing
Not all cities have been priced monthly nor
have U.S. indexes always been calculated
monthly. Prior to September 1940, indexes
were calculated for the United States and in­
dividual cities at irregular intervals. Subse­
quently, a monthly U.S. all-items index was
estimated back to 1913 based on food prices
and estimates for other groups assuming an
even rate of change between pricing dates. In
the 1940 revision, which established the 34-city
index, monthly pricing was established, with
funds provided by the Office of Price Adminis­
tration. A limited list of items in all groups was
priced monthly in 21 cities, and the full list of
items was priced quarterly in all 34 cities on
the March, June, September and December
cycle. National indexes based on all 34 cities
(56 cities for food) were published monthly, but
the indexes for the interquarterly months were
published as preliminary indexes and occasion­
ally revised, if necessary, using straight-line
interpolation between quarterly pricing dates by
group.
The serious cut in budget appropriations for
fiscal year 1948 necessitated a reduction in the
"The selection was made by a Latin Square design. No addi­
tional cities were selected for food pricing. See "Selection of Cities
for Consumer Expenditures Survey, 1950/' Monthly Labor Review,
April 1951. pp. 480-488.

10



frequency of pricing in individual cities. Month­
ly food pricing was continued in all 56 cities
formerly priced, but pricing of fuels was re­
stricted to the 34-city sample rather than the 55
cities formerly priced for fuel. Monthly pric­
ing for other groups was confined to 10 cities,
rather than the 21 cities previously priced
monthly, and quarterly pricing of the remain­
ing 24 cities on a rotating cycle was instituted.
To make possible calculation of the national
index monthly, account was taken each month
of every city in the sample by making estimates
for unpriced cities. The first such estimates
were based on price movement in one of the
priced cities; however, later, estimates were
based on the average price movement in the 10
cities priced monthly. This procedure resulted
in errors of estimate in monthly price move­
ments but no long-term error, since estimates
for unpriced cities were automatically corrected
at the next pricing in each city.
The use of a rotating cycle as a device for
spreading pricing among more cities has been
continued. In the 1953 and 1964 revisions, only
the five largest cities were established for
monthly pricing of all items. Food and fuel
were priced monthly in all cities but other
groups in other cities were priced every 3
months or every 4 months. From 1953 to early
1963, unpriced cities were estimated between
pricings on the basis of the average change in
the five monthly cities. This procedure (as
would any explicit estimating procedure) some­
times overestimated price changes, necessitat­
ing corrections in the opposite direction when
the estimates were compared with actual data
in the next pricing period. Therefore, it was
discontinued. Instead, the latest available prices
are used, in effect holding prices constant from
the last pricing for all cities not actually priced.
This tends to introduce a slight lag in reflecting
price changes in the national index, but it avoids
the necessity for making corrections to com­
pensate for overestimates.
Item Coverage
The index is designed to measure the change
in prices of a fixed market basket of consump­
tion goods and services purchased by urban
wage earners and clerical workers. Historically,
the total index coverage has been for "current
consumption expenditures" including applicable

taxes, made in retail stores and service estab­
lishments by "index families." Other outlays,
such as for life insurance, income and other
personal taxes, savings and investments, have
been excluded since they do not involve the direct
purchase of goods and services in the market
place, or expenditures necessary for continued
ownership and use of goods purchased. Before
1953, the purchase of a home was considered to
be an investment outlay, and homeowner cost
items included in the index were limited to
insurance, real estate taxes, and interest pay­
ments. After that date, following lengthy dis­
cussions, the concept of home purchase for in­
dex purposes was changed. Expenditures for
purchase of a home were added to the index
coverage, and prices of homes have been treated
in index compilation in the same way as those
for consumer durable goods. Consumption of
goods not involving cash outlay by the "index
family," such as the value of homegrown food
or the share of insurance premiums paid by the
employer, is not included in the scope of the
index.
Coverage has not been restricted to "neces­
sities," although the idea of pricing essentials
was a consideration in the selections of items
to be priced in the early history of the index.
The sample of items priced for the index has
included goods and services whose price
changes, appropriately weighted in combina­
tion, provide an estimate of the average price
movement of all items or groups of items. In
the selection of sample items and allocation of
weights, since 1935, consideration has been
given both to their importance in family spend­
ing and to the representativeness of their price
trends. In general, in the 1953 revision, expen­
diture items reported in surveys of consumer
expenditures were grouped, within major cate­
gories of goods and services, into classes of
items which were fairly homogeneous in respect
to price movement. These classes were referred
to as "price families." The most important item
or items within each class were then automatic­
ally considered for inclusion, and other items
of somewhat less importance were added to the
sample if their price movement was unique.
The weight assigned to each priced item repre­
sented family expenditures for all the items it
represented.
A considerable amount of empirical research
was carried out as part of the 1953 revision




to measure the variance in price change for a
large number of commodities and services in
order to classify them into "price families."
However, it was never possible, within the re­
sources available, to collect a completely ade­
quate body of price data for this purpose. Item
classification was, therefore, based to some de­
gree on the Bureau's price analysts' knowledge
about physical characteristics and function of
a commodity, marketing and distribution prac­
tices, and other factors assumed to be related
to price change. The number of items selected
for pricing depended on the sample size re­
quired to obtain an acceptable estimate of the
average price change for each commodity group
and on the resources available for conducting
price collections.
This method of sample selection was not a
systematic sampling method by which items
would be chosen at random with chance of selec­
tion proportionate to their importance in family
spending. In view of the thousands of different
items of all qualities, brands, sizes, etc., pur­
chased by workers, a completely random selec­
tion from a clearly defined universe of items is
not possible. However, limited probability sam­
pling was introduced in the 1964 revision.
Until the 1964 revision the particular quality
or qualities of sample items to be priced were
determined through examination of prices paid
by families, as reported in expenditure surveys.
Price ranges within which the frequency of pur­
chases were greatest were selected to identify
the appropriate qualities for the index. These
qualities were translated into specifications de­
scribing this quality or quality range through
consultation with retail dealers, manufacturers,
trade associations, and other informative
sources.
Changes in the sample of items priced for the
index over the years reflect availability of re­
sources, changes that have occurred in patterns
of family expenditures, or improvements made
in the representativeness of the items priced.29
The number of items priced has varied greatly,
but exact counts are difficult because of the mul­
tiplicity of methods of handling quality varia­
tions and special cases over the years. The
food price index compiled in 1903 was a modest
beginning; prices were obtained for 30 basic
» BLS Bulletin 699, op. cit., and BLS Bulletin 1256, op. cit., eontain detailed lists of items set up for pricing in the 1940 and 1953
revisions.

li

food items. When pricing was resumed in 1911,
back prices were obtained for only 16 foods, but
the sample was gradually increased again to 30
items by 1918. In 1919, when the Bureau began
to compile indexes for all goods and services, the
sample included 42 foods. Most of the expan­
sion was in the fresh fruit and vegetables
group which was represented in the earlier
indexes only by white potatoes. The inclusion
of fruits and vegetables reflected the increasing
importance of this group and was the first at­
tempt to include difficult-to-price seasonal items.
As many as 84 food items were priced on an
experimental basis in 1935-39, but many of
these items were dropped in the 1940 revision
after studies of comparative price trends were
made. Since then, most changes in the food item
sample other than the complete resampling of
the 1964 revision have been made to introduce
new products and specifications and to improve
the sample representation for subgroups of
foods as they assume increased importance in
family spending.
Between 1918 and 1963 the total number of
items priced for the total CPI was increased
from 166 to 325 and further increased to 400
separate specifications in the 1964 revision. In
the early 1940's as many as 300 items were
priced, but a substantial cut, mainly in the num­
ber of different qualities, was necessitated in
fiscal year 1948 by the reduction in resources.
The samples of apparel, housefurnishings, and
other items show the same process of gradual
revision as the food sample to include items
which gained in importance, to eliminate those
becoming obsolete, and generally to increase
their representativeness. Pajamas replaced
nightshirts; oxfords were substituted for high
shoes; men's separate collars were dropped;
drycleaning services were added; and modern
synthetic fabrics were introduced. Radios, vacu­
um cleaners, refrigerators, and other electrical

12



appliances were added in 1940, as were the
automobile, gasoline, and other automotive
products. Television sets, toys and modern
drugs were introduced in 1950 along with many
other important consumer goods. A number of
new consumer services were added in the com­
plete resampling of 1964, including some re­
quiring unique procedures for pricing.
Component Indexes
The classification of items into groups and
subgroups has been revised several times. Be­
tween 1919 and 1935, indexes for only six major
categories—food, rent, housefurnishings, fuel
and light, apparel, and miscellaneous,—and the
"all items" index were published. With the
increasing size of the item sample, the Bureau
was able to develop more detailed summaries for
publication. Indexes for subgroups of foods
were added in 1935 and extended back where
possible. After 1952, a major change in classi­
fication was introduced. The former miscellan­
eous group was subdivided. Indexes were pre­
pared for eight major categories—food, hous­
ing, apparel, transportation, medical care, per­
sonal care, reading and recreation, and other
goods and services—and 18 subgroups of goods
and services extended back at least to 1947. A
few years later many special group indexes,
including separate indexes for commodities and
services, were compiled. Further changes were
made in the 1964 revision.
Following World War II, in an effort to pro­
vide the maximum amount of information to
index users, U.S. indexes were calculated and
published at quarterly intervals back to 1935
for most of the individual nonfood items priced
for the index. Until 1953, they were based on
all 34 cities in the index; from 1953 to 1963,
on the 19 cities priced on the March, June, Sep­
tember, December cycle.

Chapter II. Major Features of the Revision Program, 1959-1964
The revision of the Consumer Price Index
(CPI), completed in January 1964, was the
third comprehensive revision80 since the index
was initiated in 1918. The revised index was
the culmination of a 5-year program, carried
out during the years 1959-63, and estimated to
have cost over $6 million. This chapter briefly
summarizes the various aspects of the revision
which are discussed in greater detail in sub­
sequent chapters.
Need for Revision
Minor adjustments necessary to insure that
the index reflects price changes on current
market transactions are made continually in
the course of ordinary index maintenance.
These include the introduction of revised spe­
cifications which describe new varieties of goods
and services, and the introduction of revised
samples of stores and establishments which re­
port retail prices to the Bureau. Occasionally,
adjustments are made in the relative weights
assigned to two or more specifications of a sin­
gle item (for example, full-fashioned and seam­
less hose) when their relative importance in
current sales shifts significantly, but the total
weight of the item usually is held constant be­
tween major revisions.
Item weights and the sample of items nor­
mally have been revised only upon reexamina­
tion of the entire structure of the index. Such
a major undertaking requires a comprehensive
survey of consumer expenditures and surveys
of price trends. No time schedule for such
major revisions had been established previously,
although the Bureau would have preferred,
under normal or near-normal conditions, to fol­
low a regular, predetermined schedule. With
the 1964 revision, the Bureau announced, and
hopes to maintain, a schedule of revision at
*• The first comprehensive revision was completed in 1940, retro­
actively to 1935, the second in January 1953, and the third in Janu­
ary 1964. See chapter I for a description of the various comprehensive
and partial revisions of the index throughout its history.




approximately 10-year intervals,31 unless cir­
cumstances indicate a need for an earlier re­
vision.
During the period that followed the 1953 re­
vision, dramatic changes occurred in the com­
position of the urban population, in the kinds
of consumer goods and services available, in net
incomes of urban workers, and in methods of
distribution and marketing techniques, all of
which alter the pattern of consumer expendi­
tures. Although the BLS had not conducted ex­
penditure surveys during this period, informa­
tion from other sources provided clues to the
probable obsolescence of index weights and
price patterns early in the 1960's.
The population had mushroomed, but, more
importantly, it presented a markedly different
composition than in 1950. The proportion of
persons at each end of the life cycle had in­
creased. Major changes had occurred in its geo­
graphic distribution. About 1 out of every 5
family units was moving each year, many to
the South and West, which were becoming
more industralized, from farm to city, from the
central city to the suburbs, and to peripheral
areas soon to become urbanized. Based upon
various projections of the U.S. population, it
appears that by 1975, 75 percent of the total
population may reside in metropolitan areas.
These geographic shifts naturally altered the
average distribution of expenditures by the
urban families among the different components.
Personal incomes had moved upward since
1950—about 36 percent between 1950 and 1956
—and a great part of the rise was reflected in
increased real income. Between 1952 and 1956,
incomes increased about 22 percent and con­
sumer prices about 2% percent, indicating a
substantial rise in real income of workers'
families.
81
A tentative time schedule calls for the next comprehensive re­
vision hy 1975—an 11-year span—permitting more effective use of
tabulations from the 1970 Population Census.

13

Shifts in consumer spending patterns were
already apparent by about 1957. Trade sources
indicated upgrading in the purchasing of many
commodities; for example, new instead of used
cars, and more highly processed foods. Relative
expenditures for cereals and bakery products
appeared to be lower and for meat, poultry, and
fish, higher than in 1950; public transportation
was less important and private transportation
more important. Further extension of credit
on easy terms made the consumer less and less
willing to defer purchasing a home, major ap­
pliances, an automobile, and other large ticket
items. Also, the decline of price maintenance
laws and rise of the discount house had altered
retail distribution patterns. Many new prod­
ucts or qualities had come into being. These
ranged from deep freezers to new household
items of plastics. Greater use was being made
of frozen foods, and there were important
changes in housing, including a large number
of new units, and a continuing shift from rental
to owner occupancy. Particularly significant
was the increasing share of consumer services
in the economy as a whole.
The Bureau's 1957 proposal for a revision
program did not imply that a revision was
urgent, i.e., that these economic developments
had already seriously affected the representa­
tiveness of the samples of cities, outlets, and
items in the index or of the index weights. In
terms of a 4- or 5-year revision program, the
changes did indicate the potential danger of
serious effects on the index if the program was
not launched immediately. Updating also was
urged as part of an orderly program of revi­
sions designed to maintain the quality of the
index and public confidence in the index.
Summary of Major Features of Revised Index
The revision as carried out did not change the
basic index concepts. The national index still
measures average changes over time in prices
of goods and services bought by urban wage
earners and clerical workers. The same statis­
tical formula is still employed in the index cal­
culations, and the reference base period has not
been changed since the shift was made in 1962
to the 1957-59=100 base recommended for all
Government series.
14



The major changes made in the revised index
are:
1. More comprehensive coverage in that single workers
living alone are included, as well as families of wage
earners and clerical workers.
2. A new sample of metropolitan areas and smaller
urban places (hereafter referred to as city sample).
3. Extension of pricing to suburban areas.
4. A new market basket.
5. New samples of reporters, including many new
types needed for unique items sampled for the revised
index.
6. Weighting factors based on 1960-61 expenditure
patterns.
7. Increased use of probability sampling.
8. Establishment of a replication design in order to
measure sampling error in the index.
9. Moreflexibilityin specification pricing.
10. New policy regarding the publication of city
indexes.

Appendix table III compares the important fea­
tures of the old and new series indexes.
The conceptual considerations leading to the
operational decisions made in all aspects of the
program will be discussed in detail in subse­
quent chapters. The scheduling and timing of
the major activities are summarized in the fol­
lowing section.
Scheduling and Timing
The major activities of the revision program
had to be conducted partly in sequence in order
to dovetail the various aspects. The interde­
pendence of the operations presented difficulties
in timing, particularly in the final stages. The
major activities included:
Program Planning and Analysis
Comprehensive Housing Unit Surveys (CHUS)
Consumer Expenditure Surveys (CES)
Selection of City, Item, and Outlet Samples
Initiation of Pricing and Readjustment of Samples as
Needed
Derivation of Weights
Calculation of Revised Index
Addition of Six Large Cities to Original Program

The program for the first year encompassed
most of the preliminary planning and prepara­
tion activities related to basic decisions on con­
cept, coverage, and index methodology, and de­
velopment of survey procedures and materials,
including the selection of the city sample. The
first-year program also included the actual con­
duct of housing and expenditure surveys in Cin­
cinnati, Ohio, which had been selected as a lead
or pilot survey city for the revision project

before the index sample of cities was selected.
In fiscal 1961 and 1962, housing surveys were
conducted in 49 urban places and expenditure
surveys in 66 urban places, about one-half of
the work being scheduled each year. Also,
price surveys were conducted on an experi­
mental basis for analytical use in the selection
of items and outlets and for calculation of test
indexes.
In the fourth and fifth years, expenditure and
price data were tabulated, index weights de­
veloped, index procedures formalized, and test
indexes calculated. Official calculation and pub­
lication of old and new series indexes was car­
ried out concurrently during the first 6 months
of the calendar year 1964.
After plans for the CPI revision were under­
way, the Bureau of the Budget contracted with
the National Bureau of Economic Research for
a review of Government price statistics by the
Price Statistics Review Committee. (See chap­
ter L) Even though much of the BLS planning
work for the CPI revision was well advanced
by November 30, 1960, when the committee's
final report was submitted, in several respects
the committee's recommendations had an im­
portant impact on the revision, then in process.
In the last year of the revision program, the
BLS was given funds to conduct housing and
expenditure surveys and to make plans for ini­
tiating indexes for six additional large cities,32
which had not been included in the national
sample. This work was continued during 1964
and 1965, after the revision was completed. City
indexes were first published during the calendar
year 1965 and these cities were added to the
national index beginning January 1966.
Program Planning and Analysis
Although general plans for the revision were
formulated at the time the project proposal was
submitted to the Congress, including the sched­
uling of surveys and scope of the project, im­
plementation of the general policies, develop­
ment of new sampling techniques, selection of
samples of items and outlets, and determination
of the scheme for weight derivation were car­
ried on actively throughout the 5-year project.
83

Cincinnati* Houston, Kansas City, Milwaukee, Minneapolis, and
San Diego.




Fundamental questions of concept were dis­
cussed continually, practically up to the date
when the revision was completed.
Primarily, the program was designed to pro­
vide for revision of weights; for new samples
of cities, items, and outlets; and for improve­
ment of estimating and data collection proce­
dures and price change measurement. Because
of the intensive review of concepts of the index
in the 1953 revision program, the Bureau did
not anticipate a need for any major definitional
changes in basic concepts or coverage. With
the concurrence of both the Labor and the Busi­
ness Research Advisory Committees,33 it was
decided to extend the index coverage to include
wage earner and clerical workers living alone
and to compute two indexes, one for families
and the other for families and single workers
combined. (The index for families only was
discontinued after November 1964, when it be­
came apparent that both indexes moved alike.)
It was expected also that the index would con­
tinue to be a measure of price change for a
"constant market basket" of purchases made by
urban wage earner and clerical workers. This
decision prevailed even though the Price Sta­
tistics Review Committee had expressed a pref­
erence for an index of "constant satisfaction,"
in which comparisons would be made between
different market baskets judged to provide
equivalent satisfaction, by some means yet to
be determined.
The methodology for developing index
weights was considered early in the program,
with particular reference to the special proce­
dures adopted in the 1953 revision to adjust
observed expenditures for abnormalities of the
survey year, so as to approximate more stable
expenditure patterns. It was decided that the
objective of the weight derivation processes in
the current revision would be the best estimate
of observed expenditures as of the survey date.
Some averaging of cities and other adjustments
were anticipated in order to produce better
estimates based on the survey results, but no
attempt was to be made to estimate stable pat­
terns or project them to expected patterns for
ensuing years.
88
Price Subcommittees of the Business Research Advisory Council
and of the Labor Research Advisory Council to the Bureau of Labor
Statistics.

16

Comprehensive Housing Unit Surveys
The Comprehensive Housing Unit Surveys
(CHUS) were prerequisites for several major
phases of the program. They provided the sam­
pling frame for selecting addresses for subse­
quent expenditure surveys, samples of tenantoccupied units for measurement of rent change,
and samples of owner-occupied units for meas­
urement of property tax; they also, provided
other statistical information used in derivation
of housing weights. Thus, CHUS were needed
early in the program.
Obviously, the selection of the sample of cities
had to precede the housing surveys. This was
done in the first fiscal year by the GoodmanKish controlled selection technique. (See chap­
ter VI.)
Development of the survey questionnaire was
begun in fiscal 1959 and completed in time for
the pilot survey in the fall of 1959 in Cincinnati
(which did not fall in the revised city sample,
but was 1 of the 6 cities added to the national
index in January 1966). Only minor changes
were made in the questionnaire for the fullscale surveys, which were conducted in two
stages in the fall of 1960 and the fall of 1961.
All 12 of the largest cities and 10 other large
and medium-size cities were scheduled for sur­
veying in 1960 and the remaining 11 large and
medium-size cities and 8 D strata cities in 1961.
This schedule was geared to the needs of the
expenditure surveys to be carried on in the
spring of the following year for each city. Com­
prehensive housing surveys were not made prior
to the CES in the remaining 24 smallest cities,
since sample addresses for the CES could be
drawn more economically from Bureau of the
Census records. However, subsequently, hous­
ing surveys were conducted in eight additional
D cities for other purposes.
Surveys in the five additional large cities
added to the index in January 1966 along with
Cincinnati, already surveyed, were conducted in
the fall of 1963.
Consumer Expenditure Surveys
The consumer expenditure surveys provided
the basic data for the weighting system and the
selection of the sample of items to be priced for
the index. Since they covered a sample of the

ie



total urban population, they also served im­
portant nonindex purposes in the field of mar­
keting and economic analysis of consumer in­
comes, spending, and saving. Moreover, the
surveys for 1961 were conducted simultaneous­
ly with surveys for rural farm and rural nonfarm areas, in cooperation with the U.S.
Department of Agriculture, to make possible
national estimates of consumer expenditures.
This Bulletin discusses the use of the data for
index purposes only and does not attempt to
describe the countless analytical and research
uses made of the expenditure data.
The surveys, except for Cincinnati and An­
chorage, were done in two stages, in the spring
of 1961 and 1962, covering the calendar years
1960 or 1961. In the 12 largest cities, half the
sample of households was surveyed in each of
the 2 years. Each of the other cities was sur­
veyed completely in one or the other year. Sur­
veys in five additional large cities 84 were car­
ried out in the spring of 1964 covering the
year 1963.
Selection of Revised Samples
Two important decisions with regard to sam­
pling for the revised index were made early in
the program and before the report of the Price
Statistics Review Committee was completed.
The first decision was to exert every effort to
extend probability sampling on a greater scale
than was previously thought feasible. The other
decision was to design a system for estimating
the sampling error in the index. These two
decisions underlay the procedures developed.
The strong recommendations in the Review
Committee's report gave support to the Bu­
reau's efforts to make these improvements in
sampling procedures. Following extensive dis­
cussion, and with the continuing advice of mem­
bers of the Review Committee, a system of rep­
licated samples was devised for the purpose of
measuring sampling error*
The revision work included selecting revised
samples of cities, items, and reporters. The use
of probability methods in selection of the item
sample was entirely new. Moreover, it was a
key operation, which had to be completed before
planning for the outlet samples, developing spe­
cifications, designing the scheme for weight
84
The sixth city, Cincinnati, had "been the pilot city surveyed in
1959, No new survey was undertaken,

derivation, and setting up index worksheets,
manuals, and the like. Because of this, it was
not possible to wait for the complete tabulation
of expenditure data and it was necessary to
base the selection of items to be priced on data
from nine cities surveyed in 1960. Development
of the sampling frame and experimental selec­
tion of items by probability methods com­
menced early in 1961, but the final selection
was not made until September 1962.
Consideration of the problem of reporter sam­
ples began early in 1961. The general scheme
for selection of samples was developed, sample
sizes were determined, and basic listings from
the Bureau of Old-Age and Survivors Insurance
(BOASI) were in usable form by the summer
of 1962; actual selection was undertaken after
the list of items was finalized. This work con­
tinued almost to the end of the program.
Initiation of Pricing and Readjustment of
Samples
Initiation of pricing of the revised samples
in the field was a laborious operation. It was
done very gradually, beginning in July 1962
for rent in 14 cities that were common to old
and new indexes, extending to the most im­
portant items in a few cities and then to other
cities and finally to probability items, as the list
was finalized and specifications were developed.
Several pricings were necessary to fill in gaps
and to replace unproductive outlets. First pric­
ing of the entire list of items was substantially
completed by about May 1963. Many changes
in the outlet samples which had been selected
in Washington were found necessary in the
field.
In a number of respects, the pricing proce­
dures for the revised index were different from
those used previously. The need for introduc­
ing new procedures and also continuing the old
index simultaneously by old procedures proved
burdensome. The principal changes were the
pricing of items deviating from specification
where necessary, and pricing in two different
subsamples of outlets.
Price data collected for the first few periods
were reviewed thoroughly and numerous ad­
justments of item and of outlet samples and of
processing procedures were made as needed.




Derivation of Weights
Except for the original decision that the
objective of the weight derivation was the best
possible estimate of 1960-61 actual expendi­
tures, little work on weight derivation was pos­
sible until the item sample and replication de­
sign was formulated and the final tabulations
from the CES completed. Most of the work on
this phase was concentrated in 6 or 8 months
from the end of 1962 to the fall of 1963.
Calculation of Revised Index
Recurring calculation procedures for the re­
vised index are much more complex and time
consuming than for the old series because of
separate processing of the replicated subsam­
ples, and because of the inclusion of items devi­
ating from specification in the measurement of
price change. Very careful review of the tabu­
lating and processing procedures was necessary
to uncover unforeseen contingencies, to develop
adjusting techniques where necessary, and to
routinize editing procedures. Original plans op­
timistically called for 12 months of indexes
during 1963, by which procedures would be
tested, but late completion of weight derivation
and initiation of pricing in the field forced a
rather drastic curtailment in plans. Complete
city and U.S. indexes were calculated for sev­
eral months to make possible complete testing
of the new mechanics and to make sure that all
necessary instructions for clerical processing of
new samples by new procedures had been for­
mulated.
National price changes for major groups
were computed for several months in the sum­
mer of 1963, as part of the testing; however,
because so many estimating expedients were
required the results were not considered an ade­
quate test of the possible effect of the revision
on price trend data. Beginning in the fall,
major emphasis was placed on work necessary
to publish the January revised index on sched­
ule. Problems of linking the new samples to the
old index at December 1963 were particularly
complex and a thorough review of December—
January price changes of old and new series
was made city by city and group by group be­
fore the new January index was released on
March 3, 1964. Simultaneous calculation of
complete old and new series indexes was carried
17

on for January and the next 5 months of over­
lap calculation.
Addition of Six Large Cities
The extension of the Consumer Price Index
program to six additional large cities arose out
of discussions concerning the calculation and
publication of indexes for individual cities. As

18



a matter of policy, it was decided that indexes
would be published for all Standard Metropoli­
tan Statistical Areas having a population of 1
million or more in 1960. Housing and expendi­
ture surveys were conducted, outlet samples
selected, pricing initiated, and weights derived
by procedures similar to those for cities in the
original sample.

Chapter III. Statistical and Conceptual Structure of the
Revised Consumer Price Index*
Since the Consumer Price Index (CPI) was
established, during World War I, it has under­
gone three major and several partial revisions,
as described in chapter I. The recent revision,
effective with publication of the January 1964
index, is the first to be initiated by the Bureau
of Labor Statistics on the principle that review
and revision should be a regular part of the
index program. In contrast, previous major re­
visions were carried through at long intervals
and after drastic changes in the economy raised
questions about the validity of the index meas­
urement of price change. The Bureau hopes to
review and revise the index at approximately
10-year intervals. Of course, this does not rule
out revision at shorter intervals, when economic
changes have affected significantly the buying
patterns of wage earners.
Despite many changes and improvements in
statistical procedures, the revised CPI continues
to be what it has always been—a measure of
price change, and of price change only, in items
purchased by urban wage and clerical workers
for their own consumption. Major orientation
of the index is toward use in collective bargain­
ing and as a yardstick for measuring changes
in real income of workers. The purpose of the
CPI is still to measure the shifts in the purchas­
ing power of the consumer's dollar or—in the
other way it is often expressed—to measure
changes in his cost of living, insofar as living
costs are affected by price change.
Workers Covered
Expenditures by a cross section of wageearner and clerical-worker consumers living in
a representative selection of urban places pro­
vide the basis both for the selection of items to
be priced for the revised CPI and for the
* Most of the material included in this chapter was published in
"The Statistical Structure of the Revised CPI," Monthly Labor Review, August 1964, pp. 910-924.




weighting structure. Data collected in the
1960-61 Surveys of Consumer Expenditures35
were tabulated for the CPI revision project for
the group of families, or consumer units, classi­
fied as wage earners and clerical workers.36
A family is considered within the scope of the
CPI if 50 percent or more of its total income
during the survey year came from wage and
clerical occupations and if at least one member
of the family unit worked for a minimum of
37 weeks of the year. In the old series, families
were defined on the basis of the occupation of
the head of the household only. The change was
considered necessary because of the increasing
importance of families with two or more work­
ers and of family units whose household head
was retired, but which had other working mem­
bers.
In the 1964 revision, the index's population
scope was expanded to include single workers
living alone as independent consumers. This
was done on the advice of labor and business
advisory groups in order to make the index
more representative of the total wage- and
clerical-worker population with which collective
bargaining is concerned.37 At first it had been
85
Surveys were conducted in 66 metropolitan areas and urban
places located in the United States (including Hawaii and Alaska).
The list of cities and the method of selection is described in chapter V.
86
The definition of wage earners and clerical workers is based on
the occupational classification used by the Bureau of the Census for
the 1960 Census of Population and listed in the Alphabetical Index
of Occupations and Industries. The group includes craftsmen, fore­
men, and kindred workers, such as carpenters, bookbinders, etc.;
operatives and kindred workers, such as apprentices in the building
trades, deliverymen, furnacemen, smelters, and pourers, etc.; clerical
and kindred workers; service workers, except private household, such
as waitresses, practical nurses, etc.; sales workers; and laborers, ex­
cept farm and mine. It excludes professional, technical, and kindred
workers, such as engineers and teachers; farmers and farm man­
agers; managers, officials and proprietors, except farm; private
household workers; and farm laborers and foremen.
87
Urban wage earners and clerical workers and their families com­
prised 56 percent of the total urban population in 1960. Single
workers living alone represented 10 percent of all urban wage- and
clerical-worker consumer units to which the "new series" CPI
applies. On an expenditure weight basis, however, the importance of
single consumer units is only 6 percent of the composite index, be­
cause of the lower level of their expenditures.

19

expected that additional pricing would be in­
augurated to give proper representation to
items important in the single person budget,
such as room rent, room and board, restaurant
meals, and different qualities of other items.
However, in some of these cases, there appeared
to be no major difference between the types and
qualities of goods and services bought by single
consumers and families. In other cases, exces­
sive costs of price collection precluded separate
pricing. Therefore, no special pricing is carried
out specifically for such single workers.
However, weights were computed separately
for singles. This means that the weighting dia­
gram for the composite index was affected to
a small degree by the inclusion of singles. A
parallel index, excluding single consumer units
and based on weights for families of two or
more persons, was calculated for the United
States, but not for the individual cities, from
January through November 1964. Because it
was based on the same prices, and weights were
only slightly different, it moved almost the same
as the composite index during the 11 months.
Therefore, it was discontinued after November
because such minor differences did not justify
the cost of calculation.
There is no income limitation in the new
series CPI, as there had been in the old. Aside
from generally higher income levels for occupa­
tions within the scope of the index, an income
limitation was discarded because of the higher
income per family unit resulting from the
increased number of families with more than
one worker, and greater precision in the occu­
pational classification in the survey.
Geographic Coverage
A consumer unit had to be living within an
urban place in order to be included in the urban
portion of the Consumer Expenditure Survey
for cities surveyed in I960,88 but the expendi­
tures reported were not limited to that place.
The BLS attempted to get a complete record
of the family's expenditures during the survey
year, regardless of where they were made.
Similarly, there is no limitation in the measure­
ment of price changes in the CPI to purchases
in the home city. The index is intended to meas­
ure the price changes of items bought by urban

wage and clerical workers, regardless of where
purchased.
There is, however, an operational limitation
on pricing: BLS cannot follow consumers to all
the places where they make their expenditures.
The collection of price data for the CPI is cen­
tered upon 50 (or 56) sampling points—metro­
politan areas and small urban places.39 Indexes
are published for each of the larger metropoli­
tan areas separately, and with few exceptions,
the data which go into these indexes are prices
prevailing within the sample areas. For exam­
ple, restaurant expenses and costs of operating
automobiles, although they are often incurred
away from home, are priced only in the CPI
sample areas—not in resort or vacation areas.
In the new series CPI, several additional
items are being priced to represent expenditures
generally made away from home. Hotel and
motel expenditures are represented by room
rates, averaged on a regional basis, in the
metropolitan areas of the CPI sample which in
1960 had a population of 250,000 and over.
Hotels and motels in resort areas outside the
big cities are excluded because of the prohibi­
tive cost of such extended collection. Measure­
ment of price change in college tuition fees is
based on data from independent regional sam­
ples, since it has been possible to use an annual
survey conducted by the Office of Education.
As in the case of hotels and motels, the data are
averaged over a region in computing city price
relatives for this item.
Changes in living expenses or buying patterns
of the index population, as a whole, traceable
to the movement of population, for example,
from suburbs to central cities, from one part of
the country to another, etc., are treated as nonprice factors in the index.40 Such migrations
are prevented from affecting the measurement
of price change by the assignment of fixed popu­
lation weights to each area sampling center and
by the method of collecting price data.
The Expenditures Scope
Since the CPI measures price changes corre­
sponding to all spending for family living, the
weight and price data run the gamut from bread
and butter to television and bowling fees, from
39

38

The 1961 surveys were extended to rural areas.

20




40

The selection of sampling places is described in chapter V.
Costs of moving are, in concept, within the scope of the index.

prenatal and obstetrics services to funeral ex­
penditures, from popular paperback books to
college textbooks. The CPI is not, and never has
been, limited to price changes of so-called neces­
sities. Because it has been related particularly
to wage earners and clerical workers, the allinclusive item coverage was not as evident in
its earlier history, but the growing importance
in wage-earner budgets of automobiles, amuse­
ments, recreational services, and the like has
been reflected in the composition of the index
since 1940 and particularly in the current
index.
Weights computed from Consumer Expendi­
ture data of 1960 and 1961 were introduced in
the CPI in January 1964. They are intended to
represent average annual expenditures per con­
sumer unit for the urban wage-earner and cleri­
cal-worker population.41 As in the past, they
reflect experience of renters and of homeown­
ers; of car owners and of earless families; of
families with many children, and of childless
couples. For the first time the index weights
also reflect expenditure patterns of single con­
sumers living alone.
The scope of the expenditure weights corre­
sponds exactly to the price measurement scope
of the CPI. Thus, the expenditure weights in­
clude all taxes directly associated with them—
for example, sales and excise taxes—and so do
the price data collected for the CPI. Similarly,
taxes or government fees associated with par­
ticular purchases, properties, or services related
to family living (for example, transfer taxes,
property taxes on owner-occupied dwellings, car
registration fees, and water and sewerage fees)
are within the scope of the CPI. Income taxes,
however, are excluded as not being related to the
purchase or continued ownership of consumer
goods and services.42
Among the questions to be resolved in con­
sidering the scope of the weights is whether an
expenditure is for current family living or for
investment. In the 1953 revision of the CPI,
41
The average expenditure weights referred to above are com­
puted on an annual total basis, ignoring the fact t h a t some items have
a fluctuating pattern of seasonal consumption—or, in fact, may dis­
appear from the market entirely during certain periods of the year.
For a full discussion of this problem, see "Use of Varying Seasonal
Weights in Price Index Construction,** by Doris P . Rothwell, in the
Journal of the American Statistical Association, March 1958, pp.
66-77.
** Thus, in effect, the scope of the CPI corresponds to disposable
income as used in the national income accounts, except t h a t some
additional t a x payments to government are netted out in the latter,
e.g., motor vehicle license fees.




it was decided to treat home purchases as pur­
chases for consumption rather than as invest­
ments and this decision was reaffirmed in the
1964 revision. Thus, the CPI weights and pric­
ing system include the purchase prices of such
long-lasting items as houses and cars, even
though the consumer will not consume these
items completely for many years. This contrasts
with the "space rental" approach for measuring
the value of owner-occupied housing in the na­
tional income accounts and the proposals some­
times made that cars and other durables be
priced for the CPI on a "use value" basis.
Since an expenditure is considered consum­
mated when the obligation is incurred, rather
than when the payment is made, the total pur­
chase price is included even when houses and
durable goods have been bought on mortgage or
installment credit. The effect of this treatment
is qualified by the fact that weights for the more
expensive durable goods such as houses and
cars are taken net of trade-in offsets or of sales
receipts of corresponding items. Sales are not
netted against purchases for the index popula­
tion as a whole—but only where there is a pur­
chase and a sale by the same consumer unit
simultaneously, or almost simultaneously—i.e.,
during the year for which expenditures were
surveyed. Thus, the index weights correspond
closely to net outlays or obligations, and price
changes in these components affect the index
measurement of purchasing power in a real­
istic manner.
The fact that in both the weight regimen and
in measurement of price change a transaction
is considered consummated for purposes of the
CPI when the obligation is incurred, raises the
question of treatment of credit. All costs, in­
cluding credit, associated with installment pur­
chases are within the scope of the CPI. How­
ever, it is difficult to determine exactly how to
measure these costs for weight derivation, and
how to price them on a current basis. In the
CPI, as it has operated over the last decade
(and this remains unchanged), mortgage inter­
est has been included in both the weighting
structure and in price collection. The total cost
of interest contracted for has been included in
expenditure weights of other items bought on
the installment basis, but the Bureau has not
yet introduced techniques for pricing separate­
ly the cost of these credit services.
21

Although there is general agreement that
writing of life insurance represents a service
to the consumer that is within the scope of the
CPI, problems prevent this service from being
included in the weights or the pricing plan. The
major part of life insurance premiums repre­
sents a form of investment that should be ex­
cluded. The difficulty is to devise a method for
partitioning premiums into a consumption por­
tion (the service or protection part of the plan)
which should be within the scope of CPI, and
an investment portion which must be ignored,
and to be able to repeat this division periodically
to measure price change in the consumption
portion.
Other forms of insurance (e.g., property, car,
and medical) are included in the weights and
are priced for the CPI by one technique or an­
other. In these cases, the major difficulty is in
establishing prices for policies of constant bene­
fits, as provisions of policies change and as the
current dollar value of benefits changes with
prices. This is a variant of the more general
problem of quality change which pervades the
entire price index field.
Since it is related to expenditures, the CPI
does not reflect noncash consumption. Food
grown at home, fringe benefits received as part
of a job, services supplied by government agen­
cies without payment of a special tax or fee, and
so on, are not priced. These exclusions can affect
the interpretation of the index when the rela­
tive importances of these noncash consumption
items change over time in relation to cash out­
lays. Medical care, for which employers in re­
cent years have assumed an increased portion
of the expense, is an important example. The
accuracy of a fixed-weight price index for medi­
cal care, as a measure of cash outlays required
for medical care by the index population as a
whole, is affected by changes in the employer
share of medical costs.
Index Formula
In concept, the CPI is computed by compar­
ing, at different periods, costs of a fixed set of
goods representative of all purchases made by
urban wage and clerical workers. This is popu­
larly called a "market basket" index; techni­
cally, it is a price index with "fixed" or "con­
stant" weights.
22



The CPI procedure is to measure price change
by repricing at regular time intervals and com­
paring aggregate costs of the goods and serv­
ices bought by consumers in a selected base
period. Mathematically such an index takes the
form:
(1)
L2(Poqa)J
where i is the current month
a is the period of the most recent ex­
penditure survey (1960-61) from
which current weights were derived
o is the reference base period of the index
(most recently 1957-59)
q is a derived composite of the annual
quantities purchased in a weight base
period for a bundle of goods and serv­
ices to be represented by the specific
item priced
p is the average price of a specific com­
modity or service selected for pricing
The quantity elements, q«, of the above ag­
gregative formula are considered to be defined
in sufficient detail with respect to quality and
variety that explicit prices can be attached to
them at both time periods. Thus, the index for
period "i" with respect to period "o" taken as
100, It represents the ratio (multiplied by 100)
of aggregate costs of the same items priced in
both periods. A good part of the problem of in­
dex numbers is in defining what "same" means,
first in theory, and then in practice.
In actual operations, formula (1) is replaced
by its algebraic equivalent, the dollar weighted
average of price relatives:
Ii= p(p.q.)(p,/Po)-| xl00

(2)

The dollar weights are the expenditures re­
quired in prices of the reference period "o" to
purchase each component of the weighting pat­
tern relating to period "a". If the weighting
pattern is derived from the same period used
as the time reference base, formulas (1) and
(2) reduce to the Laspeyres, or base weighted,
formulas.
An equivalent but more convenient procedure
is to average period-to-period price relatives43
41
Price relatives are computed for all individual items, separately
by city, and take the form (Pi/Pi —1). This is the average price
for an item from one period divided by the average price from the
preceding period.

for individual items of the market basket, ac­
cording to the formula given in appendix table
III. The computation of the index is a chaining
procedure in which the index for the previous
month is multiplied by the average relative
change in price from the previous month to the
current month. Computing the price change
on a month-to-month basis makes it easier to
accommodate changes in the sample of items
and specifications priced as market conditions
change—a continuing process in the new, as
in the old, CPI. Such changes would be un­
avoidable even if the BLS desired—which it
does not—to maintain an inflexible sample of
priced items between periodic major revisions
of the CPL
The quantity weights and initially, the ex­
penditure weights, of the revised CPI related
to the period represented by the expenditure
surveys, 1960 and 1961. However, before intro­
ducing the revised expenditure weights into the
CPI, the data were revalued at December 1963
prices, when the new series CPI was linked to
the former series.44 The December 1963 link-in
weights for the new series were used with the
item price relatives from December 1963 to
January 1964, to obtain the average change for
all items. The December 1963 old series index
multiplied by this average change produced the
new January 1964 index.
As the chaining process is repeated each
month, the expenditure weights are automati­
cally kept on a current price basis. When trans­
lated to percentages, the revalued weights are
called "relative importances." For example, in
calculating the June 1964 index, the relative
importances (in effect, the weights) for the
May to June CPI change were the December
1963 expenditure weights revalued at May 1964
price levels.
As the preceding illustration emphasizes,
relative importances for month-to-month com­
parisons change with prices, but the expendi­
ture weights for comparing a current period
with the link-in date do not change.
The CPI chain-computation formula does not
result in a true chain index, except in the sense
of one with rather long periods between the
links, i.e., when the major weight revisions are
44
These adjustments were made on the basis of rather broad
groups or categories of items. Only in the case of very important
individual items were price trends of items used separately.




made. In the CPI chain computation procedure,
prices of comparable items are compared from
one period to the next. However, when substi­
tutions are made, the substitute price relatives
are used with the weights for the items which
they replace. In a true chain index, each price
comparison would involve a new compilation of
weights—and a new sampling of items to be
priced to take account of changes in purchas­
ing patterns of the index population.
The CPI is sometimes referred to as a modi­
fied Laspeyres index in the sense that the weights
refer back to some earlier period. The CPI is
not, however, precisely a Laspeyres index, since
the Laspeyres approach requires that the quan­
tity weights relate to the period with which
price comparisons are being made.45
Constant Expenditure Weights
In the recent revision, the BLS modified the
ground rules underlying the CPI to take par­
tial advantage of the flexibility of the chain
index approach, while retaining the virtues of
a fixed-weight index. As the first step in orga­
nizing the 1960-61 consumer expenditures data,
a classification system was developed to allocate
each consumer outlay into 1 of 52 expenditure
classes (which later became the strata for se­
lecting the sample of items for pricing).
Until the next major CPI revision—barring
some emergency—the BLS expects to maintain
constant the base period weight relationships
of the expenditure classes. In current dollar
terms, the relative importances of the classes
will change as prices change.
The BLS may, however, choose a new sample
of items within any of the expenditure classes
for pricing in the current CPI, whenever there
has been a significant shift in the composition
of consumer expenditures within the category.
This could occur when new products or new
services within the group come into the market
in significant dollar volume. Thus, if the BLS
has access to data which show that patterns of
spending within an expenditure class have
45
The time reference base for all Government index series is estab­
lished by the Office of Statistical Standards, Bureau of the Budget.
Currently, it is the average of the years 1957 through 1969. The CPI
weights relate neither to the 1957-59 period, nor to the December
1963 link-in date; therefore, index comparisons against either of
these periods are not true Laspeyres comparisons. While the revised
weights refer back to 1960-61, price comparisons against that period
are not of the Laspeyres type either, inasmuch as the old series
index, which carries through December 1968, is a necessary part of
such comparisons.

23

changed significantly, it will be free to bring
new items into the sample of items priced for
the CPI when they become significant, or to
reweight within an expenditure class the items
formerly priced. If this action is warranted, the
BLS can select a completely new and independ­
ent sample of items to represent the particular
class. Changes in the sample or internal weights
within an expenditure class would be introduced
by a linking process, so that the changes would
not affect the index level.
In the former application of the market bas­
ket approach, the pattern and level of living of
the weight base period was effectively defined
as a group of specific items selected for index
pricing, and their associated expenditure
weights. Since the price index was intended to
hold the level and pattern of living constant,
this operational definition led to relative inflexi­
bility in the item sample. The new approach
outlined above emphasizes that the market bas­
ket of items priced for the CPI has significance
only as a sample representative of all consumer
expenditures; nothing more. When segments of
the sample cease to be representative, they may

24



be changed by a systematic procedure built into
the index framework.
In the current index, the level and pattern
of living of the base year to be held constant
in index comparisons is effectively defined in
terms of the weight base period (1960-61) dol­
lars which consumers spent on each expenditure
class relative to each other expenditure class.
Thus, if the base period dollars are interpreted
as quantity units,46 it can be said that the quan­
tities in the index system which are being held
constant are the base period dollar aggregates
spent for each expenditure class. Whether or not
there is more than a semantic difference be­
tween the new approach and the old will depend
upon the availability of reliable data to judge
whether the pattern of spending has changed
sufficiently for it to be desirable to make broad
adjustments of the sample of weights within
expenditure classes.47
48
The current period price corresponding to a base period dollar
quantity unit is identical with the price relative, i.e., the ratio of the
current price to the base period price.
47
The BLS had hoped for a program of annual expenditure sur­
veys which would have served this purpose, but plans for this have
not developed. Unless satisfactory data are available from secondary
sources, the plan to make periodic adjustments of the item samples
cannot be implemented.

Chapter IV. Sampling Aspects of the 1964 Revision*
It is axiomatic that any large complicated index
such as the Consumer Price Index (CPI) for the
United States must be based on a whole complex
of samples. A sample of cities or areas is required,
and within each area a sample of urban families or
consumer units, from whom consumer expendi­
tures are obtained. These consumption data pro­
vide the weights by which price changes in the
components of the CPI are combined to higher
levels. There must also be a sample of areas in
which to collect prices and it is usually convenient,
but not essential, that these sampling points be the
same as those in which the consumer expenditures
surveys are conducted.
Further, since it is impossible to price all the
thousands of items which consumers buy, it is
necessary to select a sample of items for pricing,
to represent price movement of all items. There
must also be samples of outlets, in each sampling
area, from which price quotations are obtained for
the selected items. Finally, pricing in any one
store is done on a specific day at monthly or
quarterly intervals so there is, in effect, a sampling
of time.
Any sampling plan should be related to a partic­
ular goal. In the case of the CPI, the primary
objective is to produce the most accurate national
index possible but, at the same time, the index
system is expected to produce accurate indexes for
large metropolitan areas and for the major com­
ponents of expenditures separately. With the
experience of many years in this field of work, the
BLS knows that in order to satisfy users it must
also provide data at the item level—both average
prices and indexes of price change. All of this,
plus the fact that the price index is based on an
aggregation of dissimilar components which do not
comprise any previously defined composite of
goods and services, leads to a very complex
sampling design, or rather a set of sampling
designs.
*Most of the material in this chapter was included in an unpublished
paper, "Sampling Aspects of the Revised CPI," by Marvin Wilkerson.




Probability Sampling
As mentioned in chapter II, in the 1964 revision
the BLS placed emphasis on the extension of
probability sampling. However, such methods
were by no means new to the CPI. The samples
of consumer units in the CES traditionally have
been selected by probability methods. So have
the samples of rental dwellings used to price rent.
From time to time as opportunities presented
themselves, probability sampling was adopted in
other phases of the CPI: in the selection of samples
of doctors, samples of properties for measuring
property taxes, samples of lending institutions for
use in measuring mortgage interest rates, etc.
In other areas, however (particularly in the
sampling of items), no comprehensive attempt to
use such methods had been made prior to the 1964
revision. This appears to have been the case in
most other countries as well, and there are good
reasons for this situation. One of the greatest
obstacles in the sampling of items is the difficulty
of constructing a meaningful sampling frame.
Samples in a price index are not used for a single
point in time but must serve for possibly 10 years.
Under the circumstances, informed judgments,
based on consideration of all the relevant infor­
mation available, has usually been preferred as a
sounder method of selecting a sample than some
random method, even though the former cannot
conceptually be called an unbiased procedure.
Samples of outlets, except possibly for food, are
typically small (4 to 8 stores per city in most cases)
and the task of attempting to represent adequately
all the possible types of outlets through a prob­
ability sample has usually not appeared feasible.
This has been expressed by a leading price statis­
tician of India as follows: " . . . it is usually possible
to cover only a very small sample of outlets for
any one specification. With the smallness of the
sample size, a representative probability sample
including all types and sizes of outlets often
becomes impractical." **
48
A. Basu, "Consumer Price Index Numbers—Sampling Problems in
Prices/' Indian Labour Journal, Delhi, June 1960, pp. 582-588.

25

Despite these well-known difficulties, however,
the BLS decided, in the 1964 revision, to attempt
extension of probability sampling as far as pos­
sible. The degree to which this objective was
achieved and the operating methods employed are
described in detail in the ensuing subject matter
chapters. The major technical considerations are
summarized here.
Estimating Sampling Error
Probability sampling is a necessity if estimates
of sampling error are to be derived in a conven­
tional manner. However, it is apparent that even
if probability sampling could be followed rigor­
ously through all the complicated CPI structure,
the mere computational load would be so extensive
that it would be impractical to compute measures
of error except by some "simple" approach.
Such methods have been used increasingly in
other series in recent years under such titles as
"interpenetrating samples," "random groups,"
"replicated samples," "ultimate clusters," etc.
While the BLS was still exploring the possibilities
of such methods, an outline of a replicated sample
design, in a staff paper49 prepared for the Price
Statistics Review Committee, became available
and it was used as the starting point.50
The BLS sample design includes two samples of
cities (or standard Metropolitan Statistical Areas),
two replicated item samples, and two replicated
outlet samples. In addition to the minimum pro­
gram, designed to produce an estimate of the
combined sampling error in the index, the struc­
ture includes more extended replication in selected
cities, aimed at permitting some evaluation of the
components of the error—those due to sampling of
cities, items, and outlets. (It is hoped that this
knowledge of error components will be helpful in
future decisions related to utilization of resources.)
A more detailed outline of the replication model is
given in a separate section at the end of this
chapter.
The design of city samples, consumer unit
samples, item samples, and outlet samples is
discussed in detail in subject matter chapters V,
VI, VII, and VIII, respectively. Only brief sum­
maries will be given here.
"See Government Price Statistics, Sampling Considerations in the
Construction of Price Indexes with Particular Reference to the United
States Consumer Price Index by Philip J. McCarthy, pt. 1, Jan. 24,1961,
pp. 197-232.
*o Professor McCarthy later served as a consultant to the BLS in expand­
ing and adapting his original outline. .

26



City Sample
The CPI sample is customarily referred to as the
"city" sample and the selected localities as
"cities," even though the sample consists of the
urban portions of Standard Metropolitan Statisti­
cal Areas (SMSA's) and urban places outside
SMSA's. This is due partly to historical usage
dating from earlier periods when CPI price data
were collected in large cities only; also, the term
"city" tends to emphasize the urban coverage of
the index. The term is used in this context
throughout this bulletin.
A core sample of 50 cities for the revised index,
supplemented by 16 additional small nonmetropolitan urban places for the family expenditure sur­
veys, was determined to be the maximum size con­
sistent with available budget. The primary
sampling units (PSU's) were Standard Metropol­
itan Statistical Areas as defined by the Bureau of
the Budget prior to the 1960 Census, except that
the Standard Consolidated Areas for New York
and Chicago were used, plus individual urban
places outside the SMSA's. The measure of size
used was estimated urban population as of Jan­
uary 1,1959. The PSU's were stratified by broad
region and by size into 12 regional-size strata.
The 12 largest SMSA's were selected with cer­
tainty, that is, they represent themselves in the
sample design. These areas each had a 1960
population of over 1,400,000. Since both Alaska
and Hawaii were covered in the revised CPI, one
sample selection was allocated to each of these two
States. The remaining 36 selections were al­
located to the 12 regional-size strata on the basis of
their relative population and the relative costs of
pricing cities of different sizes.
The exact method of selection was a matter of
considerable study and experimentation. The
method finally used was one that is generally
known as "controlled selection." 51 The procedure
is discussed in chapter V.
After the initial 50-area sample was selected, the
Bureau received funds to conduct expenditure
surveys and prepare city indexes for six additional
large SMS'As (having over 1 million population in
1960). These areas were planned for addition to
the national index in January 1966.
51
Described by Roe Goodman and Leslie Kish in the September 1950
issue of the Journal of the American Statistical Association, pp. 850-872.

Samples of Consumer Units
The samples of consumer units used for the
Consumer Expenditure Surveys covering the 2
years, 1960 and 1961, were drawn through a twostage random sampling procedure. Compre­
hensive Housing Unit Surveys (CHUS) were
conducted in each sample area late in the year
preceding the actual survey date. The CES
samples of addresses were chosen as subsamples of
the housing units enumerated in the CHUS. The
sampling procedure is described in chapter VI,
"Housing and Expenditure Surveys/'
Classification System and Item Sample
As the first step in selecting the item sample for
the revised index, a classification system was
developed to provide a logical publication frame­
work containing the traditional major expenditure
groups, subgroups, etc. In a broader sense, one of
its basic functions was to divide the thousands of
goods and services purchased by consumers into
meaningful and manageable components of the
universe. It provided the framework for the se­
lection of the item sample and for the deriva­
tion of index weights.
Two levels of the classification system were of
critical importance. These were: (1) the item
level, and (2) the level which defined the finest
stratification for the item sampling; that is, the
strata to which allocations of items were made and
within which probability samples of items were
selected. The term "expenditure class" (EC) was
given to this level.
The expenditure class was also conceived of as
the level at which base period expenditure weights
will be held constant until the next major revision
of the CPI. The Bureau reserves the right to
resample items within an EC between major
revisions if circumstances warrant. This mini­
mizes the importance attached to the sample of
items priced for the CPI, and emphasizes that the
priced items have significance only as a sample of
items selected to represent price movement of all
items.
The definition of the items was given a great
deal of thought and discussion. The list of line
items in the schedule (that is, items for which
separate family expenditures were obtained), used
in the Consumer Expenditure Surveys in 1960 and
1961, provided a logical starting point. To some
extent the existence of this list and the fact that




expenditure data were available from the CES
surveys for these line items provided a limit on the
detail in which the items could be listed for sam­
pling purposes.
It was realized from the start that the definition
of an "item" would have to be fairly broad and
that it was not feasible to list the final "speeifiedin-detail" items for which prices are collected.
The following general rules were set up as guide­
lines: (a) the item should not be so broad as to
leave most of the sampling operation in the judg­
ment area of selecting and defining specifications;
(b) the item should not be so narrowly defined,
however, that its definition unduly restricts the
selection and maintenance of specifications for
pricing; (c) the items should be as consistent as
circumstances permit in the degree of homogeneity
of the subitems included; and (d) to be of use for
sampling purposes, it should be possible to obtain
or derive some measure of the relative importance
in total family spending for each item listed.
The item sampling procedure is described in
greater detail in chapter VII, "Weighting Struc­
ture of the CPI, 1964."
Outlet Sampling
The first big problem encountered in probabil­
ity sampling of outlets from which prices are
obtained was to obtain information about the
universe of retail and service establishments in a
given area. Ideally, it would have been desirable
to have names and addresses of such places, infor­
mation as to type of store or outlet, some indica­
tion of volume of sales, and preferably (although
this is usually unavailable without a personal
contact) fairly specific information as to types of
merchandise carried.
A number of possible sources of comprehensive
establishment data were investigated. The only
one which proved fruitful was a master list of firms
which report to the Bureau of Old-Age and Survi­
vors Insurance (BOASI—Social Security Adminis­
tration, U.S. Department of Health, Education,
and Welfare). The identification codes in the
BOASI file did not permit selection by individual
city or SMSA, however, and it was necessary to
start with data assembled by the BOASI, and by
the Bureau of the Census for its publication,
County Business Patterns.
Using sampling ratios furnished by the BLS,
master samples of retail and service outlets were
selected within the counties embracing the sample
27

areas. These selections proved useful but needed
considerable supplementation by rosters, groceryroute lists, telephone directories, etc.
In the larger SMSA's, a sample of neighborhood
and suburban localities and shopping centers were
selected in which pricing was to be done, as well as
in downtown areas.52 These were usually selected
with probability proportional to sales volume,
using the best available sales data. The listings
of outlets were limited to those falling within the
sampled areas.
Although the geographic coverage of the CPI,
insofar as place of residence is concerned, is limited
to urban areas, this limitation does not apply to
price collection. It is pertinent to price wherever
urban families shop. While practical considera-*
tions limit pricing to the vicinity of the 50 (or 56)
sample areas (except for a few items like college
tuition and hotel and motel rates), important
shopping centers are included in some cases even
though they are located outside urban boundaries.
In a few small cities it is necessary to go to nearby
towns to secure price quotations.
All the various sources were used to develop a
sampling frame for selecting samples of outlets.
The frame was organized by specific type of outlet;
for example, department stores, men's clothing
stores, family shoe stores, etc. Each priced item,
in effect, required a separate sampling operation.
It was not feasible, from a cost standpoint, to
select each of these samples independently, since
this could well have spread the pricing over an
excessive number of different outlets with a very
few quotations obtained from any one outlet.
Selection with probability proportional to size
was not possible in all cases and many replace­
ments of original selections proved necessary.
Most replacements were made because the stores
did not carry any of the items BLS was attempting
to price or because they refused to cooperate in
reporting prices. Whenever possible, replace­
ments were selected by the Washington office from
the original sampling lists. By the time the final
complete outlet samples were established, so many
expedients may have been used that no claim can
be made for strict probability samples. However,
the approach is within a probability framework
"For example, in the Los Angeles-Long Beach SMSA, four "major
retail centers" were selected within the city of Los Angeles (in addition to
the central business district) from 28 such centers defined for the 1958
Census of Business. These are known as "Hollywood and Vine,"
"Crenshaw Center," "Miracle Mile," and the "Valley" area. Outside
Los Angeles proper, 7 cities out of a total of 70 were selected for pricing:
Long Beach, Anaheim, Beverly Hills, Montebello, Pasadena, South Gate,
and West Covina.

28




and it is believed that many of the main benefits of
probability sampling have been achieved: Lack of
bias, representation of different types of outlets,
sections of each SMSA, etc. Procedures used are
described in chapter VIII, "Outlet Samples, 1964
Index."
The Replication Design in the CPI
Computing a sampling error by a standard
formula requires probability sampling. The com­
putation gives a measure of the dispersion that can
be expected among many estimates made by
repeatedly sampling the same universe, using the
same sample design and estimating procedures,
and in which the variation is due only to the
chance differences in the particular cities, firms,
families, individuals, etc., which happen to fall
into the various samples. This error estimate is
derived from the basic variances in the universe
and the formula appropriate to the sample design,
and can normally be computed satisfactorily from
the results of a single sample even though it per­
tains to a whole universe of sample estimates.
The replication approach is an empirical, rather
than a theoretical, one. Repeated samples are
chosen and the variability among the sample
results is observed and a measure of sampling error
is derived. The sampler, in effect, generates a
distribution of sample results by the repeated
application of a sampling and estimating pro­
cedure and computes its variance. It is not even
necessary that the sampling be done by probabil­
ity methods as long as all the samples are selected
by the same general procedure so that they can
realistically be regarded as "replicates."
A limitation of the replication approach for such
a complicated operation as the CPI is that the
maximum number of replications which is prac­
tical is two (that is, two item samples, two outlet
samples, and two city samples). A greater
number would, of course, give better estimates of
sampling error. However, methods of cumulating
data across geographic strata and across com­
modity groups to achieve more stable variance
estimates for the all items CPI are described below.
In order that any replicated sample approach
may reflect the error contribution from different
sources, the variation due to these sources must be
built into the model; that is, they must be repli­
cated. The CPI model includes the replication of
cities, of item samples, and outlet samples.

Efforts might have been made to include the
effects of other sampling operations, for example,
selection of specifications and derivation of the
index weights from samples of consumers. Repli­
cation of index weights did not appear practical;
in any event, it is felt that the effect of different
index weights is minor compared to the variation
introduced by other factors.
Although replication of specifications was not
carried systematically throughout the design, a
certain amount is included. Where different items
appear in sample " 1 " and sample "2," any com­
parison of price trends between the two samples
necessarily includes both the effect of sampling of
items and of specifications. For those items
which appear in both samples, a planned system
of replicated specifications was used in only a few
selected items.58 In most cases the same specifi­
cation is priced for both item samples. However,
there are many alternate specifications, city
deviations, outlet deviation, etc., so that it is by no
means true that the identical specification is priced
in all cities for all items.
As pointed out earlier, the city sample design
did not explicitly include any provision for repli­
cated city samples. An ex post facto pairing of
CPI sample cities is used to simulate the selection
of two cities from each stratum. (This is some­
what analogous to the practice of "collapsing"
strata for the purpose of computing variances.)
The pairings were made by associating cities
which most logically could be considered to be in
adjoining "strata." There is no logical pairing
for two arqas, Green Bay and Bakersfield, so these
are paired with Cedar Rapids and Austin for
purposes of variance computation. Following are
the pairing of cities for replication computations:
Paired SMSA's or cities

Stratum

"B"

(2)
(1)
[Buffalo, N.Y.] i
Hartford, Conn.
♦Buffalo, N.Y.
*Dayton, Ohio
Indianapolis, Ind.
[Dayton, Ohio]»
♦Dallas, Tex.
♦Atlanta, Ga.
Nashville, Tenn.
[Atlanta, Ga.] *
♦Wichita, Kans.
*Denver, Colo.
♦Seattle, Wash.
[*Denver, Colo.]
Portland, Maine
Lancaster, Pa.
♦Champaign-Urbana, 111. ♦Cedar Rapiids, Iowa l
[Cedar Rapids, Iowa]
Green Bay, Wis.
♦Orlando, Fla.
*Durham, N.C.
Baton Rouge, La.
Austin, Tex.
Bakersfield, Calif.
[Austin, Tex.]

38
Examples of items for which different specifications are priced in the
two item samples are: steak, new automobiles, dentists' fees, sports equipmeat, and cigarettes.




Paired SMSA's or cities

Stratum

Southbridge, Mass.
Findlay, Ohio
Niles, Mich.
Crookston, Minn.
Union, S.C.
Florence, Ala.
Mangum, Okla.
Klamath Falls, Oreg.

Kingston, N.Y.
Millville, N J .
Logansport, Ind.
Devils Lake, N. Dak.
Martinsville, Va.
Vicksburg, Miss.
McAllen, Tex.
Orem, Utah

♦Cities having the extended replication program.
1
Half-sample only used for bracketed city (except for the "Food at
home" group in the " B " size stratum), since paired city has minimum
replication only.
Each city in col. (1) paired with opposite city in col. (2). I ] City in
brackets is used in special pairing to compute variance estimates for other
city in pair.

Minimum Replication Model
City

X
Item
Sample

Y

a*

Cl

c2

Ry

The two replicated item samples are described
in chapter VII. In the minimum replication
program, one of these samples, cx, is priced in one
city of a pair and the other sample, c2, is priced in
the second city. (See diagram.) Using these
different item samples (and, of course, outlet
samples), indexes are computed for some time
period, t, for the two cities: call these indexes
lix<e) and RY(e). (For simplicity, the time nota­
tion will be omitted hereafter.) An estimate of
the within-stratum variance can be made by a
comparison of ftx and ftY, and this estimate will be
influenced by the sampling of items, and the
sampling of retail outlets within cities that is, the
effects of these sampling operations have been
built into the replication procedure. The stratum
variance for stratum i based on these observations
reduces to
(Ti2 =

(ft x -ft T )t

.(1)

Although strata have been collapsed in order to
estimate between-city variances, it is assumed in
computing the United States sampling error that
one city has been drawn from each stratum and
the computed variance is used for both city X and
city Y. (Since a sample of one is used for each
stratum, the above estimate of the population
29

between-city variance is the appropriate quan*
tity.)

CPI city, then the sampling variance of the U.S.
index can be written:

The 12 largest SMSA's, or "A" cities, selected
with certainty represent only themselves; hence
their contribution arises entirely from within-city
variation. In such "A" stratum city for the mini­
mum program, both item samples must be priced,
each in its own outlet sample.

V =I>iW
(4)
where the stratum variances are computed as
outlined above. This formula will apply for the
all-items index or for indexes for major groups,
subgroups, or for any more detailed components.

As before, two indexes are computed, one for
each item sample. A variance computed from
these values will be an estimate of the variance
among indexes based on "half-samples." How­
ever, the city index is based on an average of two
such half-samples, so its variance (in the i-th city)
is
Outlet sample

oA

OB

RA

Ci

R*>

Cg

,

(RA-RB)8

<rr=*-

(2)

—

(The standard deviation, alt for the individual city
index becomes simply
(RA-RB)

""

(3)

2 —

Honolulu and Anchorage, the cities representing
Hawaii and Alaska respectively, are handled as
special cases. Both item samples are priced in
Honolulu and the within-city variance computed
as in an "A" stratum city. Since the Honolulu
SMSA includes about 90 percent of Hawaii's
urban population, the between-city variance is
ignored. No reasonable pairing is available for
Anchorage, and since the December 1963 relative
weight for Alaska in the CPI is less than 0.1
percent, Anchorage has been omitted from the
error computation.
If Wi where ]£wi==l is the weight, or relative
importance, of each stratum represented by a
30



The all-items index variance can be computed
also by regarding the major groups as strata and
appropriately combining the group variances.
This procedure is valid if the major group indexes
are independent estimates. Since there is little
overlap in the outlet samples between food, hous­
ing, apparel, etc., the assumption of independence
seems reasonable and this approach is used with
nine major groups. These are "Food at Home,"
"Food Away from Home/' "Housing/3' "Apparel/1
"Transportation/' "Medical Care," "Personal
Care," "Reading and Recreation," and "Other
Goods and Services." ("Food Away from Home"
is handled separately from "Food at Home" be­
cause the replication patterns are different.)
The assumption of independence cannot be
carried indefinitely down to successively more
detailed subgroups, EC's, etc. (the first 15 EC's,
for example, are largely priced in the same stores),
but it may be possible to set up more strata as the
computation procedures are tested and refined.
For example, although home purchase and rents
are individual "items," their price measures are
derived from samples which are unique to these
items. Since they are quite important in them­
selves, it may be deemed desirable to consider
them as separate strata.
Extended Replication
The above discussion outlines the minimum
replication program which produces estimates of
the combined error of the U.S. all-items or group
indexes. It does net provide estimates of the
components of this error, that is, the contribution
due to sampling of cities, of items, and outlets.
A somewhat more elaborate replication program,
which permits estimates of these components
through a simple analysis of variance, is used in a
portion of the CPI sample cities. The information
obtained from these selected cities may be suf­
ficient to allow a rough partitioning of the com­
bined sample variance of the U.S. index into its
cpmponent parts.

the primary purpose of the extended replication is
to obtain information on the components of
variance and it becomes more important in this
respect to avoid introducing the extraneous factor
X ! Y
of different pricing months. This is particularly
true in dealing with short-term price change.
Instead of arbitrarily switching cities to put paired
Cj
RIB
RlA
"B" cities on the same cycle, two auxiliary pairs of
"B" cities have been set up for extended replica­
tion: Buffalo-Dayton and Wichita-Denver. Al­
though both these pairs cross region lines, they are
c2
RJB
R*A
otherwise quite reasonable pairings.
Furthermore, after this design was established,
In this model, both item samples are priced in each
plans were made for publication of individual city
city and each item sample is priced in different
indexes for all cities having a million or more popu­
outlet samples. For each collapsed stratum, four
lation in 1960. Adequate city indexes require
indexes are computed. From these four values,
pricing of both item samples. Therefore, pricing
estimates are computed of the variance due to
in all published "B" cities is at the extended repli­
the sampling of cities (o^), that due to sampling
cation level. In order to estimate variances for all
of commodities (<re2), and a residual (<re2) which seven "B" areas with the more extensive replica­
is a combination of the variance due to sampling
tion, some cities are used in two different pairings;
of retail outlets and to random error. The variance
once to estimate variances for themselves and
of the collapsed stratum index, R, can be expressed
again to estimate for the "orphaned" member of
in terms of these three components:
an original pair. For example, the special pairing
of Buffalo and Dayton, both having the extended
replication, is being used to estimate variances for
these two areas. In separate computations a
However, since the previously outlined method of
"half sample" (that is, a single item and outlet
combining across strata to the United States level
sample) in Buffalo is used to compute the variance
assumes that each city represents a stratum, the
contribution for Hartford, and a half sample in
variance above (appropriate to a sample of two
Dayton is used to estimate the variance contribu­
cities from a collapsed stratum) is only half as
tion for Indianapolis.
large as that needed. The "single stratum"
For food items the extended replication is used
variance can thus be computed by the formula:
in all "B" cities. The reason for this exception is
as follows. For nonfood items, where the usual
outlet sample size is 4, the extended replication
almost requires doubling the amount of pricing
The extended replication is used in seven
since it does not appear feasible to reduce the
stratum "B" cities for nonfood items and two
basic sample size below four quotations. Since the
pairs of stratum "C" cities. No extended replicafood store samples are larger, each replicated
cation is done in the "D" stratum.
outlet sample, in cities with the extended program,
Choosing pairs of "B" cities for extended repli­
need not be full size. Under the extended replica­
cation was complicated by the cycles on which
tion program, both food item samples would have
pricing is done. In all "B" cities, pricing is done
been priced in selected "B" cities, each in samples
once each quarter (except for food and a few other
of about 15 independent food stores, whereas in
items which are priced monthly). The primary
the remaining "B" cities having the minimum pro­
criterion in assignment of cities to pricing cycles is
gram a single sample of items would be priced,
attainment of a well distributed sample in each
but in a sample of 30 independent stores. Since
month. It happens that na pair of "B" cities
the total number of quotations would have been
chosen for replication fall into the same pricing
about the same in both cases, the extended replica­
cycle.
tion is used for food in all "B" cities, that is, both
item samples are priced but in different samples of
For the computation of the overall error of the
outlets.
CPI, this is not of major importance. However,
The extended replication of the noncertaiiity
cities is shown in the following diagram:




at

In the " C " stratum, the more elaborate replica­
tion program is applied only in two selected pairs
of cities for food as well as other items. In the
four selected " C " cities, the two samples of inde­
pendent food stores in each city have five outlets
each. In the six other " C " cities, the single
sample per city has 10 outlets.
The extended replication in selected A (cer­
tainty) cities is outlined below. Both item
samples, Ci, and c2, are priced across both outlets
samples, oA and o B .
Outlet Sample

oA

OB

Ci

RlA

RIB

c2

R2A

R2B

Analysis of variance can be made on the resulting
four indexes to produce estimates of variance due
to sampling of items (<rc2), of outlets (o-02), and
random error (cre2).
The formula for the error in the i-th city index
is analogous to (5) for the stratum index for
paired cities:
2

<Tc2 , (To2 , ore2

(7)

Since even the minimum program requires two
outlet samples in each "A" city, the extended
replication means that all items are priced in two
outlet samples (rather than just the certainty
items as in the minimum replication). The ex­
tended replication is set up in three "A" cities for
nonfood items, but in all "A" cities for food.
The reason for this design for food is similar to that
explained above for " B " cities.
The fact that food and nonfood items have dif­
ferent levels of replication in some cities means
that in these cities different computational
methods are required for the food and nonfood
categories. For example, in the nine "A" cities
with the minimum "A" replication there are four
sets of aggregates for food but only two for the
nonfood groups. Consequently, it is necessary to
compute a "food" variance by the extended analy­
sis of variance approach and a "nonfood" variance
32



by the simpler comparison of two index values.
An all-items variance can be computed (for the
city) only by weighting together the variances for
two commodity groups (or more detailed groups).
A similar situation exists for pairs of " B " cities
which have the minimum replication. However,
since the all-items CPI error computation is done
through the use of at least nine groups, this offers
no particular problem.
Since each of the six additional large SMSA's
will be self-representing when added to the CPI
sample, the replication diagram and computation
procedure will correspond to the minimum replica­
tion for "A" cities (although these areas are
actually in the " B " stratum).
Since the revised index mechanism is used to
measure price change only after the link date,
December 1963, the replication design, likewise,
can only measure error in the CPI from that date
forward. Presentations of estimates made via the
replication procedure show index changes from
December 1963 (rather than the current index
itself) along with the error estimates which cor­
respond to these changes.
In addition to the long term error estimates, for
which the replication procedure was primarily
devised, error estimates for short term (for ex­
ample, quarterly) change in the CPI can be pro­
duced periodically. (This procedure is not strictly
correct, since it ignores the fact that the effective
weights for price changes over periods not dating
back to the link date are themselves sample
variates. The replicated error mechanism treats
the overall index change as an arithmetic average
of change for different samples of cities, items, and
outlets with known weights. However, the
weights, or "relative importances," used in averag­
ing current price relatives, change very slowly over
time and the continuing estimates of short term
error should give satisfactory approximations.)
It should be recognized that estimates of error
computed by replicated sample methods include
more than pure sampling error. They are influ­
enced by interviewer, supervisory, editing, proc­
essing and similar errors to the extent that such
errors are random in nature. Of course, sampling
error computed by formula can also contain the
influence of such factors, but probably not to as
great an extent. The replication procedure can­
not, however, detect any persistent bias which
may be present.

The inclusion of these other types of error is, in
many ways, desirable. If estimates are affected
by such errors, then it is appropriate that their
influence be included in the error measurement,
thereby giving a better approximation of the total




error (sampling and nonsampling) to which the
estimates are subject. Any publication of error
estimates for the CPI will, however, make clear to
the user what the estimates cover.

33

Chapter V. City Sample Selection, 1964 Index*
Selection of a new sample of cities for the re­
vised Consumer Price Index was considered
necessary because of shifts in the geographic
distribution of the population over the decade.
These shifts meant that the average pattern of
expenditures would be different from what it
had been in 1950, a factor that would directly
affect the index weights and indirectly, the
trend of prices.
A sample of 50 cities for the index was estab­
lished as the maximum number that could be
priced for the index since the resources for the
pricing program were planned to continue near
previous levels. The increase from 46 to 50 was
quite nominal in view of the additional coverage
of Alaska and Hawaii. Two alternate samples
were selected for possible future use in expand­
ing the pricing worik if later circumstances per­
mitted.
After the revision program was well ad­
vanced and consumer expenditure surveys com­
pleted in the selected sample of cities, questions
were raised concerning the availability of indexes
for individual metropolitan areas. Since it was
not possible to satisfy demands for local in­
dexes, some general policy was needed to govern
the Bureau. It was agreed, in consultation with
the Office of Statistical Standards of the Bureau
of the Budget, that indexes should be provided
for the 22 Standard Metropolitan Statistical
Areas having a population of 1 million or more
in 1960, and for Honolulu, as well as four
Alaska cities published semiannually under pro­
visions of special legislation. Of the 22 largest
areas in the country, 6 had not been drawn in
the national sample of 50 selected for the index.
These six cities are San Diego, Houston, Cin­
cinnati, Kansas City, Minneapolis, and Mil­
waukee. Funds were provided for conducting
expenditure surveys and initiating pricing in
these six cities and initiating indexes in 1965,
* Host of the material included in this chapter was published in
"The Revised City Sample for the Consumer Price Index," by Marvin
Wilkerson, in the Monthly Labor Review, October 1960, pp. 10781083; available as Reprint No, 2352.

m



with addition to the national index in 1966.
(One of the cities, Cincinnati, the pilot city
already surveyed for 1959, was not resurveyed.)
Primary Sampling Unit
The primary sampling unit for the new sam­
ple is the Standard Metropolitan Statistical Area
(SMSA) in the metropolitan segment of the
United States and the individual urban place
(over 2,500 population) in nonmetropolitan
areas.54 The SMSA is usually slightly more ex­
tensive than the "urbanized area" unit used in
the 1952 revision, encompassing some small
noncontiguous urban places that were not in­
cluded in the urbanized area. Expenditure pat­
terns and price movements in these small places
can be expected to resemble those in the metro­
politan urban segment more than those in nonmetropolitan urban places.
In planning the design of the city sample,
estimates of total and urban population for all
counties in the United States as of January 1,
1959, were obtained from Sales Management,
the Magazine of Marketing, published by Bill
Brothers Publications, since data from the 1960
Census of Population were not then available.
These estimates are projections of 1950 Census
data adjusted to less detailed Bureau of the Cen­
sus estimates for 1959. Among individual nonmetropolitan urban places, Sales Management
estimates were available only for cities of about
10,000 or more. For smaller places, the BLS
made its own estimates for all places that were
urban in 1950 and for some 200 other places
that were estimated to have grown into urban
status by January 1, 1959.
Stratification
Tests of the effectiveness of some of the more
obvious modes of stratification, such as geo54
The Standard Consolidated Areas for New York and Chicago
were used as single primary sampling units rather than being divided
into their constituent subareas,

graphic region, size of city, and climate, indi­
cated that no elaborate stratification was justi­
fiable for a sample of only 50 areas. Analysis
of variance techniques were applied to price
movements for three different time periods for
25 items and groups of items; similar analyses
utilized expenditure data from the 1950 Con­
sumer Expenditure Survey. These results were
of limited usefulness because of the small num­
ber of cities, time periods, and items for which
data were tested. The significance of the dif­
ferent classification modes varied from item to
item, but in general, classification by region and
by size of city seemed to be most effective. Since
these are the two most useful modes of strati­
fication from almost any point of view, it was
decided to use them. The four Census regions—
Northeast, North Central, South, and West—
were used as the geographic areas. The size
stratification was of particular importance be­
cause of differential cost factors. Four popula­
tion size strata, the same number used in the
previous revision, seemed to be about the maxi­
mum which the sample size would justify: Less
than four would mean excessively wide stratum
limits; even with four, the limits are far apart.
A number of possible size groupings were
considered. It was decided to retain the 12 larg­
est areas previously included in the CPI sam­
ple as certainty selections in the new sample.
These 12 areas comprise the A stratum. The
other three population strata were set up as
nearly as possible in terms of commonly used
size groups, in order to facilitate comparison
with other economic data. Accordingly, the four
strata are defined as follows:
Stratum A—Twelve largest SMSA's.
Stratum B—Other SMSA's with urban population of
over 250,000.
Stratum C—SMSA's with urban population of 50,000 to
250,000.
Stratum D—-Nonmetropolitan urban places of 2,500 to
50,000 population.

One merit of this classification is that the
first three strata correspond closely to the metro­
politan segment, and the last to the nonmetropolitan segment of the urban population. In
addition to the SMSA's which had already been
defined for the 1960 Census, other cities which
were estimated to have passed the 50,000 popu­
lation mark and which were designated as "po­
tential" SMSA's, were classified in the C
stratum. Since the few additional SMSA's that




might be established in the 1960 Census count
could not have been identified in advance, they
were classified in the D stratum.
Alaska and Hawaii posed special problems.
Although their urban population did not justify
the allocation of a sample city to each, they are
so different from the continental (48) States,
and from each other, that there appeared to be
no alternative to making each a separate stra­
tum with a sample place for each. The urban
population of Alaska is concentrated in the areas
of Anchorage, Fairbanks, Juneau, and Ketchi­
kan. Anchorage was randomly selected to rep­
resent Alaska in the CPI. Since over 86 percent
of Hawaii's urban population is concentrated in
the Honolulu SMSA, Honolulu was designated
as the sample city to represent the total
Hawaiian urban population.
Developing the CPI Sample
The certainty selection of the 12 largest cities
and the allocation of one sample place each to
Alaska and Hawaii left 36 cities to represent
the B, C, and D strata urban places in the other
48 States. These 36 were divided among strata
on the basis of the relative importance of their
urban population and the estimated annual costs
of operating a pricing program in cities of dif­
ferent size. The resulting optimum allocation is
shown in the following tabulation:
All
three
strata
36
7
li
13
5

Stratum
B
10
2
3
3
2

C
10
2
3
4
J

D
16
3
5
6
2

An important objective in selecting the spe­
cific cities was to achieve a good geographic dis­
persion. Thus, minimizing the possibility of an
undue concentration of the sample in any State
was particularly important because, for many
items, price factors are closely related to local
conditions. After considerable consultation and
experimentation, the BLS decided to utilize the
procedure usually referred to as "controlled
selection.,, m A significant advantage of this
55
An independent probability selection for each size group was
considered and discarded because it was impossible to prevent the
selection of sample cities from the several strata in the same State
where the size of the State did not warrant such extensive represen­
tation. Another method considered was a procedure whereby the
sample for all size strata could be selected by one systematic opera­
tion from an array of all primary sampling units. This method was
discarded because of the distortion of the original probabilities of
selection associated with individual cities.

35

method for the type of sample being selected is
the provision for rigorous geographic controls.
Other advantages are that it is a probability
method in which the assigned probabilities of
selection for individual cities are demonstrably
maintained, that it is a tested and reputable
system which has been in use for a number of
years by other organizations, and that it is de­
scribed in the published literature.56
The controlled selection procedure involves
the probability selection of a sample "pattern"
from a set of patterns which have been purposively established so that, taken as a group,
they give to each primary sampling unit its
proper chance of appearing in the final sample.
Each pattern is set up in accordance with con­
trols, which may be as rigid as desired, to insure
that it satisfies selected criteria of proper distri­
bution. In selecting the CPI sample, controls
were used only on size of city and geographic
location, with the latter control carried to the
State (or group of small States) level.
In order to expedite the work, patterns were
established for each of the four broad regions
of the country separately. (See, for example,
table 1.) Time did not permit the more elaborate
control that could have been maintained by se­
lecting national patterns.
The first step was to determine the proba­
bilities of selecting a given size city from each
stratum within each State or group of States.57
These probabilities are based on estimated
urban population figures previously described.
The sum of the probabilities over all States in
the region for any size group is equal to the
number of sample places allocated to that stra­
tum. The set of patterns which are derived must
exhaust all these selection probabilities exactly.
The probabilities also set limits as to the pat­
terns which may be chosen. For example, the
following illustrative tabulation for the North­
east shows that no pattern can contain more
than one sample place each in five of the States
in the region, but the other two (New York and
Pennsylvania) can have either one or two selec­
tions each.
M
The method chosen is described by Roe Goodman and Leslie Kish
in "Controlled Selection—A Technique of Probability Sampling,"
Journal of the American Statistical Association, September 1950, pp.
350-872.
w
In 5 instances* 2 or more of the less populous States were
grouped together and treated as a single State: Maine, New Hamp­
shire, and Vermont; Nebraska, North Dakota, and South Dakota;
Colorado and New Mexico; Arizona, Nevada, and Utah; and Idaho,
Montana, and Wyoming.

36



Probability of selection of cities in—
B
stratum
Maine-New HampshireVermont
Massachusetts
Rhode Island
Connecticut
New England
New York
Pennsylvania
New Jersey
Middle Atlantic
Northeast

C
D
stratum stratum

1.133

.285
.859
.225
.728
188
1.141

0.544
.312
049
.293
1.198
.757
.761
.284
1.802

0.697
.980
.305
.942
2.924
1.870
1.734
.472
4.076

2.000

2.000

3.000

7.000

0.247
256
.364
867
888
245

0.163
.421

B,C,and
D strata

In addition, a further control was used to in­
sure representation of the New England States
as a group and of the Middle Atlantic States as
a group. In most cases, the patterns must con­
tain three selections from New England and
four from the Middle Atlantic region, although,
at least one pattern must contain two and five,
respectively, in order to exhaust the 0.076 proba­
bility of having only two selections in New Eng­
land.
The patterns were set up by State-size groups
rather than individual places because the addi­
tional work was not feasible. Instead, a second
stage of controlled selection was used in States
where more than one sample selection fell.
Table 1 presents a simplified diagram of the
patterns set up for the Northeast. Many other
possible sets would satisfy all the conditions
imposed. The probability assigned to each pat­
tern is usually the smallest remaining proba­
bility of any category affected by that pattern.
Thus, pattern 1 was given the probability of
0.076 to dispose of the previously mentioned
probability of having two selections in New
England. Together, these patterns do, in fact,
exhaust the original probabilities. For example,
a B stratum city in Connecticut appears in pat­
terns 2, 5, 7, 9, 11, and 12. The probabilities
associated with these patterns add to .364, the
selection probability for that category.
Sets of patterns for the North Central region,
the South, and the West were established in the
same manner. An attempt was made to secure
additional geographic dispersion by combining
patterns for adjacent regions in such a manner
as to prevent an undesirable national pattern.
For example, patterns for the North Central
region and the West were associated so that
there was no chance of a national pattern with
no selection in Montana-Idaho-Wyoming which
also omitted a selection for North Dakota-South

Dakota-Nebraska. After patterns were estab­
lished, the four regional patterns were chosen
in a single operation by a random process. Pat­
terns were similarly selected for the two alter­
nate samples.
At this stage, specific cities were determined
only where there was a single city in the se­
lected State-size group (e.g., Denver, the only
B city in Colorado-New Mexico). In other cases
where only one selection from a State was in­
volved, the sample places were randomly selected
from arrays of all primary sampling units in
the State-size groups indicated in the selected
pattern. In New York, Ohio, and Texas, where
there were two or more sample selections and
more than one city in each selected group, a
second-stage combination was randomly se­
lected which determined the B and/or C
stratum selections, and the D places were ran­
domly selected from arrays for the designated
zone of the State.
Two alternate samples were selected in a
similar fashion and are shown with the basic
sample in table 2. Within each stratum, the
cities are arranged by geographic region, and
the year shown is that in which the Consumer
Expenditure Survey in the city was conducted,
as indicated later. Two cities, Denver and
Seattle, are duplicated in the basic and second
alternate samples. Since it was intended that
all three samples be equally representative,
some such duplication was almost unavoidable.
Consumer Expenditure Survey Sample
Analyses of consumer expenditure data from
previous surveys indicated a much higher vari­
ability in expenditure patterns among small
places than among large cities. Consequently,
the resources available to supplement the 50
city CPI sample were allocated to the D
stratum, permitting the sample size in this
stratum to be doubled (to 32 cities for the ex­
penditure surveys). The 16 additional places
were obtained by taking the D cities in the first
alternate sample.
Since the Consumer Expenditure Surveys
were to be conducted in two "waves" (in 1961,
covering 1960 expenditures; in 1962, covering
1961 data), it was necessary to divide the sam­
ple into two balanced subsamples, each repre­
sentative of the U.S. urban population. In the




12 large A cities, surveys were conducted in
both years. The B, C, and D places were alter­
nately assigned to the two subsamples, with a
random assignment of the first city in each size
group, with the places arrayed in their order
of selection (as in table 2). The 16 D cities from
the first alternate sample were similarly divided,
but with the starting assignment reversed from
that used in the D cities of the basic sample,
in order to balance each region. Because of the
special price program in Alaska, the expendi­
ture survey in Anchorage was conducted in May
and June 1960 and, with some adjustment,
served as the basis for the CPI weights. The
Honolulu survey was made in 1962. Thus, it was
not possible to have balanced representation of
Alaska and Hawaii in the 2 years of the survey
period as was the case for the rest of the United
States.
Tests of the CPI Sample
In order to test the effectiveness of the new
CPI city sample, data for sample places were
used to make estimates of several characteristics
for which data were available for the total U.S.
urban population. Among these were: 1950
urban population, 1950-58 population change,
1958 retail sales, and total 1958 population (in­
cluding farm) for the A, B, and C strata. The
differences between the actual and estimated
data are summarized in table 3. The U.S. esti­
mates are quite accurate, although some re­
gional and city-size subtotals vary fairly widely
from actual figures.
The city "weights" used in the above tests
were derived so as to provide unbiased esti­
mates. This required that the weight for each
sample city be the reciprocal of that city's orig­
inal probability of selection. These weights (and
an analogous set for the enlarged expenditure
survey sample) were incorporated into the
weighting structure for the revised CPI com­
putation procedures and for various estimates
of consumer expenditures made from the sur­
veys.
It is not possible to compute an exact sam­
pling error for the design described above. How­
ever, approximate methods can be used. For ex­
ample, it is considered that the controlled selec­
tion procedure as used in making the 36
probability selections is roughly equivalent to
37

establishing 36 strata and selecting one city
from each. The replication feature built into the
revised index procedures includes an ex post
TABLE i.

Maine-New HampshireVermont. „

2
7
12
13
16
4
5
10
1
3
15
8
9
6
11
14
.076 .020 .029 .029 .054 .072 .061 .145 .006 .082 .140 .083 .030 .027 .034 .112

.153
,544

B
C
D

.247
,421
.312

B
D

.256
.049

«.. . B
C
D

.364
.285
.293

X

B
C
D

.888
.225
.757

X

X

X

X

B
C
D

.245
.728
.761

X

C
D

.188
.284

X

Rhode Island

New York.

Pennsylvania
New Jersey

TABLE 2.

Pattern number and associated probability

C
D

Massachusetts

Connecticut..

ILLUSTRATIVE SET OF PATTERNS FOR THE NORTHEAST

Iff

Size
stratum

State group

facto pairing of probability cities, which is
equivalent to the common practice of "collatising" strata in order to compute variances.

X

X

X
X

X

X
X

X

X

X

X
X

X

X

X

X

X

X
X

X
X

X

X

X

X
X

X

X

X

X
X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

' X

X

X

X

X
X

X

X

X

X
X
X

X
X

X

X

X

X

X
X

X
It

X
X

X

X

X

X

X

X

X

X

X

X

X
X

X

X

X

X
X

X
X

X

X

X

X

X

X
X
X

X

X
X

BASIC AND Two ALTERNATE CITY SAMPLES FOR THE CONSUMER PRICE INDEX AND CONSUMER EXPENDITURE
SURVEYS
[Years shown for certain cities are those in which expenditure surveys will be conducted there!
Basic sample

Alternate 2

Alternate 1
Stratum A—Twelve largest SMSA's

New York, N.Y
Philadelphia, Pa
Pittsburgh, Pa
Cleveland, Ohio
Detroit, Mich
Chicago, 111
St. Louis, Mo
Baltimore, Md
Washington, D.C
San Francisco, Calif
Los Angeles, Calif

.
„..

1961, 1962
1961,1962
1961,1962
1961, 1962
1961,1962
1961,1962
1961, 1962
1961, 1962
1961, 1962
1961, 1962
1961, 1962
1961, 1962

No alternates; the 12 cities in the basic sample are the only ones of this sise.

Stratum B—Other SMSA's with urban population of over 250,000
Hartford, Conn
Buffalo, N.Y
Dayton, Ohio
Wichita, Kans
Atlanta. Ga
Nashville, Tenn
Dallas, Tex
Denver, Colo
Seattle, Wash
Honolulu, Hawaii

..1962
1961
1962
1961
1962
1961
1962
1961
1962
1961
1962

New Haven, Conn.
Syracuse, N.Y.
Toledo, Ohio.
Milwaukee, Wis.
Kansas City, Mo.
Norfolk-Portsmouth, Va.
Louisville, Ky.
Houston, Tex.
Phoenix, Ariz.
San Diego, Calif.

Providence, B.I.
Albany-Schenectady-Troy, N.Y.
Cincinnati, Ohio.
Grand Rapids, Mich.
Minneapolis, Minn.
Miami. Fla.
New Orleans, La.
El Paso, Tex.
Denver, Colo.
Seattle, Wash.

Stratum C—SMSA's with urban population of 50,000 to 25,0000
Portland, Me
Lancaster, Pa
Champaign-Urbana, 111
Green Bay, Wis
Cedar Rapids, Iowa
Durham, N.C
Orlando, Fla
Baton Rouge, La_>_
Austin, Tex
Bakersfield, Calif-




1961
1962
1961
1962
1961
1962
1961
1962
1961
1962

Utica-Rome, N.Y.
Altoona, Pa.
Fort Wayne. Ind.
Muskegon, Mich.
Sioux Falls, S.D.
Huntington, W. Va.
Charlotte, N.C.
Montgomery, Ala.
Waco. Tex.
Spokane, Wash.

Lawrence, Mass.
Scranton, Pa.
Evansville, Ind.
Madison, Wis.
Davenport, Iowa-Rock Istand-Moline, 111.
Charleston, W. Va.
Winston-Salem, N.C.
Mobile, Ala.
Corpus Christi, Tex.
Fresno, Calif.

X
X
X
X

TASLE 2,

BASIC AND TWO ALTERNATE CITY SAMPLES FOR THE CONSUMER PRICE INDEX AND CoNbUMER EXPENDITURE

SURVEYS—Continued
Stratum D—Cities of 2,500 to 50,000
Kingston, N.Y
MiUviUe, N . J . . .
Nttes, Mich-

„
-

......

Orem, Utah.
Khu»*tt* Fftllft, Or«,
Anchorage, Alaska a

...

1962
1961
1962
1961
1962
1961
1962
1961
...1962
1961
1962
1961
1962
1961
1962
—1961

Athol, Mass
.
Lewist own, Pa
Cambridge, Ohio
La Salle, 111
Menasha, Wis

.

.1962
.1961
.1962
1961
.1962
.1961
..1962

.

Reserve, La
Eureka, Calif

„

1
Consumer expenditure surveys in the first alternate sample of stratum
D cities were used in deriving C P I weights for smaller cities. However, no
pricing it done in these supplemental cities.




.1962
.1961
.1962

.

Manhattan, Kans
Griffin, Ga
Sebring, Ma
Cleveland, Tenn

I

1962

Danbury, Conn.
Horseheads, N.Y.
Bridgeton, N.J.
Crestline, Ohio
LaPorte, Ind.
Carbondale, 111.
Maryville, Mo.
Larned, Kans.
Sanford, N.C.
Whitmire, S.C.
Albany, Ga.
Dyersburg, Tenn.
Russellville, Ark.
Paris, Tex.
Laramie, Wyo.
Modesto, Calif.

3
Anchorage represents Alaska in the Revised CPI. The consume!
expenditure survey was conducted there in May and June 1960 as part of a
special Alaskan price program. These surveys were used to derive the
index weights for the city.

TABLE 3. PERCENT BY WHICH ESTIMATES OF U.S.1 POPULA­
TION AND RETAIL SALES MADE FROM THE N E W OPT CITY
SAMPLE DIPPER FROM ACTUAL FIGURES
1950-58
Total 1958
1950 urban population
Retail population
population change
sales, 1958 CA-B-C
cities)

Region and citysize group
Region
Northeast
North Central
South
West

...
...

-0.9
-2.4
+1.8
+6.S

Stratum A . .
Stratum B
..
.
Stratum C
Stratum D
l
United States . . . .

+0.4
-3.2
+3.6
+0.4

,

+8.9
+14.2
-6.2
-9.5

+2.0
+2.4
-1.1
-5.7

+2.8
+2.0
+7.3
+0.5

C3)
+6.3
+11.1
-13.7
(3)

(*)
+ 1.1
+J4.1
+3.2

City Sixe

(*>
-1.3
+12.8
-21.2
-1.9

*
48 States only.
8
No difference since all SMSA's in this si*? group are included in the
sample,
s
Less than 0.05 percent.

39

Chapter VI. Housing and Expenditure Surveys
Surveys of Consumer Expenditures, Income,
and Savings (CES) were conducted in 66
cities58 for the years 1960-61. The results of the
surveys determined the content of the market
basket and the quantity weights of the revised
index. The CES samples of consumer units were
chosen in most cities as subsamples of addresses
listed in housing surveys.59
Comprehensive Housing Unit Surveys
Comprehensive H o u s i n g Unit Surveys
(CHUS) preceded the expenditure survey in
each of 34 A, B, and C strata areas listed in
appendix table IV, and eight of the D cities
included in the index and were conducted sub­
sequently in the remaining eight D cities in
the CPI. The surveys in these areas were con­
ducted by personal visit to a selected sample
of housing units in the year preceding the
actual field survey work for the CES. Hous­
ing survey objectives included obtaining a
sample listing of all living quarters in sam­
ple blocks, complete descriptions, income, and
race for a sample of housing units within the
sample blocks, rental data for tenant-occu­
pied units, and market value and purchase price
for owner-occupied units. These surveys pro­
vided rental and CES samples and a sample of
owner-occupied units for the measurement of
property taxes. Information obtained was also
used in deriving weights for home purchase.
The CHUS covered the entire urban portion
of sample SMSA's including the central city or
cities, the urbanized areas surrounding the cen­
tral cities, and noncontiguous urban places
within the SMSA. For non-SMSA's, the survey
area was the city proper. The CHUS was de58
One of the 66 cities, Anchorage, had been surveyed for 1959 in
connection with a special program in Alaskan cities. Six additional
large cities which did not fall in the national sample were added to
the national index in January 1966. One of these, Cincinnati, was
the pilot city for the revision program, surveyed for 1959. The re­
maining 5—Houston, Kansas City, Milwaukee, Minneapolis-St. Paul,
and San Diego were surveyed in 1963.
59
Housing surveys were not conducted in 16 non-CPI cities; pro­
cedures for selecting CES samples in these cities are described on
p. 43.

40




signed to cover all noninstitutional living quar­
ters, including nontransient accommodations in
hotels and rooming houses.
The final boundaries of the areas had not been
established in all cases in time for planning the
surveys to be conducted in the latter part of
1960, but maps were obtained from the Bureau
of the Census on which preliminary boundaries
were indicated. Some changes, usually minor,
were made later but the discrepancies between
the sampling frame established for the CHUS
and final Census urban boundaries were rela­
tively minor. For the CHUS conducted late in
1961, more nearly final Census maps were used.
The size of the CHUS sample was determined
primarily by the size of the rental sample de­
sired and the proportion of renters in a given
area, as estimated from Census data. The num­
ber of addresses enumerated in the CHUS was
usually many times larger than the CES sam­
ple.60 The overall CHUS city sampling ratio was
determined by dividing the estimated number of
rental units in the city by the size of the rental
sample desired. This sampling ratio was applied
both to tenant and owner-occupied dwelling
units, thus maintaining tenants and homeown­
ers in the proper relationship.
Appendix table IV shows for each city the
number of units listed in the CHUS, the total
number of units in the CES, the number of
usable CES schedules, and the number of wageearner and clerical-worker consumer unit
schedules which were employed in deriving
weights for the index.
Slightly different procedures were used to
sample (a) the central city or cities of SMSA's,
(b) the urban fringe, and (c) outlying urban
places within the SMSA but outside the Census
"urbanized area." For each of the various
sampling areas, the product of all sampling
60
In all, about 130,000 dwellings were enumerated in the CHUS
compared with slightly over 10,600 CES assignments (addresses) and
an equal number of alternate assignments. (Excluded from both
counts are 24 small urban places where the CES samples were selected
from Census records.)

ratios used (block and in-block) equaled the
overall city sampling ratio.
A two-stage design was used in the central
cities; a sample of blocks was chosen and a subsample of addresses selected within blocks. A
classification of blocks by size was incorporated
in the design with variable block (and in-block)
sampling fractions for large and small blocks
(based on number of housing units) and for
apartment and nonapartment blocks. For all
SMSA's surveyed in 1960, except ChampaignUrbana, which was not a Standard Metropoli­
tan Area in 1950, 1950 Census Block Statistics
Books provided the basis for selecting a proba­
bility sample of blocks within the central cities.
For SMSA's surveyed in 1961, preliminary
copies of the 1960 Census Block Statistics Books
were used.
In all D size cities surveyed in 1961, as well
as in Champaign-Urbana, a complete BLS field
survey was made first for identification and
classification of all blocks within the survey
area, using central sources or personal observa­
tion of assigned areas, after which a selection
of blocks was made.
In the urban fringe, a first-stage selection of
Census enumeration districts (ED's) was made
with subsequent sampling of blocks (or seg­
ments) and of addresses within blocks using
the BLS field survey procedure for non-SMSA
cities. An apartment block nonapartment block
classification was also used here.
Outside the Census urbanized area, a firststage sample of urban places, with subsequent
sampling of blocks and of addresses within
blocks, was made according to the following
scheme:61
1. All places of 10,000 or more selected with certainty.
2. All places less than 10,000, sample selected as follows:
a. 1-6 places, 1:1 ratio.
b. 7-14 places, 1:2 ratio.
c. 15-29 places, 1:3 ratio.
d. 30 or more places, 1:4 ratio.
61
For example, there were 20 urban places in the St. Louis SMSA
which were outside the St. Louis urbanized areas as defined for the
CHUS. Five were over 10,000 population and were selected with cer­
tainty to represent themselves: St. Charles, Mo., and Alton, Collinsville, Edwardsville, and Wood River, 111. A sampling ratio of 1:3
was applied to the 15 places under 10,000 and the following places
selected: Ellisville and DeSoto, Mo., Cottage Hills, East Alton, and
Mascoutah, 111. Two additional stages of sampling were then used
in these 10 sample places. (The final Census urbanized area for St.
Louis included Collinsville, 111. Had this definition been used in the
CHUS, Collinsville would have been sampled as part of the urban
fringe, rather than as an outlying urban place. No change in the
total coverage was involved.)




However, in three SMSA's, surveyed in 1961,
which had 7 to 14 urban places below 10,000
outside the urbanized area, all urban places
were selected, as a matter of convenience.
For the entire survey area, sample enumera­
tion districts and blocks were numbered and
indicated on the Census maps either in the
Washington office or in the field survey. Block
boundaries were carefully described on a control
form for each block in the sample, together with
sampling instructions.
All separate living quarters in sample blocks
including private homes, apartment houses, and
hotels and rooming houses except low-income
public housing, were listed merely by street
name and address by observation, for purposes
of the CES sample. Hotels and rooming houses
were classified as transient or nontransient de­
pending upon whether or not more than half
the rooms were rented to transients. Lists of
low-income public housing were obtained from
central sources.
To the list of addresses excluding transient
structures, the appropriate in-block sampling
ratio was applied and detailed information ob­
tained for units falling in the sample, by per­
sonal interviews with occupants. If a transient
unit within a nontransient structure fell on
ratio, the next unit was selected. Housing units
that fell on ratio were classified by type of hous­
ing unit:
Type

Definition

1

With separate entrance and kitchen facilities
installed.
With separate entrance and kitchen facilities not
installed.
With separate entrance and no kitchen facilities.
Without separate entrance and with kitchen
facilities.

2
3
4

A housing unit is defined as a room or rooms
occupied or intended for occupancy as separate
living quarters by a family or other group of
persons living together or by a person living
alone, having a separate entrance and/or sepa­
rate kitchen (exclusive use by occupant) facili­
ties. Detailed information was obtained on the
schedule (BLS 2549 appended as exhibit A)
for Type 1 and 2 housing units. A Type 1 hous­
ing unit is defined as a room or rooms occupied
or intended for occupancy as separate living
quarters by a family or other group of persons
living together or by a person living alone, hav­
ing a separate entrance and separate installed
41

(or provision for) kitchen facilities for the ex­
clusive use of the occupants. The separate en­
trance need not be directly from the street but
may be from a common hall, as in apartments,
or even in private homes provided access is not
through a room or any other separate living
quarters. A Type 2 is the same but without in­
stalled kitchen facilities.
Information obtained in the survey included
address, type of living quarters, description of
structure including type, number of units, year
built, condition of the unit and neighborhood
characteristics as judged by the BLS agent,
occupancy status, tenure, race of occupants,
number of persons in housing units, a rough
estimate of family income, and detailed descrip­
tions of the facilities in the unit; also for rental
units, rent paid as of the date of the survey and
information as to the facilities and services in­
cluded in the rent; and for owner-occupied
units, date of purchase, purchase price, and
estimated current market value. Only limited
information was collected for Type 2 housing
units—type of structure, number of units, date
structure built, condition of unit, occupancy
status, tenure, race, number of persons in the
unit and family income class. No information
other than address and family income class was
obtained for Types 3 and 4.
Field collection was made by a group of
agents hired, trained locally, and supervised
by a supervisor who had been trained in Wash­
ington. Quality control checks were made of
each agent's work during the course of the sur­
vey and completed collection forms were re­
viewed in thefieldfor consistency and complete­
ness before transmission to Washington.
Consumer Expenditure Surveys (CES)
The expenditure surveys were generally com­
parable in scope and in survey methodology to
the 1950 expenditure survey.02 Although pri­
mary emphasis was on collecting data relating
to family expenditures for goods and services
used in day-to-day living, information was also
obtained on family income, change in savings
and debts, and on major demographic and eco68
For a description, see account by Helen Humes Lamale, Study
of Consumer Expenditures, Income and Savings—Methodology of the
Survey of Consumer Expenditures in 1950, (Philadelphia, Pa., Uni­
versity of Pennsylvania, 1959, pp. 41-81). A brief description of the
1960-61 survey is included in each report in the BLS Report 23?
series.

43



nomic characteristics of the consumer units.
The collection schedule used in the 1960-61 sur­
veys was adapted from the 1950 form and was
tested in a pilot survey conducted in Cincinnati
as well as in two Alaskan cities in the spring of
1960. Only minor changes were made subse­
quently.
The Consumer Expenditure Surveys covered
2 years, 1960 and 1961. About half the sample
was surveyed early in 1961, covering expendi­
tures during 1960. The remaining sample was
surveyed early in 1962, covering expenditures
during 1961. Spreading the survey over 2 years
provided a hedge against the possible abnor­
mality of spending patterns for a single year.
Operationally, the 2-year program had several
major advantages in terms of efficiency and
economy, since it was possible to reduce the size
of the staff to be recruited and trained and to
use them over a longer period.
In the 12 largest SMSA's (including the New
York and Chicago Standard Consolidated
Areas) half of the sample was covered in each
of the 2 years. In the other size strata, half of
the cities were covered in 1 year and half in
the other. Since Anchorage (surveyed for 1959
in connection with the special program for
Alaska)63 was the city selected to represent
Alaska, results from the 1959 survey were uti­
lized for the CPI weights. Honolulu was sur­
veyed for 1961.
No attempt was made to make the CES sam­
ples self-weighting, except within each SMSA
or urban place. The total urban sample included
about 12,000 assignments. Because of the indi­
vidual city indexes published for the 12 "A"
cities and for many of the "B" cities, a mini­
mum sample size in each city (area) was re­
garded as desirable. The sample in the "A"
cities ranged from 375 to 625, the sample in all
"B" cities was 250 assignments, "C" city sam­
ples 160, and "D" cities 65.
In order to select the CES sample from the
larger CHUS sample, the CHUS units were ar­
rayed by type of housing unit and location (cen­
tral city vs. suburb). Types 1 and 2 housing
units were then arrayed by race, income, and
size of family—variables known to be important
63
Funds were provided BLS under special legislation to compute
time-to-time indexes semiannually in four Alaskan cities—Anchorage,
Fairbanks, Juneau, and Ketchikan—and place-to-place comparisons
with Seattle, Wash. Family expenditure surveys were conducted for
1959 for Anchorage and Fairbanks and for 1960 for Juneau and
Ketchikan,

in their influence on consumption patterns.
From these arrays, a systematic selection was
made by choosing a random starting point and
selecting every n-th unit thereafter down the
array (n being selected to give the desired sam­
ple size). The CES units from the CHUS were
supplemented by a sample of units in lowincome public housing projects.
For each sample unit selected, a "matched
alternate" was also designated. The alternates
were as closely matched as possible with the
original selection, primarily in terms of size of
family and income level. This was done by
selecting the next address in the array. Because
of the original sequence, by location, in which
the enumerated units were numbered the alter­
nates also tend to be in the same part of the
area. These alternates were used only under
certain specified conditions: If the occupant of
the original unit could not be contacted after at
least two visits, if he refused to give even a
minimum amount of information required to
analyze nonresponses, or if the unit was vacant.
For the eight CPI and eight non-CPI "D"
cities surveyed for 1960, a somewhat different
procedure was used, primarily as an economy
measure. (See appendix table IV for list of
cities.) Samples of about 500 addresses per city
were selected from enumeration schedules used
in the 1960 Censuses.64 A double sampling pro­
cedure was used here also, to make use of Cen­
sus information on family characteristics, but
no information other than the addresses of the
selected units could be transcribed from Census
records for subsequent use by the BLS. The
same procedure was followed for the eight "D"
cities surveyed for 1961 which were not in the
CPI sample. In the remaining eight CPI " D "
places, housing surveys were conducted subse­
quent to the CES, since they would have soon
become necessary in connection with initiation
of rent pricing.
All data were collected through the voluntary
cooperation of families. Since the Bureau has
a strict policy regarding confidentiality of the
data, the cooperating family's name never ap­
pears on any records.
All of the information for the survey was ob­
tained by personal interview by a BLS agent
64
This was made possible by the cooperation of the Bureau of the
Census. Procedures were followed which avoided violation of Census
confidentiality restrictions.




hired and trained locally by a supervisor who
had had an intensive 6-week training course in
Washington. Total interview time with a single
family normally averaged 7 to 8 hours, spread
over several visits with one or more family
members, and arranged at the family's conveni­
ence. Although all consumer units 6 5 were in­
terviewed for the CES, data for wage-earner
and clerical-worker units 66 were tabulated sepa­
rately for purposes of index weights.
A complete account of receipts and disburse­
ments was collected for the preceding calendar
year. The estimated value of goods and services
received free (as gifts, public assistance, etc.),
the characteristics of housing occupied both
by homeowners and renters, and an inventory
of major items of housefurnishings owned was
requested also.
[,j
The survey schedule is a 76-page question­
naire in three parts. The first part is a two-page
Household Record Sheet, which lists all family
members and their relationship to the head of
the household and provides the basis for deter­
mining the family's eligibility for the survey.
It contained minimum data for analysis of nonresponse. The second part of the schedule, con­
sisting of 59 pages, contains 23 major sections
and a complete annual record of income, expend­
itures, and changes in savings. The individual
sections are sequenced in a logical order de­
signed to establish and maintain rapport be­
tween the interviewer and the respondent. The
detailed checklists of items printed on the sched­
ule were arranged both to provide for clearly
recording the specific information needed and to
facilitate recall. All sections are rarely applic­
able to a single family. Families were encour­
aged to refer to records whenever possible and
were interviewed to clarify ambiguous entries,
when necessary. The third part of the schedule
(for families who prepared meals at home)
covers detailed expenditures, during the 7-day
period immediately preceding the date of the
interview, for food and beverages, household
supplies, and tobacco, which are frequently pur­
chased items. Similar weekly data were subse68
The family, or consumer unit, refers (1) to a group of people
usually living together who pooled their income and drew from a
common fund for their major items of expense, or (2) to a person
living alone or in a household with others but who was financially
independent, i.e., his income and expenditures were not pooled.
66
The definition of wage earners and clerical workers is based on
the occupational classification used by the Bureau of the Census for
the 1960 Census of Population. It is described in chapter m .

48

quently obtained by mail questionnaire for addi­
tional periods during the year in order to obtain
an indication of seasonal variation in consump­
tion. These data are discussed in chapter VII.
Results of the survey were edited carefully
and processed by machine tabulation. They were
presented in a series of reports for each city
and region, showing dollar values (income, ex­

44



penditures, and changes in savings) for each
city as averages per consumer unit and cross
classifications by various family characteristics.
Average annual expenditures for all wageearner clerical-worker consumer units 67 be­
came the basis for index weights, as described
in chapter VII.
OT

These data have not been published.

Chapter VII. Weighting Structure of the CPI, 1964
The scope of the revised CPI with respect
to the definition of items represented and the
population and geographic coverage are dis­
cussed in chapter III, "Statistical and Concep­
tual Structure of the Revised Consumer Price
Index." This chapter describes the selection of
the sample of items priced and the derivation of
weights used beginning January 1964.
The weighting structure is complex. It re­
flects varying expenditure patterns (item
weights) in each city, which are weighted aver­
ages of different patterns for families and for
single consumer units. In many cities, there are
separate weights for the two independent subsamples of outlets set up under the replication
design described earlier. Population weights are
used in each city for weighting expenditure pat­
terns for families of two persons or more and
single consumer units and for combining the
individual cities for the U.S. index. In some
cases specific internal weights are assigned to
different specifications of the same item. Be­
sides the item and population weights, there are
internal weights for combining individual quo­
tations for different areas of the city and indi­
vidual stores in the sample, as described in
chapter VIII. In some cases, instead of assign­
ing specific weights, the various samples are
made self-weighting, i.e., data were stratified
for sample selection, and random selections
made and given equal weights.
Down to the expenditure class (EC) level,
the basic item weighting structure of the new
index remains much the same as it was prior to
the recent revision. As a general rule, weights
for major groups, subgroups, and expenditure
classes are expenditures incurred by index con­
sumer units during 1960-61, as observed in the
CES, but adjusted for price change to Decem­
ber 1963, the link month for the revised index.
Consumer units having zero expenditures are
included in the average along with those incur­
ring expenditures. Weights were derived sepa­
rately for families and for single persons for




each of the 50 cities and combined with popula­
tion weights described later. In some cases,
because of the small size of CES samples result­
ing in a large sampling error for the individual
items, category totals were distributed to sub­
groups on the basis of an average of cities of
similar size within the same region.
Item Sampling Frame
The item coverage includes all goods and serv­
ices purchased for family living. For purposes
of the CPI, the thousands of goods and services
purchased by consumers are organized into a
logical system of groups and subgroups which
serve as the publication framework and the
sampling frame for selection of items and
derivation of weights.
The final item sampling frame (appended as
Exhibit B) developed from extensive discus­
sions beginning early in the revision program
and from experiments with preliminary listings
containing more items and more detailed ex­
penditure classes. The development of a classi­
fication system and of a listing of items making
up the sampling frame proceeded simultaneous­
ly and almost inseparably until the final stages.
Because so many items have multiple uses
and serve several end purposes, no single tech­
nical criterion, such as form or function, pro­
vides a satisfactory way of grouping. There­
fore a dual classification was used. The basic
classification system is governed by that of the
usual family budget. Superimposed upon the
primary use or consumption base is a subdivi­
sion into an economic classification of goods and
services, of durable and nondurable goods, etc.
The lowest grouping level in the classification
scheme is termed the "expenditure class (EC)."
This level defines the index market basket to be
held constant between major revisions, but
within which resampling and adjustment of
weights may take place between revisions.
45

The statistical connotation of "item" in the
sampling frame is different from any definition
previously used by the Bureau. It was estab­
lished primarily as the level at which probabil­
ity sampling would be applied. Almost always it
is above that of the priced specification. Gen­
erally the term applies to a grouping of goods
(or of services) of similar physical character­
istics but of different qualities which serve the
same end purpose. However, the items are not
of equal homogeneity in the different classes.
Some of them are groupings of conglomerate
items, each member of which is too trivial to be
given a separate chance of selection for pricing.
Others are specified-in-detail items. In sampling
terms, the items as defined are not elementary
sampling units but clusters or first-stage sam­
pling units.
One of the earliest phases of research con­
cerned the most efficient methods of item strati­
fication into expenditure classes. For this analy­
sis, a large body of price data was utilized, in­
cluding not only price data collected for the CPI
but other data collected in an experimental pric­
ing program. The research included studies of
the variance in price movements, ranking price
changes for individual items for various periods
and combining these into a composite ranking,
and the computation of correlation matrices for
groups of items in order to define strata com­
prising items which are highly correlated in
terms of similarity of price movements. Results
of this investigation indicated that the amount
of possible meaningful stratification beyond that
necessary for publication requirements, and be­
yond what could be effected through judgment
or common sense, was quite unimportant. Con­
sequently, the expenditure classes, which define
the sampling strata, were determined primarily
by grouping items which in a general way serve
similar human needs; for example, furniture,
fuels and utilities, men's apparel, women's ap­
parel, etc.
Within the EC framework, items were listed
in an order which provided some additional im­
plicit stratification by type of item. Items
grouped together were as homogeneous as pos­
sible with respect to physical characteristics,
although not necessarily with respect to price
movements. This was in contrast to the system
of weighting set up in the 1953 revision, be­
cause the attempt made at that time to group
items into "price families," based on known or
46



expected similarity of price movements, had
proved unsatisfactory.
There were about 1,800 line items in the CES
schedule. The first proposals for the classifica­
tion system included from 62 to 90 expenditure
classes and upward of 1,400 items. The items
mostly represented individual CES line entries
or combinations of two or more such entries,
although in a few cases, important line items
were broken down into two or more separate
items on the basis of secondary data.
Expenditure data from the 1959 pilot expend­
iture survey conducted in Cincinnati were used
in the experimental work on the item sample
design. This experience and analysis of the ex­
penditure data indicated that the first classifica­
tion systems proposed carried too much detail
in some categories. For example, in one system
of 75 expenditure classes, 25 had less than 0.5
percent of the total relative importance. Since
each EC would require allocation of at least one
sample item for pricing in each replicated subsample,68 retention of all 75 EC's would have
resulted in an inefficient use of resources. Many
items also had negligible expenditures. Succes­
sive consolidations of EC's and items resulted
in the final sampling frame containing 52 EC's
and 812 items shown in Exhibit B. The list of
EC's and the number of items in each are sum­
marized in appendix table V.
Late in the revision program a change in the
method of pricing health insurance resulted in
the consolidation of two EC's. The identities of
the 52 EC's were retained in the CPI weighting
structure, however.
Selection of Items
As in past revisions of the CPI, the samples
were selected dn a national basis. Selection of
independent samples, city-by-city, was not prac­
tical from a cost standpoint since it almost cer­
tainly would have resulted in a long list of
items, each of which would be priced in at least
one city, but by no means in all cities.
The samples of items were selected with
"probability proportional to size," size being
defined as the relative importance of the ex­
penditures for the item to total expenditures for
all items within an expenditure class (EC).
98
See chapter IV, "Sampling Aspects of the 1964 Revision/' for
an explanation of the replication design.

Initially, the size of the item sample could not
be exactly specified due to uncertainties of avail­
able resources, costs of price collection for new
items, and cost increases associated with the
replicated samples of items and, ultimately, of
outlets. Consequently, in the developmental
work, a series of "dry run" sample selections
were made, involving different sample sizes and
procedures of selection.
A roughly optimum allocation of the total
number of items to the respective EC's was the
first step. Factors considered were: (a) the
relative importance of the EC's, as determined
from the 1959 Cincinnati pilot survey, and (b)
a rough measure of variability of price move­
ment, determined on a judgment basis but uti­
lizing experimental price data previously re­
ferred to.
In the final sample, every item in the follow­
ing 11 EC's were considered as certainty selec­
tions. Many of the items would, of course, have
been selected with certainty through the me­
chanics of the selection procedure. Others were
items for which price data are secured from
public records, publications, or manuals so that
collection costs are low.
15.
16
17
18
21
36
37
40
41
42
44

Expenditure class
_,

Name
_. Food away from home.
Rent.
Home purchase and finance.
Taxes and insurance.
Fuels and utilities.
Auto purchase.
Gasoline and motor oil.
Other automobile expenses.
Public transportation.
Drugs and prescriptions.
Hospital services and health insurance.

These 11 EC's contain 37 items. An additional
four items in other EC's were arbitrarily desig­
nated as certainty because of their individual
importance, the desire to maintain continuity of
pricing, or because of the availability of price
data. These are butter, margarine, college tui­
tion and fees,69 and postal charges.
For the final selection, relative importances
(in the family expenditure pattern) for the con­
densed sampling frame (52 EC's and 812
items) were obtained from CES data for nine
cities surveyed for 1960, since data for all 66
cities were not available in time for use. Ex­
penditure data for these nine places were ap69
College tuition would have been certain to fall in one of the two
subsamples. Since data are obtained from reports to the Office of
Education, there was no point in restricting it to one sample.




propriately weighted together to give prelimi­
nary estimates of U.S. average expenditures.
Relative importances for each item in the
sampling frame to the grand total of all items
were cumulated within each EC. A sample selec­
tion interval was computed by dividing the total
relative importance of the EC by the number of
sample items which had been allocated to the
EC. Any item whose relative importance ex­
ceeded this selection interval was certain to be
chosen by the sampling mechanisms. These
items were removed from the array, the rela­
tive importances of the remaining items were
cumulated and divided by the number of sample
items remaining, a new selection interval com­
puted, and additional certainty items selected.
This was repeated until no items exceeded the
selection interval. The probability items were
then determined by choosing a random start and
selecting items at regular intervals along the
array of remaining items (excluding certainty
items).
Since two replicated item samples were to be
chosen, the probability selection procedure was
repeated for the second sample, but one random
start determined both samples. The selection
points of the second sample were "half way be­
tween" those of the first, that is, the successive
selections were made at "half intervals" along
the array of items, with selections alternating
between sample " 1 " and sample "2." Any item
which had a relative importance equal to or
greater than the "half interval" was certain to
fall in one sample or the other and could fall in
both. These were designated as "half certain­
ty." Since each sample was deliberately made
as "different" as possible from the other by the
choice of random start, this may overestimate
the error by the replication approach.
The final operation produced two samples of
206 items each. These included 82 certainty
items and 11 probability items which are com­
mon to both samples. Hence, there were 319
different items in the combined samples. Dur­
ing the year or more following the sample selec­
tion, while pricing was being initiated, a num­
ber of adjustments were necessary. Pricing of
several certainty selections was discontinued, at
least temporarily, because of difficulty in finding
effective pricing methods. These included rent
of room, board, settlement charges on home pur­
chase, auto financing charges, rent of car, and
other financing charges (other than mortgage
47

interest and auto financing). In addition, pric­
ing of cocktails away from home, originally
priced as a second specification besides beer
away from home, was dropped. The weights of
the dropped items are imputed to other items in
the same expenditure class except "other financ­
ing charges" which is imputed to the all items
CPI.
Pricing of four probability items, two in each
sample, also was dropped because of methodo­
logical problems; the items are the addition of
a new room, porch, etc.; completing unfinished
room; uniforms and special work clothing; and
hats and caps. In these cases the weights car­
ried by the probability items in the affected
EC's are redistributed over the remaining prob­
ability items.
It was also necessary to make a few substi­
tutions of items. For example, greeting cards
was initially drawn in one sample. No way of
setting up a specification which would hold
"quality" constant over time was found and
stationery was selected instead. Usually the
item which immediately preceded or followed
the dropped item on the original array was sub­
stituted. Some additional adjustments may be
necessary in the future, but every effort will be
made to hold such substitutions to a minimum.
The final samples which are being priced con­
tain 309 different items represented by 396 spec­
ifications, as listed in appendix table VI. Each
replicated sample contains 198 items of which
87 are common to both samples (76 certainty
items and 11 which were drawn in both). This
count of "items" is in terms of the 812 items
defined for the sampling frame.
Selection of Specifications
The final step in selecting the item sample
was the choice of one or more specified-in-detail
items to represent the items chosen from the
sampling frame. More than one specification
was allowed for important items made up of a
variety of qualities presumed to have different
price movements. The list of items for which
more than one specification is priced is as fol­
lows:
EC 3
EC 5
EC 6
EC 13

Steak.
Pish fillets and
Fresh milk.
Coffee.

48



EC 15
EC 18
EC 21
EC 23
EC 24
EC 29
EC 31

EC 33
EC 36
EC 37
EC 40
EC 42
EC 43
EC 44
EC 47

EC 48
EC 49
EC 50
EC 51

Restaurant meals.
Between meal snacks.
Property insurance.
Electricity.
Gas.
Living room suite.
Soft surface rug.
Men's suits, year-round weight.
Men's trousers.
Men's sport shirts.
Women's winter coats.
Women's street dresses.
Women's stockings.
Women's skirts.
Women's slacks.
Men's street shoes.
Women's street shoes.
New passenger cars.
Used passenger cars.
Gasoline.
Auto insurance.
Parking; garage rent; parking meters.
Over-the-counter items and medical appli­
ances and supplies.
Prescriptions.
Family doctors' fees.
Dentists' fees.
Hospital services.
TV sets and TV radio-phonograph combina­
tions.
Radio.
Nondurable toys.
Pet foods.
Sports equipment.
Indoor movies.
Newspapers.
Cigarettes.
Beer and ale.
Whiskies and other alcoholic beverages.

In addition to these items, there are a few for
which more than one price is obtained in each
outlet for the same specification. Some of these,
e.g., appliances, are priced directly in retail
stores; others come from secondary sources and
utilize internal quality cells or classes in proc­
essing.
In many instances, the sampling frame "item"
consists of a fairly well-defined single commod­
ity or service which was used in setting up de­
tailed specifications for pricing. In other cases,
particularly where the selected item category
contains a miscellany of related but distinct
subitems, it was possible to make a second-stage
probability selection. The procedure was to
make a list of more important subitems, not nec­
essarily exhaustive, and where possible obtain
data on their relative importance.

Examples of second-stage probability selec­
tions are: psychiatrist selected from the cate­
gory "other medical specialists;" water pump
replacement, from "miscellaneous minor auto re­
pairs and services"; nails, from "miscellaneous
hardware and supplies"; and electric hand drill,
from "power tools, except lawn mower." Having
made this second-stage selection wherever it
appeared feasible, no attempt was made to carry
probability sampling any further.
Setting up detailed specifications for pricing
involves research, consultation, experimental
testing and informed judgment. The choice of
the exact specifications takes into account the
importance and representativeness of particular
qualities and the feasibility of describing a
selected item clearly enough to permit repetitive
price collection. A complete listing of the sam­
ple of priced specifications is available on re­
quest.
While an attempt is made to maximize the use
of national specifications, this is not always pos­
sible. Climatic conditions or regional prefer­
ences necessitate city deviations or, in a few
cases, different articles. For example, heavier
weight clothing is not worn in Honolulu, and
lighter weight specifications are required; also,
due to the infrequency of central heating in
Honolulu, furnace parts and repairs can not be
priced and a water heater replacement is sub­
stituted. In Anchorage, on the other hand,
heavier clothing is necessary. City deviations
which are priced in only one, or a very few,
localities are not counted as separate specifica­
tions in the total count given above.
Weighting Structure
Within the expenditure classes (EC's), the
probability-proportional-to-size method of se­
lecting the item sample necessitated a major dif­
ference in the design of index weights as com­
pared with previous procedures. There is no
explicit allocation of weights of unpriced to
priced items. Each certainty item as defined in
the sampling frame carries its own expenditure
weight70 and no portion of the expenditures for
unpriced items of the sampling frame. Prob­
ability items within each EC carry equal parts
of the remaining (probability) weight (total
EC less certainty items). Because of the way
w
Including that of the miscellany of related items as combined in
the sampling: frame.




in which the sample was selected, probability
items normally carry less weight than certainty
items in the same EC. In cities where both
item samples are priced, the expenditure total
for each EC is divided equally between the two
samples as of December 1963. When more than
one specification is priced for a single sampling
frame item, the probability weight for the item
is divided among the individual specifications
—not necessarily equally. The final index rela­
tive importances represent relative expenditures
of the expenditure classes as of the survey date,
but the relative weights of the individual prob­
ability items do not relate to actual expendi­
tures.
The starting point for weight derivation was
the tabulation by city of annual average expend­
iture data from the CES for consumer units
which meet the definition of the index popula­
tion. (See footnote 36 chapter III.) For pur­
poses of the CPI, expenditures include (1) all
items classified as current consumption expendi­
tures in the CES, excluding money lost or stolen
and food and rent received as pay; (2) expend­
itures for purchase and improvement of the
family home and other real estate for family
use (which were converted to CPI weights by
special procedures described later); and (3)
expenditures for gifts of goods and services to
persons outside the family. For items bought
for cash, the actual cash outlay is the basis for
the index weight. For items bought on the
installment plan, the weight is based on the total
price, not merely the portion paid during the
survey year. Index weights reflect average ex­
penditures of all consumer units, obtained by
dividing aggregate expenditures by the total
number of units, including those who did not
make purchases of a particular item, as well
as those who did.
The complete tabulation includes average ex­
penditures for the 1,800 individual items on the
CES schedule. These are consolidated into the
812 items finally established as the CPI sam­
pling frame. Exhibit B shows the CES schedule
line number as well as the sampling frame code
for each item.
The allocation of expenditure data to the sam­
pling frame and summation to EC totals were
carried out separately for families of two or
more persons and for single persons who meet
the definition of an index consumer unit. It was
done for each of the 66 SMSA's and cities and
49

for combinations of cities and regions. Com­
bined expenditures for 1960-61 for large cities
surveyed in both years were derived as a simple
average of average expenditures for the sepa­
rate 2 years.
Seasonal Adjustment for Food
For most items, annual expenditures were ob­
tained from consumers in the CES. For foods
and other items which are purchased frequently,
usually in grocery stores, item detail was ob­
tained only for the 7-day period just prior to the
date of the interview. From the weekly data it
was necessary to estimate annual expenditures
for individual items and adjust the total of these
separate estimates to the annual totals.
To determine seasonal variation, consumer
units surveyed in the spring were subsequently
asked to provide information on weekly food
purchases during different seasons. Mail ques­
tionnaires were sent each month to 12 inde­
pendent samples of consumer units in 16 cities
throughout the United States and similar followup questionnaires were sent in the same
cities to a subsample of families who had been
interviewed in the CES. All data obtained for
a city for a month were pooled and the aver­
age weekly expenditure per family in each
month was computed for each food item pur­
chased in each of the 16 cities. From these data,
monthly indexes of seasonal expenditures were
computed in relation to the yearly average, for
expenditure class totals, and for certainty items,
for each of the 16 cities.
Because of the small size of the samples in the
various cities, as well as lack of comparability
of families reporting, the consumption patterns
shown by these indexes are erratic. In an effort
to smooth out some of the fluctuations, the sea­
sonal consumption indexes for individual cities
were combined into four regional indexes and
the four regional indexes combined to a U.S.
index. Many of these were adjusted by com­
modity analysts on the basis of seasonal produc­
tion or marketing data from other sources—
U.S. Department of Agriculture figures on
weekly production of meats, shipments of fruits
and vegetables, etc. Monthly indexes of the
value of sales by food group and by item in six
large cities, based on confidential 1960 sales of
a large food chain, were also used to judge the
reasonableness of the adjusted seasonal indexes.

60



For items which have marked seasonal dif­
ferences in purchases, the weekly expenditures
obtained for the CES sample were blown up to
annual totals by applying seasonal factors ap­
propriate for the actual period of the original
7-day CES survey in the particular city. The
regional seasonal index was used for all cities
in a region if it seemed valid; otherwise, the
U.S. index was used. For a few items for which
no expenditures were reported, because they
were out of season during the survey week, an­
nual estimates were made from secondary
sources, based on relationships to reported
items. For items which showed only minor
seasonal changes in consumption, expenditures
for the reported week were multiplied by 52 to
obtain the annual estimates. The sum of the an­
nual figures derived in this manner were then
adjusted to the total average annual expendi­
tures for food as a whole reported in the CES.
Derivation of Item Weights
CES expenditures by city or for combinations
of cities were used directly for index weights,
except for certain categories for which special
estimates were derived. These were home pur­
chase and financing, automobile purchase and
financing, and alcoholic beverages. Special
weight derivation procedures for these items are
discussed in a later section.
The general plan of the weight derivation
process, for the majority of items, was to use
unadjusted city expenditure data for broad
groups of items and for subgroups and indi­
vidual items for which data appeared reason­
able. At successively lower levels of aggrega­
tion, where the data appeared erratic as the
result of small samples, it was necessary to use
an average of several cities as a means of dis­
tributing group or category totals to compo­
nents. This involved an element of subjective
decision but at a fairly low level in the index
classification scheme, since totals for major cate­
gories usually could be used without adjustment.
Because of the very small samples of "index"
single persons71 within a given city, expendi­
ture patterns derived directly from the survey
results are erratic and are, therefore, not suit­
able for index weights by city. The data were
n
To meet a commitment for a separate index for families of 2 or
more, it was necessary to derive weights separately for families and
singles. The separate index for families previously described was dis­
continued after November 1964.

combined to regional and national totals and
compared later with like combinations for fami­
lies of two or more. On the basis of these com­
parisons, it was decided to derive city weights
for single workers by applying to the final index
weights for families the national ratios of ex­
penditures of singles to those of families.
For convenience and ease of comparison
among cities in making decisions as to whether
or not to use unadjusted city data the observed
expenditure data for the sampling frame for
each city were converted to relative importances
at successively lower levels of aggregation,
major categories to all items, major groups
to categories, subgroups and/or expenditure
classes to major groups, and individual items
to expenditure classes, for wage-earner clericalworker families of two or more persons.
The major categories used for this purpose
are: food at home, food away from home,
shelter (less home purchase and financing), fuel
and utilities, household furnishings and opera­
tion, apparel and upkeep, transportation, and
health and recreation (less alcoholic beverages).
The items excluded are those for which expendi­
ture weights were derived independently. Al­
though CES data for total transportation were
included in the calculation of relative impor­
tances, the CES data for automobiles and auto
financing charges were subsequently replaced
by specially edited data as explained later. The
adjustments were not large enough to affect the
relative importances for the other categories sig­
nificantly and had no effect on the dollar expend­
itures accepted for the weights.
At the highest level of aggregation, major
category, the relative importances to all items
show remarkable consistency among cities for
families of two or more, especially for the large
cities. In the largest cities where surveys were
conducted in 2 years—1960 and 1961—the data
for the separate years are also generally con­
sistent. As a result, category expenditures were
accepted without adjustment for each of the
larger cities (A, B, and C strata). For the small
cities, average expenditures for all D size cities
within a region (including the 16 additional
nonindex cities surveyed), were adopted for
each D city within a region. Regional averages
for city-size strata and for all size cities com­
bined were computed by dividing aggregate ex­
penditures by the total number of index fami­
lies. Therefore, the number of usable schedules




in each city became implicit weights. In addi­
tion to the pooled average, D stratum regional
averages were computed as simple averages of
city averages. These were used in preference to
the pooled averages in the weight derivation for
D cities in order to give the cities equal weight.
For example, each of the 6 southern D size cities
in the CPI have identical expenditure weights
for food at home, namely, the average of expend­
itures for families in the 12 southern small
cities for which CES data were available. This
procedure made maximum use of the data avail­
able and tends to minimize the sampling error
in this city-size class.
The relative importances for each city were
reviewed by region at successively lower levels
of aggregation. At each stage of the review, it
was decided either:
1. To use the city expenditure data as index
weights for families of two or more without ad­
justment; or
2. To make adjustments for obvious aberra­
tions resulting from the sampling process, usu­
ally by applying an average of internal relative
importances for several cities to the expendi­
tures already determined for the city at the next
highest level of aggregation.
As expected, the data show much more vari­
ability at the lower levels. For the large (A and
B strata) cities, review of the data indicated
that the figures could be accepted as reported
for most of the expenditure class (EC) totals.
The major exceptions were durable housefurnishings and miscellaneous personal and finan­
cial expenses where the relative infrequency
of purchase results in erratic results by city
even at the expenditure class level. Total expend­
itures for durable housefurnishings by city
were therefore distributed among the four ex­
penditure classes (EC's 23-26) on the basis of
the average percentage distribution of expend­
itures in cities of all sizes within each region.
For EC-53, miscellaneous personal and financial
expenses, regional relative importances of EC
52 to the total of all other EC's were applied
to the city totals for EC 1-51, as derived in the
derivation process. In some cases, regional aver­
ages of all size cities also weise used to distribute
the transportation category expenditures among
the component expenditure classes. For some
EC's, data were accepted without adjustment
for C cities also, but in a number of cases the
major group totals were distributed to EC's on
51

the basis of regional relative importances. Gen­
erally for the D cities, the average expenditures
for all D cities in the region were used at the
EC level.
Below the EC level, i.e., the distribution to
sub-EC's and to individual certainty items and
the probability items combined, the basis for
index weights for families of two or more
varied. City data were used without adjustment
in some of the largest cities, but more generally
regional or city-size average relative import­
ances were used to break down EC totals pre­
viously determined. Appendix table VII sum­
marizes the weight derivation process for fami­
lies. Weights for probability items within an
EC are divided equally. In those cases where
more than one specification has been selected to
represent a single sample item, the distribution
of the final weight of the item to the priced spe­
cifications was made by the commodity special­
ists according to their knowledge of jiie indus­
try. In many cases, the priced specifications are
given equal weights; in others, weights are
assigned according to their estimated relative
importances in the market.
Home Purchase

Some families in the survey bought homes for
the first time during the survey year; others
sold one house and bought another, usually
larger and more expensive. The total index
weight was defined as gross expenditures for
houses bought by families buying for the first
time plus net expenditures by families for re­
placement homes (gross price of the new house
less the price received by the family for the sale
of the house which was replaced).
Estimates of purchase price for initial acqui­
sition, and for replacement, were derived
arithmetically for each city from an average
purchase price for the two types of purchasers
combined, derived in turn from the entire Com­
prehensive Housing Unit Suryey sample. Houses
purchased during the survey year and the 3jgreceding years were included in the ayerage^ Jo
improve the sample estimate. As changes in
prices of houses during this period were rela­
tively small, the reported prices were used with­
out adjustment for price change to the link
month. To take account of differences in home
prices between index and nonindex-type con­
sumer units, this average was adjusted by the
national ratio of average current market value
of homes owned by index-type homeowners to
the value of homes owned by all homeowners, as
reported, in the GES.
Using CES data for index consumer units in
all cities, a national ratio of prices paid for re­
placement houses to prices received for houses
sold by the same families (initial purchase)
was calculated. This ratio was used to derive
average prices in each city for the two types of
purchase, according to the following equation:

Because of the relative infrequency of home
purchase, the small samples in the expenditure
survey do not yield reliable averages. It was
necessary, therefore, to use a special estimating
procedure utilizing data from sources outside
the CES. As for all other items in the index, the
weight for home purchase represents, in princi­
ple, the average obligation incurred by the index
population in the survey year. The average is,
P=PJ+PrR
of course, affected by consumers who rented
homes or did not buy homes in the survey year.
x r = K J Pi
The average expenditure for families of two .
P=PiI+KPiR
or more was derived as the product of three
Pi(I+KR)-P
separate factors:
1. Estimated average price paid (or con­
P 1. . - P ,
I+KR
tracted for) by index consumer units who
bought homes during the survey year;
where P = the estimated average price paid for
2. Estimated average rate of purchase, i.e.,
all houses bought by index consumer
the average percent of homeowners who pur­
units in a 4-year period, derived
chased homes during the survey year; and
from CHUS data for a given city,
adjusted to index family basis
3. The level of homeownership among index
Pi
=
the estimated average price paid for
consumer units, i.e., the percent of consumer
initial purchases of houses by index
units who were homeowners during the survey
consumer
units in a given city
year.
52



P r = the estimated average price paid for
replacement purchases of houses by
index consumer units in a given city
K = the average ratio of replacement
purchase prices to sale (initial pur­
chase) prices, as reported by index
consumer units in the CES, for all
cities combined
I = the proportion of homeowners among
index consumer units who became
homeowners during the survey year,
derived from CES data for all
cities combined
R = the proportion of homeowners
among index consumer units who
bought homes for replacement, de­
rived from CES data for all cities
combined.
The value of K, the ratio of purchase price to
sale price, or replacement to initial purchase
price, varies with the level of sale prices. Con­
sequently, it was computed separately for four
groups of cities, i.e., those with average sale
prices (1) below $6,000, (2) $6,000-$13,999,
(3) $14,000-417,999, and (4) $18,000 and over.
The rate at which families purchased houses
during the survey year was derived in two ways
—from Census data showing percent of home­
owners on April 1, 1960, who moved into their
dwelling units during the previous year and
from CHUS. Because purchase rates derived in
this way from Census data weite so inexact and
much higher than from other sources, those de­
rived from CHUS data were used in final deri­
vation. CHUS consumer units could not be
identified as to their eligibility for the CPI.
Therefore, ratios of annual rates of purchase by
index consumer units and all consumer units for
the most recent 4 years from the CES, calcu­
lated on a regional basis, were applied to the
city CHUS purchase rates to adjust to purchase
rates of index consumer units. These composite
purchase rates were separated into rates for ini­
tial and replacement purchase on the basis of
the national proportions of homeowners report­
ing each type of purchase in the CES.
The remaining component, level of ownership,
needed for deriving weights was obtained direct­
ly from the CES, using as a basis the number of
months of homeownership reported by index
consumer units. This method takes full account
of consumer units who changed from renters to
owners (or vice versa) during the year.




The weight derivation for home purchase by
index families of two or more persons is illus­
trated in table 4, using hypothetical data from
city A in region 1.
Because of the very small sample of index
single consumer units in each city, it was not
possible to follow similar procedures to derive
weights for single workers. The percent of sin­
gles making replacement purchases, undoubted­
ly very small, was set at zero, because in the
entire sample of cities not a single case was re­
ported for the survey year. In general, the fac­
tors in the general formula for weight deriva­
tion for initial purchase were computed by
applying the ratio of the average value for
singles to the corresponding average for con­
sumer units of two or more, developed for a
group of cities, to the city values for families
of two or more. Final expenditure weights for
single consumer units were derived from the
weight derivation formula and combined with
those for families, using the relative population
weights described in a later section.
Mortgage Interest
The index expenditure weight for mortgage
interest is defined as the average amount of in­
terest incurred for the purchase of homes by the
index consumer units during the survey year
and subsequently paid. From secondary sources,
it appears that mortgages are either paid off or
refinanced at about half term. Consequently,
interest contracted for was computed for half
the average term. It was derived from the aver­
age amount and term of mortgages, the average
interest rate, the average rate of mortgaging
among homeowners, and the level of ownership,
pooling city data where necessary. As these
factors differ for different types of situations,
separate calculations were made for families of
two or more and for single workers; for new
and assumed first and second mortgages; for
new homeowners and for those buying replace­
ment houses; for conventional mortgages issued
by banks and private individuals, and for mort­
gages insured by government agencies.
Because the survey showed only the amount
of interest actually paid during the survey year,
it was necessary to compute total amount of
mortgage interest contracted for, using data
from the survey supplemented by other sources,
including those used for the home purchase
53

TABLE 4.

ILLUSTRATION OF WEIGHT DERIVATION PROCEDURE FOR HOME PURCHASE, INDEX CONSUMER UNITS OF 2 OR MORE
PERSONS, CITY A, REGION 1 (HYPOTHETICAL)

Line
No.
1
2
3
4
5
6
7
8

F9

10
11
12
13
14
15
16
17
18

19
20
21
22
23
24
25
26
27

Item
Average price of houses bought in most recent 4 years
Average market value of owned homes
Average market value of owned homes
Ratio: index to all consumer units
Estimated average purchase price of houses (F)
Proportion of home purchasers buying houses initially (I)
Proportion of home purchasers buying replacement houses (R)
Average price of replacement homes bought
Average price of homes sold by replacement
buyers
Ratio: purchase price to sale price ( K ) l
I + K R (See equation)
Average price paid for initial purchase of houses (Pi) =»P-f-(I+KR)..
Average gross price paid for replacement purchase of houses (Pr)
Average net price paid for replacement purchase of houses (Pr— Pi)Average annual rate of purchase in 4 most recent years
Average annual rate of purchase in 4 most recent years
Average annual rate of purchase in 4 most recent years
Ratio: index units to all units
Estimated annual rate of purchase:
All purchases
Initial
--Replacement
___
Percent of consumer units owning homes
Estimated annual rate of purchase:
Initial
Replacement
Estimated annual average expenditure:
Initial purchase
Replacement purchase (net)
Total
_
Average expenditure for vacation homes
Total CPI expenditure weight for home purchase in survey year

Type of consumer unit

Area

All home buyers.,-.
Index homeowners..
All homeowners

City A
City A
City A

Index home buyers
Index home buyers
Index home buyers
Index replacement home buyers.
Index replacement home buyers.

City A
U.S.
U.S.
U.S.
U.S.

All home buyers
All home buyers
Index home buyers.

City A
Region I
Region 1

Index home owners-

City A

All index.
All index.

City A
City A

All index.

City A

All indexAll index.

City A
City A

Source
of data

Amount

CHUS
CES
CES
2-5-3
1X4
CES
CES
CES
CES
8*9
6-K7X10)
5-M1
12X1.321
13-12
CHUS
CES
CES
17-M6

$16,300
$13,300
$16,000
0.831
$13,545
.683
.317
$9,874
$8,053
1.226
1.072
$12,635
$16,691
$4,056
7.00
7.60
7.90
1.039

15X18
19X6
19X7
CES

7.27
4.97
2,30
62.0

20X22
21X22

3.08
1.43

12X23
14X24
25+26
CES
27+28

$58
$447
$53
$500

1
The ratio of purchase price to sale price (replacement to initial purchase)
was found to vary depending upon the sale price level. For sales from

$6,000 to $13,999, the ratio was 1.321; for $14,000 to $17,999, it was 1.144.

weight derivation. Wherever the survey in a
given city did not provide a large enough sample
to produce reliable averages, the data were com­
bined with those for other cities to provide
regional or, in some cases, national averages.

to repay the loan if paid off after having run
only one-half its stated term, assuming no pen­
alty for early payment. Standard amortization
tables were used in this computation.
Table 5 illustrates the weight derivation pro­
cedure for a hypothetical city A in region 1, for
families of two or more persons who obtained
new first mortgages. Similar calculations were
made for other types of mortgages and for sin­
gle consumer units and the resulting amounts
were combined into an overall average, using
appropriate weights.

Operationally, the weight was calculated as
the product of two factors: (1) the amount of
interest contracted per mortgage for half the
term, and (2) the percent of consumer units
who obtained mortgages. Interest contracted
was the product of three factors: (1) average
amount of mortgage, (2) interest rate, and (3)
term of mortgage. The amount of interest paid
over the entire term of a mortgage is deter­
mined by the principal, the rate of interest, and
the term. The principal amount was derived as
a proportion of the purchase price (as computed
for the home purchase weight). The ratio of
mortgage principal to purchase price was com­
puted by region for initial and replacement pur­
chases from the CES and used for every city in
the region. Average interest rates and terms
also were available from the CES. Regional
averages were used for rates and national data
for the term. Standard amortization tables were
used to derive monthly payments. An estimate
of the amount of interest to be paid for the first
half term was derived by subtracting the mort­
gage principal from the total amount required
54



Automobile Purchase and Financing Charges
In general, the weights used in the index for
both new and used cars represent average
net family expenditures for cars purchased
(whether or not paid for) during the survey
year, based on specially edited CES data. Data
were derived separately for new and used cars
and for financing charges, for index families of
two or more and for single consumer units.
Values represent net purchases for consumer
units who purchased and sold cars during the
survey year. Stated briefly, the values of cars
traded in, sold, or repossessed were deducted
from purchase prices, to arrive at the net value.
The weights included all transactions during

TABLE 5.

ILLUSTBATION OF WEIGHT DEEIVATION PBOCEDURE FOR MORTGAGE INTEREST, INDEX CONSUMER UNITS OF 2 OR
MORE PERSONS, CITY A, REGION 1 (HYPOTHETICAL) NEW FIRST MORTGAGES

Line
No.

Item
Percentage of home buyers obtaining mortgages
Estimated annual rate of purchase
Estimated annual rate of mortgaging
Estimated average purchase price of houses
Average ratio of mortgage to purchase price
Estimated average principal amount of mortgage.
Average rate of interest on mortgages—
Average term of mortgage in years
Average monthly payments
Total payments per year
Total payments during half term of mortgage
Balance due on principal at end of half term
Total payments required to discharge loan
Total amount of interest paid in half term
Percent of consumer units owning homes
Percent of consumer units obtaining mortgages
Estimated average amount of mortgage interest...

the survey year for consumer units who pur­
chased cars; however, sales of cars by consumer
units who did not purchase cars during the sur­
vey year were not treated as a deduction for the
weights.
Where they occurred, other deductions from
purchase price such as the value of a wrecked
car also entered into the netting process. Cars
used entirely for business were excluded. Values
for cars used only partially for business pur­
poses were adjusted on a proportional basis, so
that only that portion of expenditures or sales
for family or "pleasure" use entered into the
weights.
The total CES weights for new and used cars
were then distributed to the following priced
specifications:
New c a r s . . .

„.,

Used cars
(Ford and Chevrolet)

Sample 1
Sample 2
Chevrolet Impala
Chevrolet Impala
Ford Galaxie 500
Ford Galaxie 500
Chevelle
Ford Falcon
Plymouth or Dodge
Pontiac or Buick
Rambler or Studebaker
Volkswagen or Fiat
Both samples
2 yearold
3 yearold
4 yearold
5 yearold

The breakdown of weights was made on the
basis of industry data on production or sales,
maintaining for new cars appropriate weights
of standard size versus compact and foreign
cars combined in each sample. CES data for
auto financing charges were tabulated and
edited for consistency with car purchase
weights.
Alcoholic Beverages
Previous consumer expenditure surveys have
consistently indicated a tendency on the part of




Type of
consumer unit

Area

Source of data

Replace­
Initial
ment
purchase
purchase

All index units
Index homeowners
Index homeowners
Index home buyers..
Index mortgagors
Index mortgagors
Index mortgagors
Index mortgagors
Index mortgagors...

U.S.
City A
City A
City A
Region 1
City A
Region 1
TJ.S
City A

All indexAll indexAll index.

City A
City A
City A

CES
A-20&21.
1X2
A-12&13
CES
4X5___
CES
CES
amortization table. _.
12 X mo. payments.
y* (8X10)
amortization tables..
11+12
13-5
A-22
15X3
16X14

94.8
94.8
4.97
2.30
4.71
2.18
$12,635
$16,691
.85
.75
$12,518
$10,740
5.75
6.00
20.25
16.75
$74.91
$98.87
$898.92 SI,186.44
$9,102
$9,936
$6,910
$7,766
$16,012
$17,702
$5,272
$5,184
62.0
62.0
2.92
1.35
$154
$70

consumers to understate their expenditures for
alcoholic beverages. In order to evaluate the de­
gree of underreporting of expenditures reported
in 1960 and 1961, the CES data were compared
with figures from other sources.
From the national income accounts, estimates
of aggregate personal consumption expenditures
were available for the three major categories of
alcoholic beverages—beer, distilled spirits, and
wine. National accounts estimates of average
expenditure per household for 1960 and 1961,
separately, ranged from 1% to 2 ^ times the
CES average per consumer unit for the three
items. Estimated quarterly consumption data
by States from industry sources also indicated
a general underreporting in the CES.
Because of differences in the definition of
household for consumption expenditures in the
national accounts and of consumer units for the
CES, the average expenditures per unit were
not comparable. Therefore, the relative impor­
tance of alcoholic beverages to total food expend­
itures in the two sets of data was chosen as
the basis for adjustment. Food was used in
preference to total expenditures because con­
ceptual differences between the national ac­
counts and the CES make total expenditures less
comparable than food expenditures. In 1960,
expenditures for alcoholic beverages were 14.6
percent of food in the national accounts, twice
as high as in the CES all-city average; and in
1961 they were 14.9 percent—almost 2% times
the CES.
Since it could not be assumed that the national
accounts provided the best estimate of expendi­
tures for the population represented by the CPI,
it was decided to make only a partial adjust55

ment. Some of the factors considered were as
follows:
1. The national accounts data represent the
entire U.S. population. Since they are derived
from reports of tax revenues, they may not
necessarily correspond with actual consumption
data, because they are affected by changes in
inventories of liquor dealers.
2. The CES expenditures to be used as index
weights represent urban wage earners and cleri­
cal workers only. They were derived from sam­
ple surveys of consumer units in cities of all
sizes. Expenditures reported by individual con­
sumers may reflect actual consumption more
closely than data on tax revenues.
3. The distribution of total personal con­
sumption expenditures (from the national ac­
counts) for alcoholic beverages among the three
major categories—beer, distilled spirits, and
wine was approximate. It was derived by de­
ducting rough estimates of expenditures for
business use from total expenditures.
4. The "household" used to derive average
personal consumption expenditures is not the
same as the "consumer unit" used in the CES.
However, much of the effect of these differences
has been eliminated by the use of the rela­
tionship of expenditures for alcoholic beverages
to expenditures for food rather than actual
expenditures.
5. It was not possible to make separate deter­
minations for alcoholic beverages at home and
away from home.
Although there is no exact measure of the
effect of these differences between the two sets
of data, expenditures for alcoholic beverages
were assumed to be about half way between the
expenditures reported in the CES and those de­
rived from the national accounts data. Accord­
ingly, after distributing CES expenditures
away from home proportionately to beer, dis­
tilled spirits, and wine, the dollar expenditures
in each city as reported in the CES were ad­
justed upward by the following overall factors:
Beer
Distilled spirits
Wine

I960
1.48
1.57
1.25

1961
1.72
1.53
1.46

These factors represent for each category the
ratio of the average of the relative importances
to total food expenditures in the national ac­
counts and in the CES for all cities combined
to the latter.
56



The adjusted city data were then subjected to
the same type of review as were other items.
With few exceptions, adjusted data for the indi­
vidual items of alcoholic beverages in the larger
cities were accepted as index weights. Regional
averages were used both at the EC and item
levels in the smallest cities. An additional factor
in the establishment of weights was the legal
status of alcoholic beverages in the various lo­
calities. In cities where sales of distilled spirits
by the drink are prohibited by law, no weight
was established for the index and any expendi­
ture reported for the item was allocated to beer
and/or wine in the same city or to other D cities
where drinks are sold.
Population Weights
In combining city data to U.S. totals for the
index, each of the 50 index cities carries a popu­
lation weight, derived from Census 1960 urban
population for the region-city size strata used
in the sample selection described in chapter V.
These are expressed as relative population
weights adding to 1.000; they are built into the
city item expenditure weights, so that weighting
to U.S. totals is a simple matter of summation.
The product of the expenditure weight and the
relative population weight for each item is
called a cost-population weight.
The Census data were adjusted to obtain esti­
mated wage-earner clerical-worker population.
(See appendix table VIII-A.) Estimates of the
number of equivalent persons living in con­
sumer units all year (column 1 of appendix
table VIII-A) were derived by regional citysize strata by subtracting institutional popula­
tion and armed services personnel living on
post, and making allowances for births, deaths,
migrations, etc., since 1960. The number of
consumer units (column 2) was obtained by
dividing these estimates by the average size of
consumer units by city as shown in the CES.
It was then necessary to estimate the number
of "index" (wage-earner clerical-worker) con­
sumer units by applying CES proportions of
index to total consumer units by regional citysize strata. The percentage distribution of these
estimates shown in column 3 is the basis for the
city population weights. Only the 12 certainty
A cities carry their own weight. For all others,
the region-city size stratum weight is divided

equally among the sample cities in the stratum.
Because of this design, smaller cities may carry
as much or even more weight than some of the
large certainty cities.
Because of the original plan to calculate sepa­
rate indexes for families and singles combined
and for families only, and the way in which the
expenditure weights were constructed, it was
necessary to partition the city weights between
families and singles. This was done using the
CES ratios of families and singles in A cities,
and region-city size average ratios for other
cities. Final weights for the 50 cities, additive
to the city totals, are shown in appendix table
VIIL
When the six additional B size cities were
added to the index in January 1966, population
weights were revised and revised cost-popula­
tion weights were introduced by linking in De­
cember 1965, i.e., for the six cities and the
other cities affected. Since the six additional
cities were selected purposively, they carry only
their own population weights. Therefore, the
weight of the additional cities within a region
was subtracted from the region-stratum weight
and the remaining weight divided equally
among the probability B cities. Appendix table
VIII-B shows the population weights for B




cities before and after the addition of the six
cities to the CPI sample.
Price Adjustment
The item weights as derived from the CES
relate to the average of the years 1960-61. Tech­
nically, it was necessary to adjust them for rela­
tive differences in price movement from 196061 to December 1963, chosen as the link month
for the new index. This could be done only in
a very approximate way simply because of lack
of prices for the new cities and the new items.
For the 14 overlap cities, percent changes from
the old series for appropriate commodity groups
were used to make the necessary price adjust­
ment in each city. For the new cities, national
indexes were used.
Final Relative Importances
The final U.S. relative importances as of the
December 1963 link month in comparison with
old series weights are given in appendix table
IX. In making this table, old series data were
regrouped according to the classification of
items for the new series. Data are shown sepa­
rately for the certainty items in the revised
index, but not for probability items, since they
carry equal weights rather than weights repre­
senting the importance of the items themselves.

57

Chapter VIII. Outlet Samples, 1964 Index*
The general objective and plan for selection
of reporter samples for the revised Consumer
Price Index were described in chapter IV. This
chapter describes detailed operating steps. Al­
though the outlet samples selected by the pro­
cedures described in the following pages are,
strictly speaking, not probability samples, they
have the primary merits which should come
from probability principles: Objectivity, lack of
bias in the selection procedure, and as adequate
a representation as is possible of different types
of stores and geographic sections of the SMSA's.
Probability sampling techniques are possible to
a much greater extent for those few commodi­
ties or services (such as food and rent) for
which relatively large samples are priced than
for other items, most of which are priced in very
few outlets. Nevertheless, even for the latter,
revised CPI samples were selected within a
probability framework. They are designed to be
representative of all types of stores serving the
index population and can be presumed to be un­
biased samples.
The possibility of using probability sam­
pling techniques is greatly limited due to a
lack of data for developing listings of universes
of stores of different types in which specified
goods and services are sold and which are
patronized by the wage-earner and clericalworker index population. To be more specific,
detailed information is not available for exclud­
ing particular stores which offer goods and
services of a quality that is "too high" or "too
low" to be representative of that purchased by
the index population. Secondly, no data are
available to limit universe listings to outlets
where continuous pricing of comparable items
can be assured.
In addition to the universe data limitations,
there is a problem in developing suitable sam­
pling designs for each item priced because of
lack of data on volume of sales by type of store.
Although each item, theoretically, might have its
* Prepared by Helen M. Miller.

58



own reporter sample, it is not feasible from a
cost standpoint to select each of these samples
independently. Such a procedure might well
spread the pricing over an excessive number of
outlets with very few quotations being obtained
from any single one. Moreover, only for food
are retail sales data generally available by type
of store for individual cities.
Size of Sample
The first step in selection of the reporter sam­
ples is to determine sample sizes for each expend­
iture class (EC) of items in each area, taking
into account the cost of collection and available
information on variances in price movements.
The basic sample size for most nonfoods, for
which price collection is most costly, was set at
four reporters, the same as in the previous in­
dex. This means a total sample size of eight for
certainty items in replicated cities. In three of
the largest SMSA's (Chicago, New York, and
Los Angeles) the sample sizes were increased
to 5 per sample or a maximum of 10 for the 2
replicated samples.
As indicated in appendix table X, the size of
the outlet sample differs considerably from the
basic sample size for nonfoods for a few index
components. For example, sample sizes for some
of the professional medical care services were
established slightly above the basic sample size
used for the revision, but this represents a re­
duction in the expanded sample size initiated
for medical care in 1958. In the old index, the
number of physician reporters (20 large cities)
ranged from 12 to 42 in New York for the office
and house visits. Similarly for dental and other
professional services (obstetrical, surgical, etc.)
reporter samples varied from 6 to 30 reporters.
For these same services in the "C" and "D"
strata cities, the sample consisted of six report­
ers wherever possible. In the revised index the
sample size for medical services ranges from 3
to 24.

In other instances where either the universe
of a specific outlet type is comparatively small
(e.g., utility companies), or where prices are ob­
tained by mail questionnaire (e.g., newspapers)
it is possible to price the universe of reporters
in the city proper and in selected suburban pric­
ing areas.
For new automobiles, the sample of dealers is
designed to represent the various manufactur­
ers. Therefore, the total sample for all makes
combined is larger than the basic sample size
of four, but the sample for individual makes
of automobiles is less than four. In some cases,
sample sizes are not established on a city basis.
A total sample of 200 (or 50 for each of four
regions) is used for hotel and motel accommoda­
tion rates. This total sample is allocated to the
various cities (A, B, and C strata only) based on
relative importance of the hotel receipts in an
SMSA to the total for the region. Thus, the
number priced for this component ranges from
1 in some of the "C" stratum cities to some 35
hotels or motels in New York. ("D" stratum
city hotels are excluded due to their small rela­
tive importance.) An exception also has been
made for college textbooks. In this instance,
sampling is established on a regional basis and
limited to nine universities, since uniform list
prices prevail for this item.
Lastly, for some items no specific sample of
outlets is selected. Instead, for these—home pur­
chase, college tuition, used cars, automobile in­
surance, magazines, etc., data collected by other
government agencies or private organizations
are used.72
Because of the volatility and variability of
prices, sample sizes for food are considerably
larger than for nonfood and they vary for dif­
ferent cities, depending upon number of chain
organizations priced and other factors. Basic
data on retail food store sales were obtained
from special tabulations from the 1958 Census
of Retail Trade. Number of stores and volume
of sales were reported for chains (four or more
stores) and independent stores, by type of store,
by central city and remainder of SMSA, and for
independent stores by seven sales size groups.
Determination of the number of food stores to
be sampled was based on two factors. (1) Since
all chain organizations are to be priced, there is
no explicit restriction on the number of chain
« See chapter X for a description of special procedures used.




outlets. In most CPI cities, one outlet of each
food chain is priced and more than one, if neces­
sary, to represent stores having different pric­
ing policies, such as discount division outlets
and suburban outlets. In a few cities, some
small chains are excluded if the organization
represents less than 1 percent of chain grocery
sales in the SMSA. For New York and Los An­
geles all the largest chains are included, but it
is necessary, because of cost, to sample the
smaller chains (organizations with less than 10
outlets in the area).
(2) The total cost of collecting food prices for
the revised index is planned at about the same
level as in the previous index, for which sample
sizes were large enough to permit publication of
U.S. and city average prices. The maximum
sample size for independent food stores is estab­
lished by the number of outlets required to pro­
vide a predetermined number of quotations for
meats and fresh produce, the prices of which are
usually more volatile than other food items
(sometimes referred to as dry groceries). For
example, in a particular city the nominal sample
size of quotations for meat and produce might
be 30, with 20 meat quotations allocated to regu­
lar food stores and 10 to meat markets, and with
25 produce quotations allocated to regular food
stores and 5 to produce markets. Since meat
and produce markets do not ordinarily yield
prices for other groceries, this means that the
maximum number of quotations for dry gro­
ceries is determined by the 25 regular food
stores for fresh fruits and vegetables. For most
"D" stratum cities, the size of the sample is set
at five independent grocery stores, plus chains.
The variations in a few D cities reflect the lim­
ited number of stores in a city or the importance
of chain stores. Independent food outlet sample
sizes by city are given in appendix table XI.
Allocation by Type of Store
As indicated earlier, it would be desirable to
distribute pricing across the various kinds of
stores roughly in proportion to their importance
in terms of sales of each specific item. Of course,
with a basic sample size of 4 outlets (5 in the
three largest SMSA's) for nonfood items, the
allocation of quotations by type of store is done
only in a very approximate manner, even in the
more fully replicated cities where 2 outlet sam­
ples are priced (8 or 10 outlets in total). For
59

certainty items priced in two replicated sam­
ples, it is considered more desirable to make the
best possible allocation for the combined outlet
sample and then make an ex post facto division
into the two samples. Thus, in some instances,
the replicated samples are not "balanced" ac­
cording to type of store.
For nonfood items, very little factual data are
available on which to base outlet type alloca­
tions, and data which are available have limita­
tions. With few exceptions, distribution patterns
are reported for commodity groupings and not
for specific items. Thus, the relative importance
of the various outlet types for a grouping can
be reflected, but the distribution pattern might
vary considerably for the specific items included
in the group. Also, distribution patterns change
continually and vary from city to city. In many
instances up-to-date data and/or city distribu­
tions are not available.
For some major commodity groups, such as
furniture, household appliances, and apparel,
"merchandise line" statistics are available for A
and B SMSA's and some C stratum cities indi­
vidually from the 1948 Census of Retail Trade.
These data were utilized in determining both
outlet types and the distribution of quotations
among these outlet types. In order to approxi­
mate a more current distribution, the 1948 prod­
uct line information (by area) was updated by
1958 Census sales data by type of store. In other
words, for each SMSA the 1948 distribution of
sales by merchandise line for a given outlet type
was applied to the 1958 total sales of that outlet
type, to obtain updated merchandise line sales.
The 1958 dollar volume sales thus calculated for
the various outlet types provide a distribution
pattern for selected merchandise lines by type
of outlet.
Strict outlet type allocations are impractical
for small samples. For example, using avail­
able source data, a typical distribution pattern
for the major household appliance category
might be as follows:
Type of outlet
Total
.
Household appliance store.
Department store....
Radio store
T
Furniture s t o r e . . . . . . . . . .
.
Variety store
.
T..
Hardware store......
r
Jewelry s t o r e . . , . . * . . . . . . . . . . . .
.

60



Percent
distribution
of sales
100.0
..........
56.6
.
14.3
11.0
—
6.3
4.8
...T.
3.5
...............
1.9

Type of outlet
Tire, battery, and accessory dealer
Dry goods, general merchandise
.
Gasoline service stations
.
Lumber yards
Heating and plumbing equipment dealers
Building materials dealer
Music stores..
.

.
....
....
...
...
.

.

Percent
distribution
of sales
1.4
.4
.2
.2
.2
.1
,1

It is impossible within the limited samples
to include all 14 outlet types as strata with at
least one quotation allocated to each. Rather
than consolidate the miscellaneous outlet types
into a combined stratum, cutoff sampling is em­
ployed. In the above example, the 10 outlet
types which account for less than 5 percent of
each of total sales for the merchandise line are
eliminated. Sales of the eliminated outlet types
(12.8 percent) are prorated over the four most
important outlet types. By this procedure, the
10 quotations for major household appliances
are allocated among four outlet types as follows:

Type of outlet
Total
Household appliance stores
Department stores
Radio stores
Furniture stores

.

Adjusted percent
distribution Distribution
of sales
of quotations
100.0
10
63.7
6
16.4
2
12.6
1
7.3
1

The above illustration also demonstrates the
point, mentioned previously, that outlet types
can be reflected properly for a category, but not
necessarily for individual items. For example,
radio stores naturally are insignificant as an
outlet type for sales of washing machines and
the allocation for washing machines must elimi­
nate radio stores.
Allocations for these groupings in the C and
D cities (where merchandise line data are not
available) are based on data obtained from a
BLS "where bought survey." This survey was
conducted in connection with the Consumer Ex­
penditure Surveys. A copy of the schedule is
appended as exhibit C. For other commodity
groupings, e.g., drugs, personal care, automotive
repairs and tires, etc., distributions based on
industry studies such as "Drug Topics" "Food
Field Reporter/9 "1961 Look National Automobile and Tire Survey" were used. These data
were current but usually give a U.S. distribution
pattern rather than a city or regional pattern.
For still other items, no allocations are required
since they are predominantly "one outlet type"
commodities.

In this category, major items are:
Automobiles
Gasoline
Man's haircut
Beauty parlor services

Motion pictures
Shoe repairs
Self-service laundries
Funeral services

Reporter samples for the various medical
services are allocated to the type or types of
physicians reported by the local medical or den­
tal associations as being representative of the
priced service in their locality. Data from the
various medical associations indicate that the
relative importance of the specialist and the
general practitioners differ greatly among cities
depending upon local conditions. No uniform
relationship exists. In some samples, therefore,
a 50/50, 75/25, or other proportion of general
practitioners versus specialists is used, whereas
specialists only are included in other SMSA
samples. For all reporter samples, selection is
limited to physicians or dentists engaged in fulltime private practice. For sample selection, the
list of practitioners was stratified by age and by
geographic location both in the city proper and
in the suburban pricing areas.
Allocation as to type of outlet for food has
been mentioned previously in reference to size
of sample. To be more explicit, the allocation of
the number of stores of different types—inde­
pendent groceries, meat markets, and produce
markets—was made to the three major cate­
gories, meats, fruits and vegetables, and gro­
ceries, based on the approximate distribution of
sales as reported by food stores in the previous
CPI outlet sample (on BLS 1040 Retail Food
Stores Outlet Information). For example, in
one city, BLS independent grocery store report­
ers estimated that meat sales accounted for an
average of some 30 percent of sales. Using this
percent, an estimated dollar value was computed
for meat sales in all independent grocery stores,
i.e., 30 percent of total grocery store sales as
reported by Census. A comparison of the esti­
mated dollar value of meat sales in grocery
stores and meat market sales (Census data)
produces an 80/20 ratio between the two outlet
types, i.e., 80 percent of estimated total meat
sales are made in grocery stores and 20 percent
in meat markets. Outlets in the sample for
meat are allocated in this proportion. This pro­
cedure was used also for the cities not in the
previous index by making use of the percent­
age distributions for cities with similar charac­




teristics; for example, percentage distributions
reported for Atlanta (overlap city) were used
for Nashville (new city), etc. In a few in­
stances an overall average for all overlap cities
was used. The resultant number of meat mar­
kets in the sample ranges from none in some
of the smaller SMSA's to 30 in New York, the
number of produce markets from none in some
cities to 14 in New York, and independent gro­
cery stores from 9 in some of the C stratum
cities to 37 in Chicago.
One further point should be noted regarding
food store allocations. Two substrata of inde­
pendent food stores ("large" and "small") were
set up according to sales volume, i.e., $300,000
and over for the "large" and $50,000 to $299,999
for the "small." Each substratum represented
roughly 50 percent of Census independent gro­
cery store sales volume in 1958. The exclusion
of stores with less than $50,000 sales volume is
based on practical considerations. Although
these stores are numerous, especially in a few
cities, overall they account for a very small per­
cent of total independent grocery store business,
and pricing in these little stores has always been
unsatisfactory—many items are not carried at
all, others not consistently.
Allocation by Geographic Location
Not only is the sample allocated by kind of
store, but an attempt is made to give proper rep­
resentation for nonfoods to the downtown areas,
to neighborhood areas of the central city, and to
the suburban areas, and for food to the city
proper and to suburban areas. Again, it is im­
possible within the small sample sizes to do this
in any precise fashion, except for food.
There is a wide variation in the importance of
central city and suburban area retail sales in
relation to the total retail sales of the metropoli­
tan area. For example, for five of the SMSA's
(Indianapolis, Wichita, Austin, Baton Rouge,
and Durham) suburban areas account for less
than 10 percent of total 1958 retail sales. How­
ever, for seven SMSA's (Los Angeles, Boston,
Pittsburgh, St. Louis, San Francisco, Hartford,
and Lancaster), suburban sales account for over
50 percent of total 1958 retail sales, ranging
from 50.9 percent in St. Louis to 65.5 percent
in San Francisco.
From an operational standpoint—and this is
based mostly but not entirely on cost considera­
tions, it is necessary to restrict pricing in the
61

suburban area of the SMSA to selected pricing
areas. The suburban areas were selected with
probability proportional to 1958 total retail
sales,73 from a listing of suburban communities
(incorporated communities of 2,500 inhabitants
or over) and major retail centers arrayed by
and within counties by total retail sales. For
each of the SMSA's, the number of pricing areas
selected for nonfoods equals the maximum num­
ber of quotations required outside the central
city for any category of outlets. The suburban
pricing area selections are considered a master
sample of areas with the needed quotations dis­
tributed over sampled areas on a random basis.
An exception to the above was made for six
SMSA's. For four SMSA's (Austin, Baton
Rouge, Durham, and Wichita) no suburban
pricing areas were selected, and all pricing is
limited to the central city because of the small
percentage of sales for the suburban areas. For
two other SMSA's (Cedar Rapids and Champaign-Urbana) suburban sales ranged from 10
to 25 percent; however, incorporated suburban
communities are practically nonexistent. There­
fore, in these areas, individual suburban outlets
were selected (where required) from the sub­
urban areas adjacent to the central city.
Within the central city, allocations for nonfoods are made either between the central busi­
ness district and the neighborhood area or,
where data are available, between the central
business district and major retail centers located
in the neighborhood area. No data are available
for the allocation of food outlets within the cen­
tral city. For New York, the distribution of
central city outlet types among the boroughs
was based on the percent distribution of 1958
retail sales for the five boroughs.
Allocations between the central city and sub­
urban areas for most of the commodity and
service outlets are based on 1958 census data on
retail sales by establishment type (2, 3, or
4-digit SIC codes) and on 1958 Census service
establishment receipts. Allocations for inde­
pendent food outlets are based on the special
Census tabulations mentioned previously.
For a merchandise line where several outlet
types are included, the city-suburb allocation is
made separately for each outlet type, rather
78
1958 Census of Business Retail Trade (BC-58-RA Series) and
1958 Census of Business and Central Business District Series (BC58OBD Series).

62



than attempting to maintain the proper alloca­
tion for the combined group of outlet types.
Again, using the major household appliances as
an example:

Type

Number
of
qxiolations

Total
Household appliance stores..
Department stores
Radio stores
Furniture stores

10
6
2
1
1

Allocation
of quotations
Sales Distribution
SuburCentral Suburban Central ban
city
area
city
area
{Percent) {Percent)
6.0
4.0
7
3
3.5
2.5
4
2
1.3
.7
1 1
.6
.4
1 0
.6
.4
1 0

In allocating the number of quotations between
city and suburbs by type of outlet separately for
each specified type and then summing, seven
quotations are allocated to the central city and
three quotations to the suburban area. If first
however, the sales data for the types involved
had been combined and a city-suburban distribu­
tion calculated, six quotations would have been
allocated to the city and four to the suburbs.
Allocation by the individual outlet types, in some
instances, tends to overemphasize the area hav­
ing the greater volume of sales. However, con­
sidering the many data limitations, further
refinement does not appear justified.
For other types of reporters where sales vol­
ume is not applicable, the basis for area alloca­
tions varies, e.g., area representation of hos­
pitals is based on the relative importance of the
number of hospital beds in the city and suburban
areas to the total hospital beds in the SMSA.
Other Allocation
A final stage of allocation of outlets is made
between single and multiunit establishments for
some commodity groups. These allocations are
based on 1958 retail sales for four major group­
ings—general merchandise group; apparel, ac­
cessory stores; furniture, home furnishings,
equipment stores; drug, proprietary stores. In
using these data, the assumption is made that
the relative importance of single and multiunit
establishments is similar for the various types
of stores within the groupings. Since no data
are available for service establishments or home
maintenance contractors, no attempt is made to
distribute quotations between single and multiunit establishments for these types of outlets.
For food, as previously mentioned, the universe
of chain store organizations is included in the
store sample in most cities.

Selection of Outlets
The master sample of establishments obtained
from BOASI, described in chapter IV, was
found useful. However, the limitations of the
data are numerous and considerable supplemen­
tation and improvisation were required. The
samples are most satisfactory for those kinds of
stores where the typical pattern is one of many
small independent establishments and where
quality factors of the items priced are of less
significance, from an outlet standpoint. Filling
stations, barber shops, cleaning and pressing
shops, etc., are examples.
For multiunit establishments, one report is
usually submitted to BOASI for all stores in a
State or region; therefore, the firm would ordi­
narily be listed only for the city in which the
central office of the firm is located or from which
the report is submitted. If listed for individual
cities, the central office address is given, so no
allowance is made for the actual location of the
outlet, especially if it is located in the suburban
area of the SMSA. Hence, the BOASI listings
are especially inadequate as a sampling frame
for large grocery chains and such other chains
as shoe stores, drug stores, apparel stores, and
department stores. For these types of organiza­
tions, it is necessary to compile listings from
various sources: directories of department, drug,
and variety stores; grocery route lists, shipping
center guides; and telephone books.
Another problem encountered in using the
BOASI listings is that the sampling ratios, used
to draw the master establishment samples, failed
to furnish the required number of stores from
both the outlet type and area standpoints. Area
wise, in the larger SMSA's, this was not un­
expected since, as indicated previously, pricing
is restricted to selected suburban communities
and, in some instances, to selected neighborhood
areas or major retail centers. Outlet-type diffi­
culties result from the broad classification cate­
gories used in BOASI reports, thus limiting the
number of acceptable outlet types. For example,
the listing for furniture stores include second­
hand furniture outlets, custom-made furniture
outlets, interior decorators, etc.
In other instances, particularly in small cities,
the address given for an establishment often is
not that of the outlet itself but of the person or
firm who prepared the BOASI report: account­
ing firms, lawyers, individuals whose home ad­




dress is given or the name of a completely dif­
ferent kind of store—e.g., a service station with
a barber shop address, where the barber shop
proprietor also owns the service station.
Despite the many problems and limitations,
such as those mentioned above, the BOASI list­
ings were useful as the starting point for com­
pilation of sampling lists in general.
For some major commodity and service group­
ings, no data are available from the BOASI list­
ings. In these instances other source materials
were used to develop universe listings. For ex­
ample, a list of medical service reporters was
compiled from special professional directories
such as the AMA and ADA directories of doc­
tors and dentists; hospitals from the Journal of
the American Hospital Association, and news­
papers from the N. W. Ayer and Son's Directory—Newspapers and Periodicals. For still
other items such as automobile dealers, movers,
day nurseries, babysitters, etc., telephone direc­
tories proved to be the only available source.
Automobile dealers fall into this latter group
since only certain makes of cars are sampled.
Actual selection of the sample of stores to be
priced for each item was made originally in
Washington following probability techniques.
Different items call for varying numbers of
quotations by kind of store. To illustrate, some
apparel items require only one quotation from
a department store, while others call for two,
three, or a maximum of four. A "master sam­
ple" was drawn for each type of store and serv­
ice outlet, the number drawn being the maxi­
mum needed for any item to be priced. In the
above illustration, four department stores would
have been selected. The particular stores within
this master sample in which each item was to be
priced were assigned by a random method. Simi­
larly for the replicated samples, outlets were
assigned at random with any one outlet kept
in the same subsample for all items priced.
For some categories (home maintenance,
medical care, and restaurants), where past ex­
perience has pointed up a high loss rate in
establishing outlet samples, an original and a
replacement sample were selected.
Replacements
Initiation of pricing of a revised sample of
outlets in the field is a time-consuming opera­
tion. It requires initial contact to store man63

agers to explain the CPI program and to obtain
their cooperation, contacting department man­
agers and selecting volume selling items for each
specification, and recording prices. Frequently
substitutions are required where investigation
shows that the selected outlet does not carry the
type of merchandise to be priced, e.g., high
priced stores carry only luxury merchandise.
There are also some substitutions necessary
where the store has gone out of business or can­
not be located, reporters refuse to cooperate, etc.
An effort was made in the CPI revision to re­
duce the necessity of replacements by allowing
the agents greater latitude in what they could
price.74 If the original specification was not
available in an outlet, the agent was instructed
to price whatever was "nearest to specification,"
unless the quality of the merchandise carried
was clearly outside the quality level purchased
by index families. The final decision on whether
to accept such deviations was made in Washing­
ton. In some instances the deviations reported
were not acceptable and outlet substitutions
were made.
As described earlier, area and outlet-type allo­
cations are approximate ones. Therefore, devia­
tions from original sampling patterns were per­
mitted where it was found that the designated
pattern was not typical of specific areas. Not
only is there great variation among cities in
types of stores available but there have been
important changes in retail distribution since
the 1958 Census. Consequently deviations from
the original allocations were permitted whenever
local investigations indicated that the current
distribution of sales by types of store or between
the central city and the suburbs was out of line
with the Washington office allocations for the
outlet sample.
Using the original sampling lists, substitute
outlets were supplied by the Washington office
as long as it was expedient. Eventually, because
of time and cost factors, it was necessary to have
the agents select replacement outlets, subject to
a fairly exact description of the type of outlet
wanted, for example, a family clothing store in a
suburban locality similar to the outlet being re­
placed. As pricing has progressed over time,
replacement of outlets has been a continuing
74
One of the recommendations of the Price Statistics Review Com­
mittee was for less rigid specification pricing,

64



procedure to maintain the outlet sample at re­
quired levels. Replacements are selected by the
agents based on the controls mentioned above
regarding type and location of outlet.
Summary
Outlet samples are designed to represent retail
stores, establishments and individuals selling
goods or services to wage-earner clerical work­
ers and their families throughout each SMSA.
Sample selections are made separately in the
Washington office for each item priced, using
statistical probability methods to the extent pos­
sible. Samples are selected to be approximately
self-weighting with respect to type of ownership
(single or multiunit store), type of outlet (de­
partment store, specialty store, grocery store,
etc.) and geographical location within the
SMSA (downtown business district, neighbor­
hood area, or suburban area). For food stores,
sales volume is also taken into account and the
sample is divided into three distinct strata: (1)
chain grocery; (2) large independent stores and
large chain meat or produce markets (over
$300,000 sales in SMSA); and (3) small inde­
pendent stores and small chain meat or produce
markets (under $300,000 sales in SMSA). For
analysis purposes, in some cities one master
sample was selected and divided ex post facto
into two different (replication) samples.
As stated previously, no claim is made that
the outlet sample is, strictly speaking, a prob­
ability sample. Instead a more accurate descrip­
tion might be that it is a compromise between a
completely purposive sample which might be
termed "specification determined," as it was in
the previous index, and a probability sample.
It is still too early to evaluate fully all of the
innovations and techniques used in contrast to
the procedures previously used in establishing
outlet samples. Some conclusions, however, can
be made:
1. With the inclusion of suburban outlets,
price data should be more representative of in­
dex workers' purchases.
2. With greater representation of specialty
stores, the data are more typical of merchandise
distribution patterns.
3. By use of the objective selection proce­
dures, agent bias in selecting outlets is mini­
mized to a great extent.

4. Some of the more recent merchandising
trends have been reflected, i.e., discount opera­
tions, bantam grocery, or convenience-type gro­
cery chains, etc.
5. By "nearest to specification" pricing, data
for new models, styles, and materials should be
reflected in the pricing at an earlier stage than
previously.
6. Operationally, it is a far more costly sam­
ple both from the standpoint of the number of
outlets priced and the area covered.
7. By use of predetermined outlet selections,
the overall initial workload was simplified for




the many new agents who participated in the
revision program.
8. It is the type of sample that should be con­
stantly reviewed and updated, as more recent
data are available, to assure that proportional
allocations and distribution patterns are main­
tained as accurately as possible. For example, a
greater shift to shopping centers might be in
order; a change in the boundary between the
"large" and "small" strata of independent food
stores possibly should be made if more current
data become available.

65

Chapter IX. Pricing Procedures*
The pricing and calculation procedures for
the Consumer Price Index are designed to carry
out, as precisely as resources permit, the general
concepts and principles described in chapter III.
The index, though published monthly, does
not refer to any definite date within the month.
Except for food, for which pricing is done on 3
specific days each month in all cities, and for
rents and items collected by mail which relate to
the 15th of the month, pricing is conducted
throughout a month and prices for each item
relate to the day of the agent's visit. In addi­
tion, pricing is scheduled on a regular quarterly
cycle in 45 of the 50 cities or 51 of the 56 cities.
Only the five largest cities are surveyed every
month. For the U.S. index, estimates are made
each month for all sample cities not priced dur­
ing a month by holding prices constant or by
other means, as explained later.
The index is oriented toward calculation of
price changes between adjacent pricing periods,
not toward the measurement of representative
dollars and cents prices at a point in time. Col­
lection and calculation procedures stress com­
parability of prices and quality from one date
to the next for a given reporter, and not com­
parability among reporters or among cities at
any given time.
Definition of Price
The concept of the Consumer Price Index re­
quires measurement of price change for goods
and services of constant or at least equivalent
quality, but the price data must be observed in
a market in which changes occur frequently in
the kinds of goods and services offered, in their
qualities, and in their terms of sale. This situa­
tion gives rise to persistent and complex prob­
lems of adjusting for quality.75
* Most of the material in this chapter was included in a prelimi­
nary report, The Consumer Price Index: Pricing and Calculation Procedures, op. cit.
« Ethel D. Hoover, "The CPI and Problems of Quality Change,"
Monthly Labor Review, November 1961, pp. 1175-1185,

66



The concept of "price" is theoretically clear
but in practice it defies easy and uniform defi­
nition. In its most restricted sense, "price" in
the CPI refers to the price charged for a par­
ticular brand, identifiable by style number, on
a single sale in an outlet at specific terms of
sale. The concept can be, and is, extended fairly
readily to cover different transactions made
under stated conditions and at identical prices
for each transaction. Extension beyond this
leads to the idea of "average price" for similar
sales and finally to "realized price" for a broadly
defined bundle of goods and services. For a few
commodity areas for which there is little stand­
ardization of quality, notably home purchase,
used cars, and mortgage interest, realized prices
on actual transactions are accepted, pending de­
velopment of improved methods more in line
with the index philosophy.
The goal in measurement is "pure price
change" between two points in time with qual­
ity of goods sold and terms of sale as nearly
identical as possible; unavoidably it usually be­
comes average price on similar transactions.
The dilemma for the technician is the definition
of similarity, i.e., the limits within which price
differences can be tolerated when transactions
are not strictly comparable. No hard and fast
criteria can be outlined, and decisions once made
must be constantly reexamined as market con­
ditions change or new resources are allocated to
price collection.
Pervading BLS operations is the constant
search for better methods of handling the qual­
ity problem. Up to now, quality has been defined
by the Bureau in terms of physical character­
istics, rather than anticipated durability of per­
formance or other intangible features.
In theory, prices used for the index should in­
clude all applicable taxes and credit charges, as
explained in chapter III. In practice, prices are
cash prices. Although credit charges are in­
cluded in the weighting structure, they are not
priced because of practical difficulties of collect-

ing data and measuring changes for the index
in a consistent manner, and not on conceptual
grounds. The effect of sales and excise taxes is
reflected in the index by a separate operation so
that the prices for individual items can be tabu­
lated and processed excluding taxes, to reduce
tabulating costs.
Specification Pricing76
Underlying the BLS pricing procedures is
"specification pricing," the chief tool since 1934
for defining similarity and insuring constant
quality of goods priced. Specification pricing is
practiced for the great majority of commodities
in the index and also for many of the services.
A specification is a detailed description of the
physical characteristics of an item which are
judged to determine its quality and influence its
price. It may include features which aid in iden­
tifying an item from one pricing date to the
next and from one store to another. For a few
items, the brand name or the model number, as
for automobiles, also becomes part of the defined
quality. The BLS relies on assistance from pro­
ducers and retailers as well as on its own pric­
ing experience in developing original specifica­
tions and in making changes to meet new
market situations. A specification does not de­
lineate a precise quality, since to do so would
preclude obtaining enough price quotations. In­
stead, it defines a relatively narrow range of
qualities, within which prices are averaged for
the index and outside of which they are not.
There is considerable variation among different
commodity groups as to the precision possible.
Scattered throughout the CPI complex of
items are some for which nationally advertised
brands command a distinct premium over socalled local brands. To obviate occurrence of
spurious price changes by substitution of one
type for another, a few of the specifications,
such as hosiery, dungarees, men's shirts, and
beer, specify national or local brands. The desig­
nation of "nationally advertised" brands is pre­
scribed by commodity specialists. That such list­
ings are not made more extensively is due chiefly
to the difficulty of keeping brand lists complete
and up to date. For most items in the national
index, however, it is believed that the effect of
TO
Ibid., see also Average Retail Prices: Collection and Calculation
Techniques and Problems (BLS Bulletin 1182, 1955), for a more com­
plete description of specification pricing*.




substitutions from national to local brands and
vice versa tends to balance out.
BLS specifications involve an elaborate system
of regular and alternative specifications, choices
of noncomparable features within a general spec­
ification, and city and outlet deviations. The
system has evolved as a means of coping with
the variety of kinds of goods and services and
qualities offered in the market place. The par­
ticular nomenclature of the system signifies for
pricing agents and tabulating clerks the appro­
priate procedures for making price comparisons
from one period to the next.
The Price Statistics Review Committee en­
dorsed specification pricing in principle but rec­
ommended adoption of a more flexible system.
Partly because of its recommendation, but also
because of the fact that independent selection77
of outlets without regard to availability of par­
ticular specifications was incompatible with
rigid specification pricing, the Bureau revised
its procedures for the new index to some extent,
while maintaining the fundamental aspects of
specification pricing.
Rules for the new index permit pricing of
regular and alternate specifications, or choices
within specification, and even volume-selling
items deviating from specification if necessary.
However, in measuring price change for the
index, the same quality is compared from period
to period within any one outlet. For example,
for men's shirts, the agent may choose within
any one outlet nationally advertised or not na­
tionally advertised brands, and within each,
wash and wear finish with a range of thread
count of 136x60-68 or residual shrinkage 1 per­
cent or less with a thread count of either 136x60
or 128x68. However, she prices the same qual­
ity at subsequent periods. Although the agent
attempts to price to a preferred specification, she
is permitted to price an item fitting an alter­
nate specification or deviating from specification
in major or minor respects. For example, the
specification for upholstered living room furni­
ture calls for inexpensive grade covering and
includes a sofa with two cushions. Medium or
good grade covering would be judged a major
deviation; a sofa with three cushions a minor
deviation. The agent records the nature of the
w See chapter VIII, "Outlet Samples, 1964 Index," for a descrip­
tion of sampling techniques.

67

deviation in detail. In effect, this implies a sys­
tem of outlet specifications described by the
pricing agent.
Items deviating in minor respects are treated
as meeting specifications. Price differences aris­
ing from alternate specifications or major devia­
tions are not considered legitimate price change
for the index. Such prices are introduced by
"linking"; in other words, the difference in price
level between the specification quality and the
deviating quality is not treated as a price change
at the time of introduction.
This adaptation of pricing principles has ma­
terially reduced the usefulness of the index price
data for purposes of intercity and international
comparisons, budgets and family allowances
studies, and market analyses. Supplementation
of prices collected for the index is essential to
achieve reliable measures of average prices at
a point in time.
Price Collection
The majority of priced items can be grouped
into classes—food, rent, commodities other than
food, and services. Prices may be obtained by
mail, by personal visit, or telephone, or from
manuals and other secondary sources. Prices are
obtained from the same sample of reporters in
successive periods. Any necessary changes in
the sample are introduced by linking. In all
cases, cooperation of reporters is entirely volun­
tary.
Pricing by personal visit is governed by gen­
eral rules but with numerous adaptations and
exceptions for particular situations. These are
prescribed in detailed specification and collec­
tion manuals.78 These manuals are looseleaf, and
are kept up to date by regular monthly revisions
to take care of market changes or to introduce
better procedures. At the first pricing of a
specification, the agent, by consultation with de­
partment managers, determines the volume-sell­
ing item within the specification in each outlet,
or for a few items, notably appliances and rayon
dresses, several volume-selling brands. By this
means, a sampling of different brands and styles
is accomplished for the outlet sample as a whole.
The agent continues to price the identical item
in each outlet each pricing period as long as it
78
These manuals are available for study by interested persons, on
request to the Bureau*

68



is stocked and sold in reasonable quantities. If
the item previously priced is not available in a
store, the agent selects a substitute, also con­
forming to the same specification if possible.
Except for those items for which pricing is con­
trolled by brand name, any difference in price
between the item originally priced and a new
one meeting specification is reflected as price
change for the index.
Food
Food prices are collected in 1,775 food stores
by personal visit by part-time agents during 3
specified consecutive days each month in each of
the 56 index cities (50 prior to January 1966).
Even though all prices are not identical for all
stores of a chain organization, in most cases only
one outlet is priced to represent all stores of the
chain organization within an SMSA. This ap­
pears to be the most efficient allocation of re­
sources among stores. When chains are known to
operate stores having different pricing policies,
more than one store may be priced. A few major
chains require collection of prices from the cen­
tral office or district warehouse.
The agent relies on a manual of specifications,
which are generally less detailed and precise
for food than for other commodities and serv­
ices. The price is observed by inspection of price
markings of items on the shelf or, where items
are not marked, prices are obtained from the
store manager or a designated clerk. In most
cases the price is the one prevailing on the day
of the agent's visit, including multiple unit
prices, sale prices, and specials. For fresh meats,
poultry, and fish, which are sold only on week­
ends in some stores, the agent is permitted to
report the price in effect for the previous week­
end. Prices are collected on Tuesday, Wednes­
day, and Thursday of the given week in order
to incorporate both first-of-the-week and end-ofthe-week prices. A store is usually scheduled for
pricing on the same day each month. Stores
priced on Thursday represent a little less than
half the total weight and about three-fourths of
the chain weight.
The agent carries a schedule, containing
prices collected for the previous month, together
with complete descriptions of the item priced.
At subsequent visits, the agent prices the iden­
tical items, except that if another meeting spec­
ification outsells the one originally selected by

as much as 2 to 1, she substitutes the new item,
prices of which are compared directly. The 2 to
1 rule is a rule of thumb, set up to balance the
need for maintaining comparability against the
requirement for pricing popular items. Substi­
tutions and unusual price changes are carefully
explained in agent's narrative reports.
Other Commodities and Services
Pricing of other commodities and services is
costly; largely for this reason, the outlet sample
size for any one item is considerably less than
for food, usually 4 to 6 outlets. However, for
nonfood commodities and services as a whole,
about 16,000 outlets are contacted by Bureau
agents. Pricing is done on the regular monthly
or quarterly cycle shown in appendix table XII.
Identification and control of quality is more
troublesome for these items than for food, and
the specifications and pricing rules are more
complex. Field work is carried out by full-time
agents who undergo very intensive training and
work under the immediate supervision of a
more experienced agent before undertaking in­
dependent pricing. In addition to the specifica­
tions, agents are provided with lists of illustra­
tive brands, swatches, and other guides to qual­
ity for purposes of identification. Collection in a
single city and its suburbs extends over a period
of several weeks to a month and occupies the
time of several agents in the largest cities. A
few items—fuel, telephone, public transit, etc.
—for which identification of quality is fairly
simple, are reported monthly by mail question­
naire.
In addition to the rule for initial selection of
the largest selling item conforming to specifica­
tion, the one chosen must be regularly sold upto-date merchandise in good condition, and it
must be available in a reasonable assortment. If
nothing to fit specification is available in a store,
the agent prices an item "nearest to specification" and describes in detail the features which
deviate from the specified quality. She distin­
guishes carefully between trivial points which
really do not affect price or quality and major
deviations which do. The agent's notations gov­
ern decisions in the Washington office on the
proper method of introducing these prices into
the index.




Brand name within specification is an addi­
tional controlling factor in the pricing of auto­
mobiles and heavy appliances. For appliances,
a volume-selling brand from each of two manu­
facturers is selected in each store and price com­
parisons are restricted to identical brands.
Manufacturers' descriptions at each year's
model changeover provide information as to
which specific model shall be compared with the
last year's model. For automobiles, prices for
each dealer include customary optional equip­
ment.
Once chosen, the same item is priced as long
as it is available in reasonable supply. If it is
only temporarily out of stock, the agent asks
for the last regular price in effect since the 15th
of the previous month. In most cases, she re­
ports the regular price-tag price, apparent from
personal inspection of the merchandise. Sale
prices are accepted provided the sale lasts at
least 1 week and the merchandise is available in
good assortment and in good condition. Clear­
ance sales are excluded. These restrictions on
use of sale prices insure use of popular, gener­
ally available items. Discounts and concessions
are deducted from the price recorded if they
apply generally to all customers. For automo­
biles, electrical appliances, TV and radio sets,
and tires, for which bargaining between buyer
and seller is customary, the agent seeks the re­
porter's estimate of the average concession (or
overallowance on a trade-in). No acceptable
procedure has yet been developed for handling
"tie-in sales," trading stamps, or special deals
which is consistent with the constant-qualityconstant-quantity concept of the index.
Services
Pricing procedures are substantially the same
for most services as for nonfood commodities.
In many service establishments, for example,
beauty and barber shops, bowling alleys, movies,
laundry and dry cleaning establishments, etc.,
fees are standard posted prices just as in other
retail establishments. There are many service
reporters, however, such as doctors, hospitals,
repairmen, contractors, lawyers, and funeral di­
rectors, for whom prices cannot be observed but
must be obtained by interviewing the reporter.
For some of these, prices depend upon the
nature of a particular job. In such cases the
69

agent requests the typical or most common rate
for a specified service.
Rent
The great variation in characteristics of hous­
ing units makes specification pricing of rents,
home purchase, and property tax an unsuitable
technique. Techniques for dealing with home
purchase and property tax are discussed in
chapter X. Change in rents is measured from
large samples of rental units which include the
same units at successive periods. No substitu­
tion of nearby units is permitted. Until the
1964 revision, BLS followed a practice of mak­
ing recurring dwelling unit surveys in each city
about every 3 years for the purpose of gradual
revision of rent samples. Units from newly
developed areas of the city were added system­
atically; existing areas were relisted and new
units substituted. During the 1964 revision, rent
samples were completely redrawn and a system
similar to the recurring dwelling unit surveys,
but less costly, is being planned as a means of
replenishing rent samples.
Since new units are customarily linked into
the sample the difference in rent level between
new units and existing units is not reflected as
rent change for the index. This gave rise to the
"new bias" in the rent index for which correc­
tion was made as part of the interim adjustment
in 1950 described in chapter I.
In the same direction as the new unit bias in
the rent procedure is the "aging bias" arising
from the obvious fact that a rental unit is 1
month older each month and can be presumed to
be worth less. It follows that prices for units of
identical quality would be higher. No adequate
means of adjusting for this have been developed
but it is not considered serious for short-run
movements.
Monthly rent charge is obtained in each city
by part-time agents every 2 months or every 3
months by personal visit or telephone inquiry
to tenants of specified units in different samples.
In most cities, two subsamples of up to 500
rental units each are drawn, with each sample
priced semi-annually in different months. Thus,

70



information is acquired for 1 of the 2 samples
each quarter. In the five largest cities, three subsamples of 500 each are contacted semiannually
in different calendar months, providing data for
one of the subsamples every 2 months. When­
ever a change from the last collection is reported
for a particular rental unit, the date of rent
change is noted. This method permits calcula­
tion of rent change retroactively for any desired
period and provides a basis for incorporating
an adjustment of earlier "error" in earlier subsamples in the current month's index level.
Since back data are not corrected, this procedure
results in a less exact measurement of current
price change, although it yields a more correct
long-term measure.
Rent measurement is complicated by the need
for data in each period concerning the particular
facilities furnished and included in the rent
charge. Inability to obtain such data success­
fully by mail, as well as a high nonresponse rate,
accounts for abandoning the mail questionnaire
formerly used. The agent uses a detailed check­
list covering fuel, gas and electricity, telephone,
garage, furniture, water, maid service, switch­
board service, etc. The Bureau either adjusts
the monthly rent for the estimated value of any
changes in facilities included or links out the
effect of the changes. The estimated values are
for the specific housing unit when they can be
ascertained. Otherwise predetermined factors
representing typical or average values derived
from other sources apply.
Mail Collection
Price collection by means of a mail shuttle
schedule is followed for a few items in the in­
dex, including fuel, gas and electricity, public
transit, water rates, and newspapers. These are
items for which identification of quality is fairly
simple. A few other items, notably automobile
and property insurance rates, are obtained from
published rate manuals.
Prices for a number of items in the index re­
quire special techniques or come from secondary
sources. Pricing procedures for these will be
discussed, along with processing procedures, in
the following chapter.

Chapter X. Calculation Procedures*
As explained in chapter III, the index is a
base-weighted index of price relatives. In prac­
tice it is not calculated directly with reference
to the base period but, for convenience, by up­
dating the previous month's index according to
the formula:
T

r£(Pi-iqa)(Pi/Pi-i)1

T

in which:
Ii = the index for the current month (currently
with reference to 1957-59=100)
i = current month
p = price of an individual sample item
q as physical quantity weight allocated to the
sample item
a = period to which index weights relate (cur­
rently average of 1960-61.)
This is mathematically equivalent to a quantity
weighted index:

i 1== §^4xioo

S(p«q.)
It is not exactly equivalent to a true Laspeyres
index:

! E^xl00
S(Poqo)

where the weights of the index refer to the
mathematical base of the index (o).
Although constant physical quantity weights
are implicit in the index, in reality the constant
q's are not calculated separately. Rather, the
price relatives pi/pi-i or Pi/pi-3 are applied to
the previous month's (pi-iqa) or previous quar­
ter's (Pi~3qa) values to derive the current
month's values, pi qa. This practice is used be­
cause chaining together monthly or quarterly
price changes to construct current pq values
provides requisite flexibility for substitutions of
items, outlets, and weights.
* Much of the material in this chapter was included in a prelimi­
nary report, The Consumer Price Index: Pricing and Calculation
Procedures, op. cit,




Since the market basket priced for the index
is a sample of all items and services purchased
by consumers, each "q a " is in some sense a com­
posite of the bundle of goods and services repre­
sented by each priced item. The "p's" refer to
the sample item priced. The "q's" are not addi­
tive, as are the expenditures, from the basic
family expenditure survey. Synthetic q's could
be computed, if there were any reason for so
doing, by dividing the base price of each sample
item into the total expenditures by consumers
for the bundle of items represented. But such
q's would not be meaningful in view of probabil­
ity selection of items and the way in which
weights were allocated to the probability items.
(See chapter VII for explanation.) These im­
plicit q's will remain fixed as long as the sample
of items and reporters remains unchanged.
Since links of one sort or another due to specifi­
cation revisions, outlet replacements, etc., occur
frequently as part of ordinary maintenance,
calculation of the index by multiplying prices
by quantities is much more difficult than chain­
ing price relatives.
Until January 1966 the indexes were not ad­
justed for seasonal variation in prices.79 Sea­
sonal factors based on the old series were pub­
lished, permitting users who wished to do so to
calculate seasonally adjusted indexes.80 With the
January 1964 revision, these factors became of
questionable value for adjustment of the new
series indexes, except for major group totals.
With the January 1966 release the Bureau initi­
ated regular publication of seasonally adjusted
U.S. indexes for those groups and subgroups
having significant seasonal variation. The all
TO
Victor Zarnowitz, in Staff Paper 5, "Index Numbers and the
Seasonality of Quantities and Prices," prepared for the Price Sta­
tistics Review Committee of the National Bureau of Economic Re­
search in 1959-60, pointed out that the real problem was how to
reflect seasonal variation in prices and consumption, not how to re­
move the seasonal element. He also maintained that use of annual
weights amounted to partial adjustment for seasonal variation, at
least for quantity consumption weights.
80
See Seasonal Factors, Consumer Price Index: Selected Series,
June 1953-May 1961 (BLS Bulletin 1366, 1963).

71

items index was not adjusted because there is
so little seasonal variation.
Unking Old and New Series
Although the Consumer Price Index has un­
dergone at least three major weight revisions,
three changes in base period, as well as con­
tinuing changes in reporter samples, specifica­
tions priced, and processing procedures, con­
tinuous indexes are shown in historical tables
from 1913 to date. This is made possible by
"linking" or "splicing." This means double cal­
culation for a single date of old and new sam­
ples, with old samples or weights used for
comparison with earlier periods and new sam­
ples or weights for comparison with later
periods. In theory, this presents no problems,
but in practice it presents difficult operational
problems.
In the comprehensive revision of the index
completed in January 1964, indexes for the
U.S., both for families and single workers com­
bined and for families only, for all items and
for component groups, start at the level of the
December 1963 (link month) indexes previously
computed. New (1960-61) expenditure weights
and average prices, as well as the old series
weights and prices, were adjusted to or com­
piled as of December 1963. New data were
substituted for the old December data and as­
signed the numerical level of the previously pub­
lished indexes.
Linking may be accomplished either by multi­
plying the link month (December) index by
subsequent price changes or by computing a fic­
titious reference base period value comparable
with revised weights. This reference value is
computed by dividing the revised expenditure
weight for the link month by the index for the
link month previously computed with old
weights. Subsequent indexes can be computed
with reference to this value. In actual prac­
tice,81 current month's indexes are usually com­
puted by multiplying current index pq values
by a base reciprocal obtained by dividing the
81
To illustrate, assume link month index of 125.0 and revised
weights of $500 and current period weights of $600, then the current
period's index is 150.0, calculated:

(W

600
600
500-4- 125 or 4
(c) 600 X 4H or 600 X .25

72



link month index by the new expenditure
weights.
The average price change for all items from
the 1957-59 base to any date after December
1963 is the product of the average change up to
December with old weights and after December
with new weights. All other links—of stores or
specifications—which pervade the index system,
are based on the same principle.
Staggered pricing cycles, unavailability of
prices for all items in the link month, reclassifi­
cations of items, and other difficulties greatly
complicate the linking process in its details.
Built into the U.S. indexes for December 1963
are estimates for cities last priced in October
and November. In the going index, quarterly
changes from October or from November are
applied directly to the October or November
index cost-weights—thus eliminating, from the
longrun index movements, errors in December
estimates. Across the December 1963 link
month, only an imperfect correction is possible,
due to differences in the sample of items and the
relative weights in the two indexes. Quarterly
change for the new samples of items, outlets,
and cities straddling the December 1963 U.S.
link is conceived of as replacing quarterly
change measured by the old samples.
In the linking operation, overlap items and
overlap cities presented different problems from
new items and new cities. Gasoline, for ex­
ample, caused serious difficulties in the overlap
cities because of its large weight and the preva­
lence of "price wars." Depending upon the
particular date of price collection for old and
new limited samples of gasoline stations, re­
ported price changes varied greatly. Special
care was necessary to make sure that both the
substantial reductions to sale prices and the
subsequent returns to regular prices were
handled consistently in the old and new series
across the link month. For example, a reduction
might be reflected in the old index sample but
not in the new, if the former sample was priced
early in the month and prices returned to regu­
lar price before the new sample was priced.
Seasonal items not available in the link month
were, in most cases, imputed to the movement
of other priced items until prices became avail­
able for two pricing periods. Overlap items in
overlap cities could be treated consistently for
old and new series.

Processing, General
After collection and, in some cases, tabulation
of prices, the processing steps for the index
include all or some of the following:
1. Editing and adjustment of prices for com­
parability from one period to the next.
2. Calculation of price relatives pi/pi-* for
each item, based on average prices for the same
reporting outlets in two consecutive periods.
3. Calculation of current values (pi q»)—base
quantity at current prices—generally termed in­
dex cost weights or expenditure weights.
4. Aggregation of pi q* values by city.
5. Estimation of unpriced cities and aggre­
gation of city pi q* values to U.S. totals.
6. Calculation of indexes on base period.
If dollars and cents prices are to be published,
as for food, additional processes are required.
Editing
The first major step in processing prices is a
careful review of each individual quotation and
quantitative adjustment of prices, if necessary
and possible, for comparability of quality with
the last price reported for the same outlet. This
is a key operation. The accuracy of the final
index depends in large measure upon the com­
posite of all the individual decisions made in this
phase. Editing rules are made as objective as
possible, but professional judgment also comes
into play. When possible, blanket instructions
are issued to insure consistent handling of like
situations at every occurrence.
If identical items (brand, size, etc.) are
priced in an outlet in adjacent periods, any dif­
ference in price is true price change and is so
reflected. If, however, a new item has been
substituted, a decision is made as to whether
(1) the prices of the two shall be compared di­
rectly, (2) the price difference shall be linked
out of the current period's comparison, or (3)
preferably, but contingent upon required data,
a quantitative adjustment of the reported price
for the value of the difference in quality shall be
made. Direct comparison of prices of different
items may incorporate in the index the effect of
quality difference. Linking so as to eliminate
the entire price difference may exclude some real
price change, since, characteristically, many
price adjustments are effected by manufacturers




concurrently with product changes. Use of quan­
titative adjustments is limited because of the
difficulty of obtaining the information for deter­
mining appropriate factors for each of the hun­
dreds of substitutions occurring throughout the
country.
Changes in quality occur for goods produced
at different times which are seemingly iden­
tical. In the main, the changes creep in so
gradually as to be undetectable. Moreover, the
changes normally lie within the range of quality
defined in the specification.
In keeping with the theory of specification
pricing, prices of substitute items fitting a given
specification almost always are compared di­
rectly with the previous month's price. Any
change in price due to the small spread in
quality within specification is accepted for the
index. Occasional exceptions are made for items
such as furniture, for which identification of
quality is especially difficult, when substitutions
are made from one end of the range to another.
In such cases, the linking procedure described
in the following paragraph may be used.
Substitutions outside the quality range of the
specification usually are linked, because quanti­
tative evaluation of quality differences is seldom
possible. When prices are obtainable concur­
rently for old and new items, the market price
differential can be considered a measure of the
value of the difference in quality. More often
than not, however, substitutions occur because
the previous item has been discontinued. Care­
ful decisions must be made by commodity spe­
cialists on the basis of their own knowledge and
field agents' notations as to the degree of varia­
tion from the quality last priced. Frequently,
offsetting quality changes are involved. Minor
deviations from specification are disregarded
and their price differentials accepted as price
change: major deviations are either excluded at
the first pricing or adjustments made for the
quality difference. In subsequent periods, if
prices of the same deviating item are available
for two consecutive periods, major deviations
are included in the calculation of price relatives
along with those conforming to specification.
This is done by the ordinary linking procedure,
i.e., the average price for the previous period is
recomputed to include the price of the deviating
item which had been excluded, because of lack
of comparable prices for the previous period,
73

and the recomputed price is compared with the
price for the current period which includes the
deviating item.
In most instances, it is not possible to obtain
precise evaluations of quality changes, even for
commodities which have easily identifiable
physical characteristics. The refrigerator, which
is subject to annual model changes, illustrates a
number of the problems encountered. The cur­
rent CPI specification describes a refrigeratorfreezer with two outside doors and a true
freezer at the top. The specification permits a
size range of 11.5 to 14.5 cubic feet and allows
the pricing of both "frost-free" and conven­
tional defrost models. It is obvious that such a
specification permits a fairly wide latitude in
selecting models for continuous pricing within
any one outlet and that all models priced are
not comparable. The inclusion of more than one
"quality" in the specification is necessary in
order to obtain a sufficient number of quotes for
all brands carried in reporting outlets. The
problems caused by substitution of a model with
automatic defrost for one without this feature
are taken care of through provision for separate
choices, identified in the specification. Substi­
tutions of one for the other are routinely linked
into the index, in the same way as alternate
specifications. Substitutions of different size
refrigerators within the range permitted by the
specification are somewhat more difficult to
evaluate, since other features may also be added
or deleted at the same time the size is changed.
Other things being equal, however, a size dif­
ference between models of 0.4 cubic feet or less
has been considered a minor quality difference
and prices are compared directly. Greater dif­
ferences in size are considered major and prices
of the new models are linked into the index,
even though they may both meet the specifica­
tion.
Because of the larger sample of quotations for
food than for other commodities and services,
as well as greater uniformity in goods pro­
duced, editing of food prices has been routinized
to a great extent. Prices for foods within an
accepted range of weights are mathematically
converted to a common weight. Prices for dif­
ferent size cans of processed fruits and vege­
tables, for example, are adjusted to a common
size on the basis of the estimated relative weight
of contents. In a few cases, special factors based
74



on usual market differentials are employed to
adjust to a given standard quality.82 Prices
seriously out of line are excluded from the calcu­
lation and followup inquiries are made to the
field agent for the next pricing. For nonfood
items, the small number of quotations requires
careful editing and inclusion of every quotation
reported, if at all possible.
Calculation of Price Relatives by City
To avoid the complications of positive and
negative changes, the Bureau utilizes price rela­
tives rather than percentage increases and
decreases. A price relative is the ratio of price
in the current period to that of a previous period
times 100, e.g., if a $10 item advances to $12,
the price relative is 120; if it drops to $8, the
price relative is 80.
Price relatives between two dates for a specifi­
cation can be computed in at least three differ­
ent ways: (1) the relative of average prices
for identical outlets, (2) the relative of aver­
age prices for all reporting outlets, (3) the
average of price relatives for reporting outlets.
All three methods may be employed with or
without internal weights. The three yield dif­
ferent answers. Although instances of each can
be found within the CPI labyrinth, the first
method (without internal weights) has been the
most commonly used.
Method 2, the relative of average prices for
all reporting outlets, is generally considered less
efficient for measuring price change than
method 1, because any difference in average
prices due to the difference in the outlet sample
is reflected as a price change for the index.
For example, if the highest priced outlet drops
out of the sample between period 1 and period
2, a price decrease would be reflected in the in­
dex, even if no prices changed. Therefore, this
method is used only when the universe or at
least a large number of reporters is included in
the sample, as for foods in large cities. In such
cases, no one reporter exerts undue influence on
the average.
The choice between the relative of average
prices (method 1) and the average of outlet
relatives (method 3) rests largely on the range
of prices. Method 1 implies equal weights in
83
For example, prices of U.S. good grade rump roast or rib roast
are adjusted upward by 10 percent to make them equivalent to U.S.
choice grade.

terms of quantities of the given specification
represented by that outlet. As for any arith­
metic average, the relative price levels are
implicit weights in the average. Method 3 im­
plies equal weight in terms of dollar sales,
rather than quantities for each outlet, and in the
first of each of two periods being compared. It
is inconsistent with the Laspeyres fixed quantity
weighted formula, since the implicit quantity
weights (dollar sales divided by price) are not
held constant. The relative of average prices
will show a greater average change than the
average of outlet relatives when the highest
priced reporters experience the greatest change,
and conversely a smaller change if the lowest
priced outlets have the greatest change.
Most often, the average of relatives method
yields larger average increases than the rela­
tive of average prices. Evidently this is because
in actuality, both price increases and decreases
tend to be proportionately greater for lower
priced than for higher priced goods.
Method 1 compares the sum of prices for
matched outlets in two periods, i.e., those for
which comparable and usable prices are avail­
able both for current and previous periods.
Average prices are not needed nor are they ordi­
narily computed. If items conform to specifica­
tion in all outlets the spread of prices is small
and outlets receive approximately equal weight.
If, however, there is a wide disparity among
outlets in the level of prices, as when deviations
from specifications are encountered, the highest
priced outlets have a disproportionate effect on
the average change. Relaxation of specification
pricing for the revised index has made this a
more serious problem. However, in preference
to using method 3, averaging price relatives for
each outlet, to cope with the problem of wide
price ranges, which is not consistent with the
fixed quantity weighted formula, quality adjust­
ment factors are being used more extensively
to adjust prices to a comparable basis, permit­
ting use of method 1.
Normally sale prices pose no problem. They
are treated as legitimate price change. De­
creases to sale prices and subsequent returns to
regular prices are handled automatically. It
may happen though, that the initial price col­
lected for a store is a sale price. An increase
from such a "first" at the next pricing period
is eliminated from the index on the basis that




the decrease to the sale price had not been re­
flected previously.
Special Procedures
Collection and calculation procedures for
many items in the index do not correspond to
the general ones described. Among these are
restaurant meals, women's dresses, home pur­
chase, mortgage interest, property tax, property
insurance, telephone, used cars, health insur­
ance, college tuition, college textbooks, maga­
zines, paperbacks, transportation rates, and
hotels.
Restaurant Meals
For restaurant meals, first priced for the in­
dex in 1953, specification pricing and regular
procedures apply only for breakfasts. For
lunches and dinners, specification pricing is im­
practical. Instead, a procedure is used which
bases price changes on a comparison of a num­
ber of identical meals within each restaurant.
This is analogous to the procedure followed for
rents. A master listing of 46 entrees, classified
as beef and veal, pork and lamb, poultry, fish
and seafood, meat-food combinations and meat
substitutes, hot sandwiches, and cold sand­
wiches, has been established. Where printed
menus are available, the agent usually picks up
the menus for each day of a given week. If
printed menus are not available, she copies from
records the prices of all meals on the master list­
ing which were offered on the day of the visit.
Restaurant meals are treated as a monthly
item in all cities. The total sample for each city
has been divided into three independent subsamples. Because prices in any one restaurant
change infrequently, only 1 of the 3 subsamples
is priced each month; prices in the unpriced
subsamples are held constant. This procedure
enables the Bureau to use a larger sample of
outlets than it could handle if all outlets were
priced each month.
The tabulating clerk selects a specified num­
ber of entrees from the master list for each
class, up to a maximum of 11 for lunch and 10
for dinner in each restaurant. The total price
of combination meals—entree, one or two vege­
tables, beverage, and dessert—is built up from
combination meal prices plus a la carte prices
75

of any of the specified components not included.
Sandwich meals include sandwiches and bever­
age only. Prices are compared from one quar­
terly period to the next for the particular meals
available in the two periods in each restaurant.
After a lapse in pricing, if a particular meal
is again offered and used in the index, the cur­
rent price is compared with an "implicit" price
for the previous period, which is computed by
adjusting the last reported price by the price
change for the restaurant reflected in the index
during the intervening period.
Outlet relatives are computed based on the
sum of prices for the particular meals offered in
a restaurant in two consecutive pricing periods.
Adjustments in the total price of a combination
meal are made so that the components of a meal
are comparable in the two periods; no attempt is
made to adjust for changes in size of portions.
Women's Dresses
Style and workmanship, rather than fabric
and other physical characteristics, predominate
as price-making factors for women's inexpen­
sive dresses. Consequently, normal specification
procedures have proven to be ineffectual. Two
broad workmanship standards, with little if any
restrictions on fabric, have been established for
the pricing of women's inexpensive street
dresses, which are made chiefly of manmade
fibers. Four volume-selling items are selected
separately in each store at each pricing and
direct comparisons made between the average
prices of all quotations combined in successive
periods.
Homes and Used Cars
In sales of homes land used cars, identifica­
tion of quality is difficult. Specification pricing
is not feasible and neither is the system of
identical units used for rents. For these items,
transactions reported to secondary sources
furnish a basis for prices realized on actual
sales in the market. Average prices for broad
quality classes combined with fixed weights
minimize the effect of quality on the realized
price. Prices of homes are converted to price
per square foot and reflected in the index by
3-month moving averages to eliminate erratic
fluctuations in each month's index.
76



The data for used cars are not available for
individual cities. Averages for a single State,
or for two States, depending upon the cities'
market areas and the number of sales per
month, are used. Data are confined to two
makes, Ford and Chevrolet, but they are not
standardized in terms of equipment and acces­
sories included. The reporting source edits out
prices of cars in poor condition. In addition,
other extreme values are edited out either by
the reporting agency or by BLS. Similarly,
average prices of homes are not standardized
and prices for larger areas sometimes replace
the limited city samples.
For houses, age breaks (based on year built)
within two major classes, newly built and previ­
ously occupied, are maintained; for used cars,
specific price series and body styles are desig­
nated for 2-, S-, 4-, and 5-year old models of
standard-size Chevrolet and Ford cars. Once a
year a shift is made to homes and used cars 1
year newer. For both, BLS takes account of the
gradual aging within a calendar year through
a system of averaging prices a year apart
in age with gradually shifting weights every
month. Looked at another way, the difference
in price between houses or cars a year apart in
age is considered a measure of the annual de­
crease in value due to aging. Adjustment for
this factor is made in comparing prices for a
given age from 1 month to the next.
Mortgage Interest
Mortgage interest for the CPI is considered
the price of borrowing money for purchase of a
home, or the total interest obligation originally
incurred. The index strives to measure the,
change in the amount of interest required in
current markets at current rates to buy houses
of the same quality and at the same ratio to
purchase price as in the base year. The change
incorporated in the index is the product of
changes in rates on new loans and in purchase
price.
Rates on conventional loans for each SMSA
and State totals for non-SMSA's, are obtained
from special tabulations from the comprehen­
sive monthly survey of loans on purchases of
homes by all types of lending institutions, con­
ducted by the Federal Home Loan Bank Board.
Average interest rates on loans in designated

fuality cells—specified: purchase price classes of
newly built and of previously occupied homes—
are combined for each CPI city with weights
representing, the total amount of loans in these
classes in a reference period. The two rates
for conventional loans are averaged with FHA
and VA rates* the weights representing the total
amount of interest contracted on the four types
of loans in a reference period. These internal
weights are revised annually as new data are
compiled. The base period amount of the loan
is adjusted every month by the CPI change in
purchase price of homes. By this means, the
amount of money borrowed is kept at the same
proportion of the value of a house as in the base
period. A further description of BLS procedures
appears in "Housing Costs in the Consumer
Price Index/' Monthly Labor Review, February
and April of 1956. The procedures described
have remained essentially the same.88

for comprehensive homeowners^ insurance are
treated separately, but in a similar way.
Automobile insurance rates are also obtained;
from rate manuals* Rates for companies con­
forming to standard casualty bureau rates and
for deviating companies are represented in the
sample. Replicated samples of companies and
rating territories within each SMSA are drawn
and rates are combined with internal weights
representing relative volume of business. Rates
are obtained for specified models priced as new
cars for the index and judged to be comparable
from year to year.
Telephone

Tax assessors or other central sources in each
city provide property tax data annually for a
predetermined probability sample of addresses
selected from the Bureau's comprehensive hous­
ing unit surveys. The tax change which is re­
flected in the index once a year is based on a
comparison of the total annual taxes paid in
successive years for the fixed sample of ad­
dresses. Adjustments eliminate the amount of
change in taxes due to special assessments or
capital changes in the property.

In cooperation with all the telephone com­
panies operating in index areas, it has been
possible to set up complete pricing of all tele­
phone services. A base weight structure, as of
June 1963, was developed from complete rate
and usage data obtained from telephone com­
pany records. The base weight structure was
classified in accordance with the appropriate
rate groupings applicable in each CPI metro­
politan area. Changes in rates for all classifica­
tions of services in an area, including intra- and
interstate toll calls, are obtained monthly by
mail from the reporting telephone company.
These changes are applied to the fixed base
period values to derive average price change
monthly in each city. It is expected that these
internal weights will be revised periodically.

Property and Automobile Insurance

Health Insurance84

Property Tax

Property insurance rate manuals on file at
BLS for every rate jurisdiction within the 56
urban places in the CPI make possible complete
monthly pricing of property insurance rates.
Three-year premiums for fire insurance for
brick and frame houses of a specified value of
house are added to those for extended coverage.
The specified house value is median value of
owner-occupied homes from the 1960 Census,
adjusted regularly by changes in purchase price
of homes. Changes in insurance premiums due
to changes in house prices as well as rates are
reflected as price change for the index. Rates
** Except that base weight represents interest contracted for halfterm. See chapter VH.




The weight for health insurance in the CPI,
as for all other items, represents direct expendi­
tures by the index population and excludes the
portion paid for as fringe benefits by the em­
ployer.
Until 1964, premium rates for the most
widely held family group hospitalization and
surgical insurance plans in each city were re­
ported monthly by the Blue Cross and Blue
Shield Organizations. Commercial carriers were
not represented. In addition to the rates, local
plans furnished an evaluation on a propor­
tionate basis in terms of the three major reasons
84
For a more complete explanation, see article by James C.
Daugherty, "Health Insurance in the Revised CPI," Monthly Labor
Review, November 1964, pp. 1299-1800.

77

for the changes in rates: those attributable to
changes in costs of covered services; those asso­
ciated with changes in utilization, i.e., changes
in total claims paid; and changes in benefits.
Changes in rates which were associated with
differences in benefits were considered quality
differences and, as such, were factored out of
the index calculation.

Transportation Fares

In the 1964 revision, following extensive dis­
cussions with insurance representatives, the
use of premiums was dropped, mainly because
of inability to make accurate adjustment for
changes in benefits covered, and the difficulty of
determining and pricing representative plans
offered by private carriers. Instead, health in­
surance is represented by prices for hospital
and professional services and drugs for which
insurance benefits are paid, plus an adjustment
for the retained earnings or overhead cost (the
excess of premium income over claims paid
out).

College tuition and hotel rates are not priced
on a city basis, since expenditures by families
for these are customarily made outside of the
home city. Regional probability samples of
hotels selected from the large CPI cities are
priced by BLS agents. Four regional samples of
hotels were drawn and allocated to the rep­
licated A and B subsamples. Hotels (and
motels) are priced on a regular quarterly pric­
ing cycle for each city; rates are held constant
between quarterly pricing periods. All hotels
are combined to obtain monthly price change
for each regional subsample as a whole. The
appropriate regional subsample price changes
are weighted differentially for the cities in each
region according to the importance of travel
from the city to the four regions, according to
the Travel Survey of 1957 by the U.S. Bureau
of the Census.

Expenditure weights for the claims portion
were assigned to medical services, most of which
were already in the sample to represent direct
expenditures. The base weight for the overhead
portion was divided between Blue Cross and
Blue Shield plans and private carriers on the
basis of data for 1960. The overhead weights
will be escalated for price change from month
to month in each city on the basis of the aver­
age change in prices of the claims portion.
The annual adjustment of changes in the ratio
of retained earnings to income will be based on
national financial data for Blue Cross-Blue
Shield plans and for the commercial insurance
carriers, as reported to the Social Security Ad­
ministration. For convenience, the calculation is
carried out by expressing retention ratios as a
proportion of benefits rather than total income,
since the cost weights for priced services cor­
respond to benefits. These ratios are termed re­
tention factors. The annual relative of change
in the retention factor will be based on weighted
averages of the two ratios for each year, using
data of the previous year as weights. The rela­
tive weights of the two will be adjusted annually
by linking, in such a manner that the change in
proportions of Blue Cross-Blue Shield and com­
mercial carriers is not reflected directly as a
price change.
78



The General Accounting Office furnishes rail,
intercity bus, and plane fares for selected
specific trips from each CPI city. These fares
are converted to rate per mile.
College Tuition and Hotel Rates

The total weight for college tuition was di­
vided into a regional and a local weight for each
A and B size city. Because of the absence of
more specific data, the local weight represents
fees for students who are residents of the State
rather than the city, as determined from a spe­
cial study.85 The entire weight in C and D size
cities was considered regional, that is, the same
regional average tuition fee relative was used
for all the C and D cities of the region.
Fees for undergraduates are reported annual­
ly to the Office of Education by public and pri­
vate institutions accounting for more than 90
percent of total college enrollment. For the CPI,
probability samples of 50 institutions each were
selected for each of four regions from an array
of "reliable" institutions (those reporting con­
sistently) arranged within region by type of in­
stitution, by State, and by the amount of fulltime enrollment. Those selected for each region
** The Home State and Migration of American College Students,
Fall 1958, American Association of Collegiate Registrars and Ad­
missions Officers, Committee on Research and Services, Ohio Univer­
sity, Athens, Ohio, 1959.

were assigned alternatively to A and B subsamples for purposes of replication. These institu­
tions were supplemented by a few additional in­
stitutions in the A and B size cities, for use in
the State samples only. State samples were not
replicated.
Some institutions (mainly public) charge dif­
ferent fees for residents and nonresidents. In
such cases, nonresident fees were used in re­
gional samples and resident in State samples.
The appropriate regional subsample relatives are
used for all cities within a region; the same
"local" relative is used for all cities within a
State.
Books and Magazines
Pricing of books and magazines by specifica­
tion is not practical. Prices of college textbooks
are obtained annually from a small national
sample of nine university and college bookstores,
by BLS agents. Prices are obtained, if possible,
for the principal textbook for a beginning course
in each of 16 subject fields. If no one text can
be designated the key text, prices are reported
for one or more important textbooks, excluding
workbooks and required reading books. When
the text changes from 1 year to the next,
prices of old and new texts are compared direct­
ly, unless one is a paperback and the other a
hardback, which are considered noncomparable.
Prices of magazines and paperback books
come from secondary sources, (1) the ABC
Statement (the semiannual circulation report of
the Audit Bureau of Circulations), and (2)
monthly listings of paperbacks released, pub­
lished by Bestsellers Magazine.
Because of their specialized character and lim­
ited distribution, organizational publications,
publications by professional societies and edu­
cational institutions, and trade journals were
eliminated from consideration for the CPI.
A national sample of 16 magazines of those
issued for general circulation was selected by
probability proportional to importance, using
average circulation in 1960 from Ayer^s Directory of Newspapers and Periodicals as the meas­
ure of importance. Magazines having circula­
tions of 6 million copies or more were considered
certainty selections.
Pricing of magazines in conventional retail
outlets was ruled out because subscription sales,
which are more important than sales of single




copies, are sold primarily through other chan­
nels and because single copies are traditionally
sold at publishers' preticketed prices. Circula­
tion and price data are available from publish­
ers' statements for 6-month periods ending June
30 and December 31. Price categories include,
in addition to single copy prices, two major
types of subscriptions—basic price and reduced
price.
In the index calculation, the quantities of
various types of subscriptions and single copies
sold in the base period (June to December 1960)
are held constant. The index change is based on
a comparison of estimated receipts for these
quantities at current and preceding period
prices. It is planned to review base quantities
about every 3 years and to introduce revised
quantities by linking as necessary.
The paperback industry comprises mass-pro­
duced titles for general circulation and trade or
quality editions for selected uses. The former,
comprising about 85 percent of the total busi­
ness, were selected for the CPI. Only limited
consideration was given to establishing specifica­
tion pricing, since relative prices for paperbacks
do not reflect quality differences in a literary or
physical sense.
Monthly price change is measured by utilizing
a 12-month moving average price of new titles
released each month, in order to reflect the price
movement of previously issued as well as newly
issued mass paperback books. The internal
weight pattern represents the 1962 value of sales
for five major categories of mass paperback
books—novels, factual, mystery-suspense, west­
erns, and classics. Monthly prices used in the
CPI calculations are obtained from the major
trade publication, Bestsellers Magazine.
Calculation of Current Expenditure Weights
Price relatives are applied to index values for
the previous period (Pi-iqa) to derive current
period values (Piq*) for each item in each city.
Totals of all items are compared for the two
pricing periods to determine the average price
change for each city. This procedure is mathe­
matically equivalent to a weighted average of
price relatives with relative values of the pre­
vious period as weights.
If prices are missing for an item, estimates
are made either by holding constant or by im­
putation to the price change for other items in
79

the same grouping. For seasonal items for which
long-term relatives are computed, the current
index expenditure weight (PtqO is calculated
by applying the relative to the pq value at the
end of the previous season. By this means the
interim estimates are canceled and the current
month's value is brought to the correct level.
Aggregation to City and U.S. Totals and
Estimation of Unpriced Cities

in overestimates for some cities, requiring cor­
rections in the opposite direction at the next
regular pricing for the city. For new cars, a
special procedure utilizes cities priced quarterly,
as well as those priced monthly, for estimating
unpriced cities. This is particularly necessary
at the time of introduction of new models to
avoid a lag in reflecting the price change which
normally occurs uniformly throughout the coun­
try at the time of model changeover.
Calculation of Indexes

Item expenditure weights are totaled by sub­
classes within each city, and city class totals
(for prices obtained or estimated) are aggre­
gated to U.S. totals—published groups and sub­
groups and other unpublished analytical or sub­
sidiary groupings. In this process the individual
cities are weighted by population weights based
on the 1960 Census of Population. City weights
equal the proportion of wage-earner clericalworker population represented by the sample
city. The 18 largest cities carry their own
weights; all other cities represent a group of
cities. City population weights are shown in
appendix tables VIII-A and B. In operation,
weighting is done by simple aggregation, since
the relative population weights for the U.S.
index have been built into the expenditure
values for each item in each city. For sim­
plicity, these combined weights have been called
"cost-population" weights. The U.S. totals in­
clude actual or estimated values for all cities.
Inability to do a complete pricing of every
item in all cities each month results in a some­
what imperfect measure of month-to-month
change, but no long-term error. Expenditure
values for quarterly groups in unpriced cities in
interquarterly months are held constant at sub­
class levels from the latest time priced (except
for new cars). This means that quarterly change
for about a third of the cities, instead of monthly
change for all cities, is reflected in the move­
ment of each month's national index. This
method may at times introduce a temporary lag
in measuring price movements for these groups
but, even if so, it is considered preferable to the
alternative technique of estimating price change
in some way such as on the basis of the five
large cities priced monthly. This alternative (in
use for most groups from 1953 to 1962) resulted
80



Indexes are issued for the all items composite
and for commodity groups, for the U.S. average
and for individual Standard Metropolitan
Statistical Areas having a 1960 population of
a million or more. They are computed with
reference to an established base — currently
1957-59 = 100. In actuality this is done, for
convenience, not by division by comparable base
period totals, but by multiplying by an appropri­
ate base period reciprocal.86
Aggregation to U.S. item totals is an extra
operation not essential in computation of group
and all items totals. This operation is carried
out regularly for food, which is processed by
automatic data-processing equipment, and U.S.
indexes based on all cities are published regu­
larly for the individual foods. Monthly U.S.
totals based on all cities are not possible for
most nonfood items because they are priced on a
quarterly cycle, and estimates for unpriced cities
are not made at the item level. Until 1964, in­
dexes for individual nonfood items were com­
puted apart from the regular index work and
published for the month of March, June, Sep­
tember and December based only on the cities
priced in those months. Since not all items are
priced in all cities for the revised index, the
total population weight priced for the probabil­
ity items was too small to permit continuance
of this system. Consequently, from 1964 for­
ward, these indexes are based on all cities, using
the latest available data for each city. Immedi­
ately following completion of the revision, in­
dexes were computed only semiannually; in
March 1966, publication on the quarterly cycle
was resumed.

Jroq»

Jroqa

Chapter XL Publication of Data
The policy of the Bureau is to make the Con­
sumer Price Index and supporting data avail­
able as generally as possible. However, limita­
tions both of the data and of staff resources
make it impossible to satisfy all demands for
price data. Prices and other data are reported
to the BLS in confidence. BLS pledges to its
reporters that their data, and particularly their
identity, will not be disclosed to anyone outside
the Bureau. Consequently, price information is
published in the form of averages and not for
individual firms. The Bureau's success in ob­
taining cooperation of reporters is due in large
part to this policy.

Expenditure
class

29-35
29-30
31-32
33
36-41
36-40
41
42-52
42-44
45-46
47-49
50-52

Major groups and subgroups

Apparel and upkeep
Men's and boys'
Women's and girls'
Footwear
Transportation
Private
Public
Health and recreation
Medical care
Personal care
Reading and recreation
Other goods and services
Special groups

The National Consumer Price Index is released
monthly from the Washington office by means of
a regular press release and a formal press con­
ference late in the month following that to which
the data refer. The release contains a brief
analysis of price movements during the month,
as well as the latest available indexes and per­
cent changes over selected periods. A more com­
plete report is issued about 2 weeks later. U.S.
average indexes are published monthly for the
following list of 28 major groups and subgroups
and 20 special groups:

All items less shelter
All items less food
Commodities
Nondurables
Durables
Services
Commodities less food
Nondurables less food
Apparel commodities
Apparel less footwear
Nondurables less food and apparel
New cars
Used cars
Household durables
Housefurnishings
Services less rent
Household services less rent
Transportation services
Medical care services
Other services

Expenditure
class

City Indexes

National Indexes

EC 1-15
1-14
1-2
3-5
6
7-9
10-14
15
16-28
16-20
16
17-20
21
22-28

Major groups and subgroups

Food
Food at home
Cereals and bakery products
Meats, poultry, and fish
Dairy products
Fruits and vegetables
Other foods at home
Food away from home
Housing
Shelter
Rent
Homeownership
Fuel and utilities
Fuel oil and coal
Gas and electricity
Household furnishings and operation




The Bureau cannot satisfy all demands for
local data. Since the 1964 revision, the Bureau
has adopted a policy of publishing indexes for
only those individual SMSA's having 1,000,000
or more population in 1960. To make this pos­
sible, full samples of items are priced in all such
cities, whereas in most of the smaller cities only
one subsample of items is priced. City indexes
are published on cycle (quarterly priced cities
in appendix table XII) for all items and the 28
major groups and subgroups listed above. Be­
cause many users misinterpret the city indexes
81

as measures of intercity differences in prices,
each report cautions the user of these indexes
as follows: "Comparisons of indexes for indi­
vidual SMSA's show only that prices in one loca­
tion changed more or less than in another. The
SMSA indexes cannot be used to measure dif­
ferences in price levels or in living costs between
areas."
Besides publication of city indexes in the na­
tional press release, statements for the indi­
vidual cities are issued from the Bureau's re­
gional offices on the same day as the national
release. These contain price indexes and anal­
yses of price movements in individual cities
within the region.
Mimeographed tabulations of historical in­
dexes for all available periods for all items and
groups are available back to 1913 ( or the earli­
est available date) for the United States and
individual cities. Whenever the official base
period is changed, the Bureau computes and pub­
lishes new historical tables back to the begin­
ning of the series, as rapidly as possible. In
addition, conversion factors are published for
the convenience of users.
Indexes for Individual Items
Indexes for individual items are published for
the United States only, either in the regular
CPI report or in subsequent reports. Those for
food items are included in the more complete
report on the Consumer Price Index issued about
2 weeks subsequent to the press release.
Until 1964, indexes (for individual articles
and services other than food and fuel) based on
the subsample of cities priced in March, June,
September, and December were calculated quar­
terly and published in a regular report, Quarterly Price Indexes for Selected Items and Groups.
Beginning in 1964, procedures were adjusted to
include estimates for cities not priced in these
months and frequency was reduced temporarily
to semiannual — in June and December. In
March 1966, quarterly publication was re­
sumed. The report is now entitled Consumer
Price Indexes for Selected Items and Groups.
It includes indexes for most of the items priced
for the index.
Generally, the actual price data underlying the
published indexes are not published. This is be­
cause the Bureau's techniques are designed for
measurement of price change, rather than for
82



calculation of representative average prices, and
the samples of reporters are too small for calcu­
lation of representative averages. Under certain
circumstances, however, and with some adapta­
tions, the price data collected for the index are
useful for secondary purposes.
This is the case for U.S. and city average
prices for individual foods, which have been pub­
lished by the Bureau since 1890. The January
1964 revision of the Consumer Price Index in­
troduced changes into price collection procedures
which complicated the calculation and publica­
tion of dollars and cents prices, and required
initiation of a system for estimating prices
described in chapter X.87 Prices for the United
States and 12 large SMSA's are released month­
ly in Estimated Retail Food Prices by Cities.
A similar report, Retail Prices and Indexes of
Fuels and Electricity contains U.S. prices and
indexes and city prices for individual items of
fuel and utilities.
Except for food and fuel, there is no regular
publication of actual average prices. The last
publication of prices of nonfood items was in
Bulletin No. 1197, Average Retail Prices, 1955,
published in 1956.
Correction Policy
The Bureau's consumer price indexes are pub­
lished as "final" when first issued, rather than
as preliminary, as is done for many statistical
series. This is done not only to avoid complica­
tions for users who have based policy decisions
or fulfilled contractual arrangements on the
basis of published indexes, but also because so
little additional data relating to a given month
are obtained after publication. Occasionally,
however, errors of reporting and computation,
serious enough to warrant correction of previ­
ously published indexes, occur. These correc­
tions are made when the magnitude of error
reaches predetermined levels. These levels have
been set forth in an official statement.88 When
the amount of error is less than the standard
warranting correction, the level of the current
month's index with respect to the base period is
corrected. This method means, of course, that
the index does not reflect exactly the correct
w
"Calculation of Average Betail Pood Prices," ljy Doris P. Rothwell, in Monthly Labor Review, January 1965, pp. 61-66.
88
Consumer Price Index, Procedure for Correction of Consumer
Price Indexes and Prices, Aug. 1, 1958*

change from the previous month's published in­
dex, but there is no cumulative error.
Interim Extension of Old Series
Recognizing the problems that revision of the
index in 1964 posed for users having escalation
contracts based on the index, the Bureau ar­
ranged for a 6-month overlap period, January
through June 1964, during which both new
series and old series indexes were calculated and
released. It was hoped that this would provide
sufficient time for conversion of contracts to the
new series* This did not prove to be the case.
Most users continued to use the old series index
for existing contracts as long as it was available.
Moreover, the Bureau was asked to extend the
old series index after June 1964 (when pricing
of old samples was dropped) by estimation. The
Bureau agreed to furnish such an estimate by
letter on joint request of labor and management
as often and as long as needed. The estimate
was not published in any Bureau release. In




June 1964, the old series index was 0.2 index
points above the new series on a 1957-59 base
and 0.3 index points on a 1947-49 base. The
estimate was made by projecting the old series
forward by movements of the new index. The
simple arithmetical basis for this projection was
carefully spelled out for all persons requesting
this extension. Because of the method of calcula­
tion, the differentials over the published new
series indexes were expected to remain +0.2
index points on the 1957-59 base and +0.3
index points on the 1947-49 base in the foresee­
able future.
Descriptive material
The Bureau publishes official and unofficial
statements, Monthly Labor Review articles, spe­
cial reports, bulletins, and papers for technical
journals. A bibliography of historical publica­
tions dealing with the Consumer Price Index
appears on page 114.

83

APPENDIX TABLE I. CHARACTERISTICS OP WAGE EARNERS AND CLERICAL WORKERS WHOSE EXPENDITURES WERE USED IN THE
DERIVATION OP INDEX WEIGHTS

Surrey date

Period weights used in index

Average
family

Average
family
income
after taxes
$827

1901

1890-1920 >.

5.3

1917-19

1913-24...
1925-29 K.

4.8

1934-36

1925-29 »
1930-49
1950-52 "old" series

3.6

1947-49 «.
1950
1960-61..

1950-52 "adjusted" series.
1953-63
1964 forward

Survey date

3.3
«3.7

Source and amount of family income

Family composition

Two or more persons..

$1,485 I Minimum of husband, wife, and one
II child who was not a boarder or
lodger. No boarders nor more than
3 lodgers present.
$1,524 Two or more persons. Not more than
2 boarders or lodgers, or guests for
more than 26 guest weeks.

$4,160
« $6,230

Two or more persons
Families of 2 or more persons and single
workers; at least 1 full-time wage
earner.

Length of employment

Earnings of chief earner

Salaried worker earning $1,200 or less
during year. No limitation on
wage earners.
Salaried worker earning $2,000 or less.
No limitation on wage earners.
At least $300. Salaried worker earn­
ing less than $2,000 during year or
less than $200 during any month.
No upper limitation on wage
earners.
No limitation. (Family income not
in excess of $10,000.)
No limitation.

Economic level

Length of residence,
nativity and race
No limitation.
White only; in area entire year
and in the U.S., 5 years or
more; no non-English-speak­
ing families.
White only, except where
Negro population was signifi­
cant part of total; in area 9
months or more.

1901....
1917-19

No limitation
At least 75 percent from principal
earner or others who contributed all
earnings to family fund.

No limitation.
No limitation.

NoKmitation
No slum or charity f amilies-

1934-36

At least $500. Less than one-fourth
from interest, dividends, royalties,
speculative gains, rents, gifts, or
income in kind. No rent in pay­
ment of services. Less than 3
months' free rent. No subsidiary
clerical worker earning $2,000 or
over.
Family income under $10,000 after
taxes in the survey year. No mini• mum income limit, except that
families with no income from wages
or salaries were excluded.
More than half of combined family
income from wage-earner or cler­
ical-worker occupation.

At least 1,008 hours spread
over 36 weeks.

No relief families, either on
direct or work relief.

No specific requirement, but
major portion of income of
family head must be from
employment as wage earner
or salaried clerical worker.
A minimum of 37 weeks for
at least one family member.

No exclusion for receipt of
relief as such, but only
families with wage or salary
earnings included.

1947-49
1950
1960-61.

1
8

Food price index only.
Indexes between 1925 and 1929 recomputed retroactively with group
weights based on the average of 1917-19 and 1934-36 survey data.
APPENDIX TABLE II,

Year

1890-1912 »1913-34
1913-17
1918-24
1925-29—
1930-34
1935-49.
1935-39
1940-42
1943-49-1950-52—
1953-63
1964 forward—-

A
SA-Q
SA
SA
Q

-

XL::
M
M
M
M.

Census
providing
population
weights




8
7 selected cities only.
* Families of 2 or more persons; average income of single workers was
$3,560.

Survey providing expenditure weight
Group
weights

Item
weights

1901
1917-19
1917-19
Av. 1920-30.. 1917-19
Av. 1920-30-- Av. 1917-19 and 1934-36
1930
1934-36
8
1934-36
1934-36
1930
1930
1940*
«1934-36
1950
1947-49
«1950
1950
1950
1960-61
1960-61
1960
None

*A
■■ annually; SA ■» semiannually; Q=quarterly; M = monthly.
1
Food price index only.
8
During World War II, weights were adjusted to account for rationing
and
shortages.
8
51-56 cities included in the food index.
4
1940 Census data were supplemented by ration book registration data.

84

No limitation.

GENERAL DESCRIPTION OP U.S. CONSUMER PRICE INDEX COVERAGE

Frequency
of
publication*
A

No restriction other than the
wage-earner clerical-worker
definition.

No limitation.

None

Reference
base
period

Number
of cities
included

1890-99
1913

39-171
32

1936-39

8

1935-39
71947-49
1957-59

8

Number of
group and
subgroup
indexes
published
1
7

34

18

34
46
»50

18
45
48

5
Item weights were revised for only the 7 cities for which 1947-49 ex­
penditure data were available.
• Data were adjusted to 1952 for weight derivation.
* The base period was changed to 1957-59 in 1962; indexes also calculated
on8bases of 1947-49 - 1 0 0 and 1939 = 100.
Six additional cities added in 1966.

APPENDIX TABLE III.

COMPARISON OF OLD AND NEW SERIES CONSUMER PRICE INDEX

Title
Formula (Simplified expression).
Reference base period.
Duration
Population coverage:
Place of residence...
Family size.
Occupation.

Length of employment
Income
Definition of index expenditure weights.

City coverage:
Sample of priced cities..

New series index

Old series index

Item

Consumer Price Index—U.S. City Average­

Ii«Ii-

' s (Pi-*qa) ( P i / P i - i ) ~

2(pi-iqa)

J

1957-59 ■■100. Series was changed from the 1947-49
base period in January 1962, but continued to be
published on this base as well as 1939 = 100.
Discontinued after June 1964
Urban places of 2,500 or more in 1950; excluding Alaska
and Hawaii.
2 or more persons; single person consumer units ex­
cluded.
Wage-earner and clerical-worker families. (Head of
household must have been employed in wage-earner
or clerical-worker occupation.)
No specific requirement, but major portion of income of
family head must have been from employment as
wage earner or clerical worker.
Family income under $10,000 after taxes in 1950. No
lower income limit, except that families without in­
come from wages or salaries were excluded.
Average family expenditures for urban wage earners and
clerical workers derived from the 1950 Consumer
Expenditure Survey in 91 urban places, adjusted for
changes in prices and income between 1950 and 1952.

46 urbanized areas, selected to represent urban places in
the U.S. having populations of 2,500 or more in 1950,
excluding Alaska and Hawaii.

Pricing cycle -

Prices of foods, fuels and a few other items collected
monthly in all cities.

Population weights-

Prices of most other commodities and services collected
monthly in the 5 largest cities, and quarterly in
remaining cities.
Based on 1950 Population Census; Alaska and Hawaii
excluded. Proportion of population in wage-earner
and clerical-worker group covered by index was based
upon BLS expenditure surveys.
U.S. and 20 cities, for families only

Published indexes..

Item sample:
Commodity coverage

Goods and services purchased for family living, in­
cluding necessities and luxuries; excluding personal
insurance, income and personal property taxes, but
including real estate taxes and sales and excise taxes.
About 325, priced in all cities

Number of items priced
Most important items in family spending
Basis of item sample selection

Direct allocation of unpriced to priced items based on
expected similarity of price movements.

Basis for allocation to priced items.
Reporter samples:
Location
Number of reporters

Within boundaries of central cities of 46 urban areas...

Number of quotations obtained.
Pricing techniques




Consumer Price Index—U.S. City Average for Urban
Wage Earners and Clerical Workers.
Same.

About 1,500 food stores
30,000 tenants
5,500 other reporters of all kinds
About 1 million food prices per year
About 60,000 rent charges per year
About 230,000 quotations per year for items other than
food and rent.
Personal visit of BLS agent except for a few items col­
lected by mail or from secondary sources.
Specification pricing; same quality priced in all stores
in a city.

1957-59 = 100. Series also published on 1947-49 and
1939 bases.
January 1964 forward.
Urban places of 2,500 or more in 1960; including Alaska
and Hawaii.
No restriction; single consumer units included.
Wage-earner and clerical-worker families and single
individuals living alone. (More than half of total
family income from wage-earner or clerical-worker
occupations.)
At least 1 family member must have been employed for
37 weeks or more during the survey year in wageearner or clerical-worker occupations.
No criterion as to family income except the qualification
above.
Average expenditures for urban wage earners and
clerical consumers (including single workers) derived
from the 1960-61 Consumer Expenditure Survey in
66 urban places, adjusted for price changes between
the survey dates and December 1963 except for 6
cities added in 1966.
50 metropolitan areas and cities selected originally to
represent all urban places in the U.S. including
Alaska and Hawaii, with populations of 2,500 or more
in 1960. Six additional areas added in 1966.
Same.

Based on 1960 Population Census; Alaska and Hawaii
included. Proportion of population in wage-earner
and clerical-worker group covered by index was based
upon BLS expenditure surveys.
U.S. and 17 large metropolitan areas for families and
single consumer units combined. Indexes for 6
more large metropolitan areas available in the latter
part of 1965.
Same.

About 400 represented in U.S. index and in published
city indexes. Certainty items priced in all unpublished
cities; other items in 1 or 2 subsamples of other
unpublished cities.
Probability proportionate to importance in family
spending.
Expenditures classified into 52 expenditure classes*
Certainty items assigned their own importance; re­
mainder of expenditures assigned equally to proba­
bility selections within expenditure classes.
In central cities and selected suburbs of 56 metropolitan
areas (50 areas in 1964 and 1965).
About 1,775 food stores (1,525 for 50 areas).
40,000 tenants (34,000 for 50 areas).
16,000 other reporters of all kinds (15,000 for 50 areas).
Over 1 million food prices per year.
About 80,000 rent charges per year (68,000 for 50 areas).
About 375,000 quotations per year for items other than
food and rent (350,000 for 50 areas).
Same.
Specification pricing but agent is permitted to price
deviations from specification under prescribed condi­
tions.

85

APPENDIX TABLE IV.

SUMMARY OF SAMPLES FROM COMPREHENSIVE URBAN HOUSING UNIT SURVEY AND URBAN SURVEY OF

CONSUMER EXPENDITURES, 1960-61

Standard Metropolitan Statistical Areas (SMSA) or cities

U.S. urban total
Areas Having CHUS Prior to CES
Stratum A-SMS A 1,400,000 & over
Baltimore, Md ' 1
Boston, Mass
Chicago-Northwestern Indiana *-Cleveland, Ohio
Detroit, Mich .
Los Angeles—Long Beach, Calif
/New York, N.Y. *
{Northeastern, N.J.*
Philadelphia, Pa
Pittsburgh, Pa
.
St. Louis, Mo
San Francisco-Oakland, Calif.
Washington, D.C
.'
Stratum B-SMSA 250,000-1,399,999
Atlanta, Ga
——----Buffalo, N.Y
Dallas, Tex—
Seattle, Wash

CHUS
year

„,.,

-

1960

-

__

__
-__--.--_

-r,,,,. _.

_-

.-




4,860
4,220

6,127
3,459
4,087
4,554
8,078
5,041
3,786
1,256
7,039
5,031
4,445
4,280
3,627

375
375
500
375
375
500
625
500
375
375
375
375
375

313
268
371
294
290
388
448
356
313
323
319
302
323

192
132
219
156
161
182
242
189
144
165
171
152
142

1961
—do
„.do
—do
—do
...do

4,453
3,481
3,060
3,698
4,079
3,522

250
250
250
250
250
250

180
204
175
215
201
189

91
98
90
117
107
79

1960
—do
—do
—do
__.do

1960
—do
—do
—do
...do

1,510
2,557
1,253
3,472
1,567

160
160
160
160
160

110
125
126
106
135

49
58
47
44
81

1961
—do
—do
—do
...do

1961
—do
—do
—do
...do

2,459
1,659
1,245
. 1,843
2,063

160
160
160
160
160

120
112
135
130
151

51
49
71
77
85

1961
do
do
—do
—do
—do
—do
—do
1959

1961
—do
—do
—do
—do
—do
—do
—do
1959

646
782
769
780
419
951
1,545
342
1,744

65
65
65
65
65
65
65
65
275

61
54
50
50
55
56
54
53
134

23
24
28
12
29
29
26
30
67

1961
—do
do
—do
—do
—do
—do
...do

1960
—do
—do
—do
—do
—do
—do
—.do

5,281
557
685
646
713
711
734
648
587

* 1,560
65
65
65
65
65
65
65
65

1,230
49
55
47
44
38
61
60
55

640
23
36
24
29
20
33
29
21

65
65
65
65
65
65
65
65

52
43
58
61
55
41
48
64

25
20
28
41
32
25
25
41

65
65
65
65
65
65
65
65

60
43
42
56
45
58
48
57

37
17
17
26
20
40
18
13

_--

------------_
-

Including low income public housing obtained from central sources.
* Including 5,281 addresses in CHUS in 8 D stratum cities surveyed in
1961 and excluding 16 D stratum cities not having CHUS, for all of which
CES samples were selected from Census records.

86

9,476
8,246

1961
—do
—do
—do
—do
___do

Vicksburg, Miss..
-. D Stratum Cities Not Having CHUS (Not in CPI but Used for CPI
Weights) (CES Sample from Census)
Burlington, Vt - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - Cleveland, Tenn
«.,.»-,„■•.,._,--*■ -,^„™™ ^. r „ ^rGallup, N.Mex
...
Griffin, Ga
LaSalle, 111
Lewistown, Pa
Owatonna, M i n n . - - - - - -__.__-- - - - - - Reserve, La —----___----------------

1

12,205
10,645

104
123
97
94
93

-

--

2 137,198
131,917

198
199
178
173
209

Anchorage, Alaska
-------- D Stratum Cities Having CHUS Subsequent to CES (CES Sample

Athol, M a s s - . - . . .
_--.
Cambridge, OhioEureka, Calif
Gainsville, Tex -Manhattan, K a n s - - . - . - . . - — - - - - - - - - - - - Menasha, W i s . — - - - ----------Okmulgee, Okla_
-_---Sebring, Fla

Usable
schedules

250
250
250
250
250

---

Findlay, Ohio

Assignment
addresses

Number of
wage-earner
clericalworker con­
sumer units

4,043
4,091
3,437
4,406
5,231

-

-

Number of consumer
units (CES)

1960
—do
—do
—do
...do

.......-,-*,,-.

Portland, Maine

Number of
separate
living
quarters
(CHUS) *

1960
—do
— —do
—do
...do

Bakersfield, Calif

Lancaster, Pa
Stratum D-Urban Places 2,500-49,999
Crookston, Minn
Florence, A l a . - .
-_Logansport, Ind

1960-61

—

Dayton, Ohio

Wichita, Kansas
r,Stratum C-SMSA 50,000-249,999
Austin, Tex

CES
year

(5)
(5)

(5)
(5)
(5)
(55)
(5 )

()
(55)
( 5)
(5)
( 5)
( 5)
()
(65)
()

1960
—do
—do
—do
—do
—do
—do
—do
1961
—do
—do
—do
—do
—do
—do
—do

(5)
(5)
(S)
(5)
(55)
( 5)
(6 )

()
( 65)
(5)
(6 )
( 5)
(5)
(6)
( 5)
()

*4 Standard Consolidated Area.
CES samples in the 8 CPI D cities surveyed in 1960 were selected
from
Census records rather than from CHUS.
6
Not available.

APPENDIX TABLE V.

NUMBER OF ITEMS IN SAMPLING FRAME AND NUMBER OP ITEMS PRICED BY EXPENDITURE CLASS

Expend­
iture
class
number
All items

.

.

.

Food
EC1
EC 2
EC 3

EC 4
EC 5
EC 6
EC 7
EC 8
EC 9
E C 10
ECU
EC 12
EC 13
E C 14
E C 15

j

Food at home:
Cereals and bakery products:
Cereals and gram products
Bakery products
Meats, poultry, and fish:
Meats:
Beef and veal
.
Pork
Other meats
Poultry
Fish
Dairy products:
Dairy products . .
Fruits and vegetables:
Fresh fruits

.

.

.

Processed fruits and vegetables
Other food at home:
Eggs
Fats and oils
....
Sugar and sweets
Nonalcoholic beverages
Prepared and partially prepared foods
Food away from home

.
-.
-

.

-

...

Housing
EC 16
EC 17
EC 18
E C 19
E C 20
E C 21
EC 22
EC
EC
EC
EC

23
24
25
26

EC 27
EC 28

Shelter:
Rent
Homeownership:
Purchase ana financing . . .
.
Taxes and insurance. T
Maintenance and repairs:
Commodities...
Services
.
Fuel and utilities
Household furnishings and operation:
Housefurnishings:

.

Textile hotisefnrnishiTigs. - -. - ^

-^ ™ -

Furniture and floor coverings:
Furniture
Moor coverings
.
Appliances
Other housefurnishings
Household operation:
Housekeeping supplies
Housekeeping services

„

-

EC 31
EC 32
E C 33
EC 34
EC 35

.

.

EC 36
EC 37
EC 38
EC 39
EC 40
EC 41

.

_

_

Health and recreation
EC 42
EC 43
E C 44
EC 45
EC 46
EC 47
EC 48
EC 49
EC 50
EC 51
EC 52

Medical care:
Drugs and prescriptions
Professional services
Hospital services and health insurance
Personal care:
Toilet goods
Services
Reading and recreation:
Recreation:
Recreational goods
Recreational services
Reading and education
Other goods and services:
Tobacco products
Alcoholic beverages
.
Financial and miscellaneous personal expenses




...
„

.

-

396

93

105

19
16

4
5

4
5

12
12
14
4
6

7
6
6
3
4

9
6
6
3
5

19

6

7

15
20
48

8
11
10

8
11
10

1
8
12
8
50
3

1
3
4
5
8
2

1
3
4
6
8
9

212

73

81

4

2

2

3
2

2
2

2
3

14
30
6

6
5
6

6
5
10

6

6
11
4
8
8

28
14

8
8

184

8
7
64

77

30
23

12
4

15
4

....

39
31
21

19
8
9

26
9
11

..__

26
14

6
6

6
6

34

21

34

2
2
6

2
2
2

12
3
2

13
5 i
6

6
4

6
6
5

115 i

58

99

2
11
2

29
2

20
12
6

28
9

8
4

8
4

29
13
11

13
6

20
7
7

2
3

5
7
3

Transportation
Private:
Autos and related goods:
Auto purchase
Gasoline and motor oil
Auto parts, etc.
Automobile services:
Auto repairs and maintenance
Other automobile expenses.,
Public

309

267

10
3
8
8

.

-__.

812

20

Men's and boys' apparel:
Boys' apparel
Women's and girls' apparel:
Women's apparel
Girls' apparel
._
Footwear
Other apparel:
Commodities
Services

Number of
specifications
priced

31
7
21
32

,

m

Apparel and upkeep
EC 29
E C 30

Number of
items sampled

Number of
items

Class

.

_

3
3
4

!
1

6

3

87

APPENDIX TABLE VI.

EC No.

LIST OP ITEMS PBICED FOR THE REVISED CONSUMER PRICE INDEX AS OP DECEMBER 1963
Priced items

Groups, subgroups, expenditure classes
Sample A

Sample B

Food
Food at home
Cereals and bakery products
Cereals and grain products

EC-l

Bice, long and short grain
EC-2

Bakery products

Meats, poultry, and fish
Meats
Beef and veal

EC-3
3A

3B

Pork

3C

Other meats

EC-4

Poultry

EC-5

Fish

EC-6

Dairy products

EC-7

Fruits and vegetables
Fresh fruits

EC-8

Fresh vegetables

EC-9

Processed fruits and vegetables

Other food at home
Eggs

EC-10
EC-11

|

Fats and oils

EC-12

Sugar and sweets

EC-13

Nonalcoholic beverages

EO-14

Prepared and partially prepared foods

See footnotes at end of table.

88



White bread
Whole wheat bread
Layer cake, p l a i n . . . .

Flour, white, all-purpose.
Cracker meal.

...

White bread.
.
Cookies, cream filled.
. . . . . . . . . . J Cinnamon rolls, frosted.

Hamburger, preground..
................|
Steaks, round, bone-in..
.--..---.-.....-Steaks, porterhouse, b o n e - i n . . . . . . . . . . . . . . . J
Rump roasts, s t a n d i n g . . . . . . . . . . . . . . . . . . . . .
Chuck roasts, bone-in. . . . . . . „ . — . . . . . . .
Veal cutlets, b o n e - i n . . . . . . .
.........

Hamburger, preground.
Steaks, round, bone-in.
Steaks, sirloin, bone-in.
Rump roasts, standing.
Bib roasts, bone-in.
Beef liver, sliced.

Pork chops, center c u t . . . . . . . . . . . . . . . . .
Bacon, sliced
.
.
Pork roasts, loin halves
.
Picnics, smoked
.-................

Pork chops, center cut.
Bacon, sliced.
Pork sausage.
Ham, whole.

Lamb chops, loin. . _ . . . „ . . . . . . . . - . _ . . - - . - Bologna sausage, sliced*
Liverwurst sausage, sliced or whole.
Salami sausage, sliced... . . . . .
-Frankfurters, s k i n l e s s . . . . . . .
. . . . . . . . . . . . Ham, canned, domestic or imported.
Frying chickens, r e a d y - t o - c o o k . . . . . . . . . . . . . . Frying chickens, ready-to-cook.
Turkey, fresh or frozen.
Chicken breasts, f r e s h . . . . . . . . . . . . . . . .
Fillets or steaks, fresh or frozen * . . — . . _ . . . . . Shrimp, raw, frozen.
Tuna fish, chunk: s t y l e . . . . . . . . . - . . . . - . Sardines, Maine.
Milk, fresh, g r o c e r y . . . . . . . . . . . . . . . . . . . . . . . .
Milk, fresh, delivered
....
.—...-..
Milk, fresh, s k i m . . . . . . . . . . . . . . . . . . . . . . . . . .
Ice cream, prepackaged
.....
-...-.-.Butter, s a l t e d . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Milk, fresh, grocery.
Milk, fresh, delivered.
Milki evaporated, canned.
Cheese, American process.
Butter, salted.

Apples, all p u r p o s e . . . . . . . . . . . . . . . . .
.Bananas, yellow variety..
.
...... Oranges, except Temple or King
...
Grapes, Thompson s e e d l e s s . . . . . . . . . . . . . . . . .
Grapefruit, fresh, pink or w h i t e . . . . . . . . . . . . .
Orange juice, f r e s h . . . . . . . . . . . . . . . . . . . . . . . . .

Apples, all purpose.
Bananas, yellow variety.
Oranges, except Temple or King.
Grapes, Thompson seedless.
Strawberries, fresh.
Watermelons, whole or sliced.

Head l e t t u c e . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Head lettuce.
Potatoes, white.
Tomatoes
...
.-...-.-.--...
. . . Tomatoes.
Asparagus, g r e e n . . . . . . . . . . . . . . . . . . . . . . . . . . Cabbage, all varieties except red.
Celery, Pascal, stalk.
Carrots, topped, prepackaged
.. .....
Onions, yellow.
Spinach, p r e p a c k a g e d . . . . . . . . . . . . . . . . . . . . . . Peppers, sweet green.
Pears, Bartlett, can or jar
.. .
.....
Lemonade, concentrate, f r o z e n . . . . . . . . . . . . . .
Beets, sliced, can or j a r . . . . . . . . . .
....
Tomatoes, can or j a r . . . . . . . . . . . . . . . . . . . . . . .

Eggs, fresh, large, Grade A

...

Fruit cocktail, canned.
Pineapple-Grapefruit juice drink, canned.
Orange juice concentrate, frozen.
Peas, green, can or jar.
Broccoli spears, frozen.

. Eggs, fresh, large, Grade A.

Margarine, c o l o r e d . . . . . . . . . . . . . . . . . .
. . . . Margarine, colored.
Salad dressing, I t a l i a n . . . . . .
. ...
Salad or cooking oil, vegetable.
Sugar, white, granulated
. . . . . . . - . . — _ - . . Grape jelly, pure.
Chocolate bar, plain m i l k . . . . . . . . . . . . . . . . . . Chocolate flavored syrup.
Coffee, can or bag.
Coffee, can or bag
._.-........—-.—-Carbonated drinks, fruit-flavored. . . . . . . . . . Coffee, instant.
Cola drink, carbonated.
Bean soup, canned, c o n d e n s e d . . . . . . . . . . . . . . .
Spaghetti, in tomato sauce, c a n n e d . . . . .
.
Mashed potatoes, instant.
.
.
Potatoes, French fried, f r o z e n . . . . . . . . . . . . .

Chicken soup, canned, condensed.
Baby foods, strained.
Sweet pickle relish.
Pretzels, hard, salted.

APPENDIX TABLE VI.

EC No.

LIST OP ITEMS PRICED FOR THE REVISED CONSUMER PRICE INDEX AS OF DECEMBER 1963—Continued
Priced items

Groups, subgroups, expenditure classes
Sample A

EC-15

EC-16

EC-17

Food away from home

Housing
Shelter
Rent
Homeownership
Home purchase and financing

EC-18

Taxes and insurance

EC-19

Maintenance and repairs
Commodities

EC-20

EC-21

EC-22

Services

Fuel and utilities

Household furnishings and operation
Textile housefurnishings

EC-23

Furniture

EO-24

Floor coverings

EC-25

Appliances

EC-26

Other housefurnishings

EC-27

Housekeeping supplies

EC-28

Housekeeping services

See footnotes at end of table.




Restaurant meals
Lunch
Breakfast
Between meal snacks
Coffee, cup
Carbonated beverages, cup
Frankfurter on roll
Ice cream, dish

Sample B

Restaurant meals.
Lunch.
Dinner.
Between meal snacks.
Coffee, cup.
Carbonated beverages, cup.
Pie, slice.
Candy bar.

.

Rent of house or apartment,
Hotel, motel room rates

! Rent of house or apartment.
Hotel, motel room rates.

Home purchase
Mortgage interest rates

| Home purchase.
Mortgage interest rates.

Property taxes, residential
Property insurance rates
Fire and extended coverage
Comprehensive homeownership policy

. Property taxes, residential.
Property insurance rates.
Fire and extended coverage.
Comprehensive homeownership policy.

Exterior house paint
Furnace air filters
Packaged dry cement mix

Interior house paint.
Shelving, Ponderosa pine.
Shrubbery, evergreen.

Residing houses
ReshingHng roofs
Replacing sinks

Residing houses.
Repainting living and dining rooms.
Repairing furnaces.

Fuel oil and coal:
Fuel oil, #2.
Coal, anthracite or bituminous
Gas and electricity:
Gas, 3 bills per city
Electricity, 3 bills per city
Other utilities:
Residential telephone services
Residential water and sewerage services

Fuel oil, #2.
Coal, anthracite or bituminous.
Gas, 3 bills per city.
Electricity, 3 bills per city.
Residential telephone services.
Residential water and sewerage services.

Pillows, bed, polyester or acrylic filling..
Curtains, tailored, polyester marquisette.
Drapery fabric, cotton or rayon/acetate.

Sheets, percale or muslin.
Bedspreads, chiefly cotton, tufted.
Slipcovers, ready made, chiefly cotton.

Bedroom suites, good or inexpensive quality. _
Living room suites, good and inexpensive]
quality
Lounge chairs, upholstered
Sofas, dual purpose
Sleep sets, Hollywood bed type
Aluminum folding chairs

Bedroom suites, good or inexpensive quality.
Living room suites, good and inexpensive
quality.
Dining room suites.
Sofas, standard, upholstered.
Box springs.
Cribs.

Rugs, soft surface
Broadloom, wool
Broadloom, nylon
Rugs, hard surface

Rugs, soft surface.
Broadloom, wool.
Broadloom, nylon.
Tile, vinyl.
Refrigerators or refrigerator-freezers, electric.
Washing machines, electric, automatic.
Vacuum cleaners, cannister type.
Air conditioners, demountable.
Garbage disposal units.

Refrigerators or refrigerator-freezers, electric.
Washing machines, electric, automatic
Ranges, free standing, gas or electric
Clothes dryers, electric, automatic
Room heaters, electric, portable

Dinnerware, earthenware
Flatware, stainless steel.
Carpet sweepers, manually operated
Table lamps, with shade.
Venetian blinds, white, steel or aluminum] Lawn mowers, power, rotary type.
slats
—™.
Nails, 8d (penny) common.
Electric drills, hand held
Detergent, liquid, laundry
Laundry soap for fine fabrics
Scouring pads, steel wool
Toilet tissue

Detergent, granules or powder.
Air deodorizers, spray ty;
Paper napkins, emboss<
Stationery, envelopes.

Domestic service, general housework
Baby sitter service
Postal services
Laundry flatwork, finished service
Licensed day care service, pre-school chUd..
"Washing machine repairs

Domestic service, general housework.
Babysitter service.
Postal services.
Laundry flatwork, finished service.
Reupholstering furniture.
Moving expenses.

APPENDIX TABLE VI.

EC No.

LIST OP ITEMS PRICED FOB THE REVISED CONSUMER PRICE INDEX AS OP DECEMBER 1963—Continued
Priced items

Groups, subgroups, expenditure classes
Sample A

EC-29

EO30
EC-31

EC-32

EC-33

EC-34
EC-35

EC-36

Apparel and upkeep
Men's and boys' apparel
Men's apparel

Boys' apparel
Women's and girls' apparel
Women's apparel

Girls' apparel

Footwear

Other apparel
Commodities
Services

Transportation
Private
Auto purchase

EC-37

Gasoline and motor oil

EC-38

Auto parts

EC-39

Automobile services
Auto repairs and maintenance

EC-40

Other automobile expenses

See footnotes at end of table.

90



Sample B

Suits, year round weight, 2 q u a l i t i e s * . . . . . . . .
Topcoats, wool
..
... . ...
Suits, tropical weight
........
...
Slacks, wool or wool b l e n d . . . . . . . . . . . . . .
Shirts, work, c o t t o n . . . . . .
..
.-......-!
Shirts, sport, cotton, short s l e e v e s . . . . . . . . . . .
Shirts, sport, cotton, long sleeves...
----T-shirt

Suits, year round weight, 2 qualities.
Jackets, lightweight.
Trousers, work, cotton.
Slacks, cotton or manmade blend.
Shirts, business, cotton.
Socks, cotton.
Handkerchiefs, cotton.

Coats, all purpose, cotton or cotton blend
Sport coats, wool or wool blend,
Dungarees, cotton or cotton blend. . . . . . . . . i under shorts, cotton.
Coats, heavyweight, wool or wool blend, 2 j Coats, heavyweight, wool or wool blend, 2
qualities.
qualities.
Carcoats, heavyweight, c o t t o n . . . . . . . . . . . . . . Coats, lightweight, topper.
Skirts, wool or wool blend
.
.
Sweaters, wool or acrylic.
Dresses, cocktail, street length.
Dresses, street, chiefly manmade fiber, 2 Dresses, street, chiefly manmade fiber, 2
qualities.
qualities.
Dresses, street, wool or wool blend.
Dresses, street, c o t t o n . . . . . . . . . . . . . . . . . . . . . . Dresses, street, cotton.
Housedresses, cotton.. . . . . . . .
...
— Blouses, cotton.
Slacks, lightweight, cotton and corded Bathing suits, 1 piece.
cotton.
Slips, nylon... . . . . _ . . . . . _ . . . . . - .
. . . . . Girdles, manmade blend.
Brassieres, c o t t o n . — . . . . . . . . . . . . . . . . . . . . . . . Panties, acetate.
Hose, nylon, full fashioned and seamless, 2 Hose, nylon, full-fashioned and seamless, 2
styles.
styles.
Anklets, c o t t o n . . . . . . . . . . . . . . . . . . . . . . . . . . . Gloves, fabric, nylon or cotton.
Handbags, rayon faille or plastic.
Coats, lightweight, topper.
Skirts, wool or wool b l e n d . . . . . . . . . . . . . . . J Slacks, cotton.
Slips, cotton b l e n d . . . . . . . . . . . . . . . . . . . . . . . . . Shorts, cotton.
Handbags, plastic
-- . . . . . — — . .
J Dresses, cotton.
Robes, duster style, quilted tricot, or percale.
Men's:
Shoes, street, oxford, 2 q u a l i t i e s . . . . . . . . . . . Shoes, street oxford, 2 qualities.
Shoes, work, high.
Women's:
Shoes, street, pump, 2 s t y l e s . . . . . . . . . . . . . . Shoes, street, pump, 2 styles.
Shoes, evening, p u m p . . . . . . . . . . . . . . . . . . . . Shoes, evening, pump.
Shoes,' casuairpump.*
Houseslippers, scuff.
Children's:
Sneakers, boys', oxford t y p e . . . . • - . . . . . . Shoes, oxford.
Dress shoes, girls', strap.
Diapers, cotton g a u z e . . . . . . . . . . . . . . . . . . . . . . Wrist watches, men's, imported movement.
Yard goods, c o t t o n . . . . . . . . . . .
. . . . . . . . . Wrist watches, women's, imported movement.
Zipper, skirt or neck placket.
Earrings, Pearl, simulated or imitation
Dry cleaning, men's suits and women's dress­ Dry cleaning, men's suits and women's dress
es.
es.
Shoe repairs, women's heel l i f t . . . » . . . . . » . - - . Automatic laundry service.
Tailoring charges, hem adjustment.
Laundry, men's s h i r t s . . . . . . . . . .
.....
New cars:

Chevrolet, Impala, 2-door hardtop.
Ford, Falcon, Futura, 4-door sedan.
Chevrolet, Chevelle, 2-door hardtop
Ford, Galaxie 500, 2-door hardtop
. . . . . . Ford, Galaxie 600, 2-door hardtop.
Plymouth, Fury III, 4-door s e d a n . . . . . . . . . Pontiac, Catalina, 4-door sedan.
Volkswagen, Deluxe, 2-door hardtop.
Used cars:
2-years old, Chevrolet and Ford.
2-years old, Chevrolet and Ford
3 years old,
Do.
3 years old, . . . do
.
....
.
4 years old,
Do.
4 years old,
do
.
5 years old,
Do.
Gasoline, regular and premium.
Motor oil, premium.
Tires, tubeless, r e t r e a d . . . . . . .

....

. . . . . . Tires, tubeless, new.

Water pump replacement.
Motor-tune-up.. . . . . . . . . . . . . . . . . . . . . . . . . . . Replacing muffler.
Front end alignment.
Automatic transmission repairAuto insurance rates, liability and physical
damage.
1 Auto financing charges * . . . . . . . . . . .
..
Auto registration and inspection fees
Driver's license f e e s . . * . . . . . . . . . . . . . . . . . . . . . .
Parking fees, private and municipal......

Auto insurance rates, liability and physical
damage.
Auto financing charges.*
Auto registration and inspection fees.
Driver's license fees.
Parking fees, private and municipal.

APPENDIX TABLE VI.

EC No.

LIST OF ITEMS PRICED FOR THE REVISED CONSUMER PRICE INDEX AS OF DECEMBER 1963—Continued
Priced items

Groups, subgroups, expenditure classes
Sample A

EC-41

EC-42

Public transportation

Health and recreation
Medical care
Drugs and prescriptions

Sample B

Local transit f a r e s . . . . . . . . .
.............
Taxicah fares
.
.
Railroad fares, coach...
..... ......
Airplane fares, chiefly coach.
—... . . . . .
Bus fares, i n t e r c i t y . . . . . .
.-..---... .
Over-the-counter items:
Multiple vitamin c o n c e n t r a t e s . . . . . . . . . . . .
Liquiu tonics
......................
Cold tablets or c a p s u l e s . . . . . . . . . . . . . . . .
Prescriptions:
Anti-infectives:
Penicillin G buffered t a b l e t s . . . . . . . . . . . .
Sulfisoxazole tablets.
Sedatives and hynotics:
Phenobarbital t a b l e t s . . . . . . . . . . . . . . . . . .
Ataractics:
Chlordiazepoxide-hydrochloride capsules. .
Antispasmodics:
Cardiovasculars and antihypertensives:
Antiarthritics:

EC-43

EC-44

Professional services

Hospital services and health insurance
Hospital services

Health insurance *

EC-45

EC-46

EC-47

EC-48

Personal care
Toilet goods

Personal care services
Reading and recreation
Recreation
Recreational goods

Recreational services

Cough preparations:
Elixir terpin hydrate with codeine

Local transit fares.
Taxicab fares.
Railroad fares, coach.
Airplane fares, chiefly coach.
Bus fares, intercity.

Aspirin compounds.
Cough syrups.
Adhesive bandages, package.
Tetracycline capsules.
Secobarbital sodium capsules.
Meprobamate tablets.
Phenobarbital and belladonna extract.
Crystalline digitoxin tablets.
Chlorothiazide tablets.
Prednisone, tablets.

Family doctor, office v i s i t s . . . . . . . . . . . . . . . .
Family doctor, house v i s i t s . . . . . . . . . . . . . . .
Pediatric care, office v i s i t s . . . . . . . . . . . . . . . .
Psychiatrists, office v i s i t s . . . . . . . . . . . . . . . . .
Routine laboratory t e s t s . . . . . . . - - - - - - - - Examination, prescriptions and dispensing
of eyeglasses.
Dentures, full u p p e r . . . . .

......

Family
visits.
Family doctor,
doctor, house
office visits.
Obstetrical cases.
Chiropractors and podiatrists, office visits.
Herniorrhaphy, adult.
Examination, prescriptions and dispensing of
eyeglasses.
Fillings, adult, amalgam, one surface.
. . . . . . Extractions, adult.

Daily service charges:
Daily service charges:
Semi private r o o m s . . . . . . . . . . . . . . . . . . . . .
Semiprivate rooms.
Private room
—......-.-. . . . . .
Private room.
Hospital services:
Daily service charges,
room.

semiprivate

Hospital services:
Daily service charges, semiprivate room.

Daily service charges, private room.
Operating room... . . . . . . . . . . . . . . . . . . .
X-ray diagnostic series, upper G.I.
Nonhosp'tal services:
Nonhospital services:
Family doctor, office visit.
Family doctor, office v i s i t . . . . . . . . . . . . . .
Surgeon's fees, tonsillectomy/ adenoidecSurgeon's fees, herniorrhaphy, adult.
tomy.
Obstetrical cases.
Prescriptions and drugs
-.-----..—.Prescriptions and drugs.
Retained earnings (overhead)
. . . . . . . . . . . Retained earnings (overhead).

Toothpaste, standard d e n t r i f i c e . . . . . . . . . . . Toilet soap, hand milled.
Hand lotions, l i q u i d . . . . . . . . . . . . . . . . . . . . .
Face powder, p r e s s e d . . . . . . . . . . . . . . . . . . . .
Cleansing t i s s u e s . . . . . . . . . . . . . . . . . . . . . . . . Home permanent refills.
Men's h a i r c u t s . . . - . . . . . . . - - - - Women's h a i r c u t s . . . . . . . . . . . .

. . . . . . . Men's haircuts.
Shampoo and wave sets, plain.
Permanent waves, cold.

.......

Radios, portable and table models, AM
band only.
TV replacement t u b e s . . . . . . . . . . . . . . . . . . . .
Sports equipment:
Golf balls, liquid center
.............
Outboard motors
.
Tricycles
. ..... . . .
Dolls
Stuffed animal- . . . . . . . . . . . . . . . . . . . . . . . . Dog food, canned and boxed.
Indoor movie admissions:
Adult
Children's
.
TV repairs, picture tube replacement
Bowling fees, e v e n i n g . . . . . . . . . . . . . .
Golf green f e e s — . - . - - . - . - - _ - - . - . . . — - - _ .

TV sets, portable and console.
Radios, portable and table models, AM band
only.
Tape recorders, portable.
Sports equipment:
Fishing rods, fresh water spincast.
Bowling baits.
Phonograph records, stereophonic.
Bicycles, boys', 26".
Movie cameras. 8-mm, fully automatic lens.
Film, 35-mm, color.
Indoor movie admission:
Adult.
Children's.
Drive-in movie admissions, adult.
Bowling fees, evening.
Film developing, black and white.

See footnotes at end of table.




91

APPENDIX TABLE VI.

EC No.

LIST OP ITEMS PBICED FOR THE REVISED CONSUMER PRICE INDEX AS OP DECEMBER 1963—Continued
Priced items

Groups, subgroups, expenditure classes

Sample B

Sample A
EC-49

Reading and education

EC-50

Other goods and services
Tobacco products

EC-51

Alcoholic beverages

EC-52

Financial and miscellaneous personal ex­
penses

Newspapers, street pale and delivery
College tuition and fees, undergraduate
Magazines, single copy and subscription....
College textbooks, undergraduate..........

Newspapers, street sale and delivery.
College tuition and fees, undergraduate.
Paperback books, not school or technical.
Piano lessons, beginner.

Cigarettes, nonfilter tip, regular size. pack.. Cigarettes, nonfilter tip, regular size, carton.
Cigarettes, filter tip, king size, pack.
Cigars, domestic, regular s i z e . . . . . . . . . . . . . Cigars, domestic, regular size.
Beer, at home, local and national brands
Whiskey, spirit blended and straight
bourbon.
Wine, dessert and t a b l e . . . . . . . . . . . . . . . . . .
Beer, away from h o m e . . . . . . . . . . . . . . . . . . .

Beer, at home, local and national brands.
Whiskey, spirit blended and straight bourbon.

Funeral services, adult
........
Bank service charges, checking account

Funeral services, adult.
Legal services, short form will.

1
Two of the largest volume sellers among the following types offishare
priced within each city, since within any given city, all varieties offishare
not available: Frozen ocean perch and haddock; fresh cod, catfish, king
salmon, halibut, sole, and haddock.

Wine, dessert and table.
Beer, away from home.

* Not actually priced; imputed from priced items.
* Four items are priced only for health insurance: Operating room,
X-ray, tonsillectomy, and retained earnings; prices for the remaining items
are also included as directly priced professional and hospital services.

APPENDIX TABLE VII. REVISED CPI WEIGHT DERIVATION PROCEDURES FOR MAJOR EXPENDITURE CATEGORIES, INDIVIDUAL
EXPENDITURE CLASS TOTALS, CERTAINTY ITEMS, AND PROBABILITY ITEM TOTALS l WITHIN EXPENDITURE CLASSES, BY
REGION 2 AND CITY-SIZE STRATUM, FAMILIES OP TWO OR MORE PERSONS, 1960-61 CES
Expenditure data code, by region-city size stratum
EC No.

1-14
1
2
3
3a
3b
3c
4
5
6
7
8
9
10
11
12
13
14
15

16,18-20
16

17

18

Item
No.

03
01
03
13
18
01
01
19
01
02
07
09
11
14
01
02
02
01
02
03
01
02
03
01
02
03

01
02
19-20
19
Seefo otnotes

Description

Food at h o m e . . . . . . . . . . . . . . . . . . . .
. ..
Cereal and grain p r o d u c t s . . . . . . . .
.
_. . . . .
Bakery p r o d u c t s . . . . . . . . . . . . — . _ . - . - . . . .
........
White bread
Meats:
Beef and veal..
....
..
Steak
Ground beef
.
....... .
....
Pork
.
Pork chops
.
...
.
........
' Bacon
...
. ............
...
._
Other meats
.
Poultry . . . . . . . .
_..__.._.._._._....
.....
Frying c h i c k e n s . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Fish
Dairy p r o d u c t s . . . . . . . . . . . . . . . . . . . . . _ - . . . . . . - . . . . . .
Fresh m i l k . . . . . .
...-_.... .
.
Buttci
—
Fresh fruits....
.
......
.
..
Apples. . . . . . .
.
.... ......
....
._
Bananas....
.
..
...
.
Oranges..
..
...
.....
..........
Fresh vegetables
....
.
..........
Lettuce and g r e e n s . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Potatoes, white
...... .....
...... ..
Tomatoes.
Processed fruits and vegetables . . . . . . .
.
..
Eggs
.
. .
Fats and o i l s . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Margarine..........-...—-.--.-...
Sugar and s w e e t s . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Nonalcoholic beverages
... .
.
.
Coffee...
Prepared and nartiillv prepared foods.
.
Food away from h o m e . . . . . . . . . . . . . . . . . .
.....
Board *'
Restaurant m e a l s . . . .
Snacks.....
..................................
Shelter (less home purchase and financing)
Rent
Rent of houpe or apartment...
...
. .....
Rent of rooms * . . _ . . . . . . . . . . .
.......
...
Hotel, motel rentals... . . . . . . . . . . .
.
Home purchase and financing * _.
.
.
Purchase................................-..--._
Settlement charges * . - . . . . . - .
.
Mortgage i n t e r e s t - . . . . . . . . . . .
Taxes and i n s u r a n c e . . . . . . . . . . . . . . . . . - - _ . - .
.
Pronertv tax, special a s s e s s m o n t s . . . . . - . . _ - . _ . . . _ . .
Homeowners insurance . . . .
Maintenance and repairs..
.
Commodities
................
...--.._-.-.
at end of table.

92



West

South

Northeast

North
Central

A B C D

A B C D A B C D

A B C D

1
1
1
1

1 1 2
1 1 2
1 1 2
1 3 3

1
1
1
1

1
1
1
1

1
1
1
1

1
1
1
1

1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
3
3
3
1
1
1
1
1
1
1
1
1

1
1
1
1
3
3
1
1
3
1
1
1
1
1
3
3
3
1
3
3
3
1
1
1
1
1
1
3
1
1
2
2
2
1
1
3
3
3
1
1
1
1
1
1
1
1
1

1
3
3
1
3
3
1
1
3
1
1
3
3
1
3
3
3
1
3
3
3
1
1
1
3
1
1
3
1
1
2
2
2
1
1
3
3
3
1
1
1
1
1
1
1
1
1

1
3
3
1
1
1
1
1
3
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
3
1
1
1
1
1
1
1
3
3
3
1
1
1
1
1
1
1
1
1

1 1 2
3 3 3
3 3 3
1 1 2
3 3 3
3 3 3
1 1 2
1 1 2
3 3 3
1 1 2
1 1 2
1 1 2
1 1 2
1 1 2
3 3 3
3 3 3
3 3 3
1 1 2
1 1 3
1 1 3
1 1 3
1 1 2
1 1 2
1 1 2
1 3
3
1 1 2
1 1 2
3 3 3
1 1 2
1 1 2
1 3
3
1 3
3
1 3
3
1 1 2
1 1 2
3 3 3
3 3 3
3 3 3
1 2
2
1 2
2
1 2
2
1 2
2
1 1 2
1 1 2
1 1 2
1 1 2
1 1 2

1
3
3
1
3
3
1
1
3
1
1
1
1
1
2
2
2
1
3
3
3
1
1
1
1
1
1
3
1
1
1
1
1
1
1
3
3
3
1
1
1
1
1
1
1
1
1

1 1 2
3 3 3
3 3 3
1 1 2
3 3 3
3 3 3
1 1 2
1 1 2
3 3 3
1 1 2
1 1 2
1 1 2
1 1 2
1 1 2
3 3 3
3 3 3
3 3 3
1 1 2
3 3 3
3 3 3
3 3 3
1 1 2
1 1 2
1 1 2
1 1 2
1 1 2
1 1 2
3 3 3
1 1 2
1 1 2
1 1 3
1 1 3
1 1 3
1 1 2
1 1 2
3 3 3
3 3 3
3 3 3
1 2
2
1 2
2
1 2
2
1 2
2
1 1 2
1 1 2
1 1 2
1 1 2
1 1 2

2
3
3
2
3
3
2
2
3
2
2
3
3
2
3
3
3
2
3
3
3
2
2
2
3
2
2
3
2
2
2
2
2
2
2
3
3
3
2
2
2
2
2
2
2
2
2

1
1
1
1

2
2
2
2

1 1 1 2
1 1 1 2
1 1 1 2
1 1 1 2
1
1
1
1
3
3
1
1
3
1
1
1
1
I
2
2
2
1
1
1
1
1
1
1
1
1
1
2
1
1
2
2
2
1
1
3
3
3
1
1
1
1
1
1
1
1
1

1
3
3
1
3
3
1
1
3
1
1
1
1
1
3
3
3
1
3
3
3
1
1
1
1
1
1
3
1
1
2
2
2
1
1
3
3
3
1
1
1
1
1
1
1
1
1

1 2
3 3
3 3
1 2
3 3
3 3
1 2
1 2
3 3
1 2
1 2
1 2
1 2
1 2
3 3
3 3
3 3
1 2
3 3
3 3
3 3
1 2
1 2
1 2
1 2
1 2
1 2
3 3
1 2
1 2
2 2
2 2
2 2
1 2
1 2
3 3
3 3
3 3
2 2
2 2
2 2
2 2
1 2
1 2
1 2
1 2
1 2

1
1
1
1

2
2
2
2

Anchor­
age
1
1
1
4
1
4
4
1
4
4
1
1
4
1
1
4
4
1
4
4
4
1
4
4
4
1
1
1
4
1
1
4
1
1
4
4
4
1
1
1
1
1
4
4
4
4
1
1
1
1
1

Hono­
lulu

APPENDIX TABLE VII. REVISED CPI WEIGHT DERIVATION PROCEDURES FOR MAJOR EXPENDITURE CATEGORIES, INDIVIDUAL
EXPENDITURE CLASS TOTALS, CERTAINTY ITEMS, AND PROBABILITY ITEM TOTALS 1 WITHIN EXPENDITURE CLASSES, BY
REGION 2 AND CITY-SIZE STRATUM, FAMILIES OP TWO OR MORE PERSONS, 1900-61 CES—Continued
Expenditure data code, by region-city size stratum
EC No.

Item
No.

Description

Northeast

West

North
Central

South

A B C D A B C D A B C D A B C D
20
21

22-28
22
23-26
23
24
25
26
27
28
29-35
29
30
31

32
33
34
35
36-41
36-40
. 36
37
38
39
40

41

42-50
42-44
42
43-44
43

44a
44b
45-46
45
46
47-49
47
48
49
50

01
02
03
04
05
06

01
14
01
01

05
13
07
01
11
32
01
05

04
05

01
02
01
02

01
02
03
04
05
01
02
03
04
05
06
01
02
01
02
07

01
01
04
08
01
06

Services....... . .
..... .....................
Fuels and u t i l i t i e s . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Fuel o i l s . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Gas
.
.
Electricity..... . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
T e l erp h o n e . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Wate and s e w e r a g e . . . . . . . . . . . . . — . . . . . . . . . . . . . . . .
Household furnishings and o p e r a t i o n . . . . . . . . . .
......
Textile honsefurmshiiigs
.... .
....
Durable housefurnishings....
... ..
Furniture
..................
.............. 3 3
Living room suite
....
. . ... .
. 3 3
3 3
Bedroom s u i t e . . . . . . . . . . . . . . . . . . . . . . . . . . .
3 3
Rug. soft s u r f a c e . . . . . .
...................... 3 3
Appliances*
..............................
.. 3 3
Refrigerator.................................. 3 3
Other h o i i s e f u r n i s h i n g s . . . . . . . . . . . . . . . . . . . . . . . . 3 3
1
Housekeeping s u p p l i e s . . . . . . . . . . . . . . . . . . . . . . . . . . .
1
1
Domestic help
..... ......................
1
1
Apparel and u p k e e p . . . . . . . . . . .
..................
1
1
Suits, w i n t e r . . . . . . . . . . . . . . . . . . .
..............
1
Boys' a p p a r e l . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1
1
Coats, w i n t e r . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1
Dresses, street- . . . — . . . . . . . . . . . . . . . . . . . . . . . . . .
1
1
1
Footwear...............................
......
1
1
Street shoes, women's.. . . . . . . . . . . . . . . . . . . . . . . . . . .
Other apparel:
1
Commodities... . . . . . . . . . . . T . . . . . . . . . . . . . . . . . . . .
1
Services. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1
Dry cleaning, men's. . . . . . . . . . . . . . . . . . . . . . . . . . .
1
Dry cleaning, w o m e n ' s — . . . . . .
..............
1
Transportation ' . - - . . . . . . . . . . . _ . . . . . . . . - . _ . . . . . . . . . .
1
Private..........................................
3
Auto p u r c h a s e . - . - . - . - . . . . . - . . . . - - - . . . . . . . . . . . . .
3
New o a r s . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3
Used c a r s . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3
Gasoline and motor o i l . . . . . . . . . .
.......
....
3
3
Motor o i l . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3
Auto p a r t s . . . . . _ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3
Auto repairs and maintenance..
..........
Auto insurance. . . . . . . . . . .
.----Auto financing charges * . . . . . . . . . . . . . . . . . . . . . . . . ! 3
3
Driver's license fees.. . . . . . . . . . . . . . . . . . . . . . . . . . . 3
Parking, garage r e n t - . . . . . - . - . - . - - .....-...- 3
1
Local t r a n s i t . . . . . . . . . . . . . . . . .
........ 1
Taxicabs................---- .
..............
Rent of car * . . . . . . . . . . . . . . . . . . . . .
..... 3
Train fares...
.....................
..! 3
Airplane, steamship f a r e s . . . . . . . . . . . . . . . . . . . . . . . . . 3
Bus, intercitv fares . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
Health and recreation (excluding EC 51-52)
...
1
Medical c a r e . . . . . . . . . . . . . . . . . . . . . . . . . . _ _ . . . - . . . - . . 1
Drugs and p r e s c r i p t i o n s . . . . . . . . . . . . .
....
I 1
Over-the-counter i t e m s . . . . . . . . . - - - . . . - . . . . . - . . . 3
3
Prescriptions
.........
.
...
Medical care services..
.
.
..
i 1
Professional services------- . . . . .
....
3
Family d o c t o r . - - - ._
.
3
Dentists* f e e s . . . . . . . .
.....
3
Kye care
.
.- 3
3
TTosoital services.------..
.
.
Health insurance... . . . . . . . . . . . . . . . . . .
...... 3
Personal care. . . . . . . . . . . . . . . .
.....
......
Toilet goods..
-.
.
Personal care s e r v i c e s . . . . . . . . . . . .
.
Men's haircuts and s h a v e s . . . . . . . . . . . . . . . . . . . . .
Reading and recreation..
.
.
Recreational g o o d s . . . . .
... ... ...
.
TV sets and combinations......
...
Recreational s e r v i c e s . . . . . . - - . .
...
......
Indoor movies
.... .
.
Indoor sports
.... ...
.
Reading and education
.
.
.
..
Newspapers
« ...
3
Tuition fees
.........
.....
..... 3
1

3
3
3
3
3
1
1
3
3
3
3
3
1
I
1
3
3
1
3
3
3
3
3
3

3
3
1

1
1
1
1
1
1
1
1
1
3
3
3
3
3
3
3
3
3
3
3
3
3
3
1
1
3
1
1
3
3
3
1
1
3
3

2
2
1
1
1
1
2
2
2
3
3
3
3
3
3
3
3
3
3
3
3
3
3
2
2
3
2
2
3
3
3
2
2
3
3

1 1 2
1 1 2
1 1 2
1 1 2
1 1 2
1 1 2
1 1 2
1 1 2
1 1 2
1 1 2
1 1 2
3 3 3
3 3 3
3 3 3
3 3 3
3 3 3
3 3 3
3 3 3
3 3 3
1 1 2
1 1 2
1 3
3
1 3
3
1 1 2
1 1 3
1 1 3
1 1 2
1 1 2
1 3
3
1 3
3
1 3
3
1 1 2
1 1 2
1 3
3
1 3
3

1
1
1
1
1
1
1
1
1
1
1
3
3
3
3
3
3
3
3
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1

1 1 2
1 1 2
1 1 1
1 1 1
1 1 1
1 1 1
1 1 2
1 1 2
1 1 2
1 1 2
1 1 2
3 3 3
3 3 3
3 3 3
3 3 3
3 3 3
3 3 3
3 3 3
3 3 3
1 1 2
1 1 2
3 3 3
3 3 3
1 1 2
1 1 2
1 1 2
1 1 2
1 1 2
1 1 2
1 1 2
1 1 2
1 1 2
1 1 2
1 1 2
1 1 2

1
1
3
3
1
1
3
3
3
3
3
3
3
3
3
3
3
3
3
3
1
1
3
3
3
3
3
1
1
1
3
3
1.
3
3
3
3
3
3
1
1
1
1
1
1
3
1
1
I
1
3
3
1

2 1 1 1 2
2 1 1 1 2
3 1 3 3 3
3 1 3 3 3
2 1 1 1 2
2 1 1 1 2
3
3 1 1 3
3
3 1 1 3
3
3 1 1 3
3
3 1 1 3
3
3 1 1 3
3
3 1 1 3
3
3 1 1 3
3
3 1 1 3
3
3 1 1 3
3
3 1 1 3
3 3 3 3 3
3 3 3 3 3
3 3 3 3 3
3 3 3 3 3
2 1 1 1 2
2 1 1 1 2
3 3 3 3 3
3 3 3 3 3
3 3 3 3 3
3 3 3 3 3
3 3 3 3 3
2 1 1 1 2
2 1 1 1 2
2 1 1 1 2
3 3 3 3 3
3 3 3 3 3
2 1 1 1 2
3 13 3 3 3
3 3 3 3 3
3 3 3 3 3
3 3 3 3 3
3 3 3 3 3
3 3 3 3 3
2 1 1 1 2
2 1 1 1 2
2
2 1 1 3
2
3 1 1 3
2 1 1 1 2
2 1 1 1 2
3 3 3 3 3
2 1 1 1 2
3 1 3 3 2
3 1 3 3 2
2 1 1 1 2
3 3 3 3 3
3 3 3 3 3
2 1 1 1 2

1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
3
3
3
3
1
1
3
3
3
3
3
1
1
1
3
3
1
3
3
3
3
3
3
1
1
1
1
1
1
3
1
1
1
1
3
3
1

1 1 2
1 1 2
3 3 3
3 3 3
1 1 2
1 1 2
1 1 2
1 1 2
1 1 2
1 1 2
1 1 2
1 1 - 2
1 1 ?
1 1 2
1 1 2
1 3
3
3 3 3
3 3 3
3 3 3
3 3 3
1 1 2
1 1 2
3 3 3
3 3 3
3 3 3
3 3 3
3 3 3
1 1 2
1 1 2
1 1 2
3 3 3
3 3 3
1 1 2
3 3 3
3 3 3
3 3 3
3 3 3
3 3 3
3 3 3
1 1 2
1 1 2
1 1 2
1 1 2
1 1 2
1 1 2
3 3 3
1 1 2
1 3
3
1 3
3
1 1 2
3 3 3
3 3 3
1 1 2

1
1
1
1
1
1
1
1
1
1
1
3
3
3
3
3
3
3
3
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1

1
1
1
1
1
1
1
1
1
1
1
3
3
3
3
3
3
3
3
1
1
3
3
1
1
1
1
1
1
1
1
1
1
1
1

1 1 2
1 1 2
1 1 1
1 1 1
1 1 1
1 1 1
1 1 2
1 1 2
1 1 2
1 1 2 1
1 1 2
3 3 3
3 3 3
3 3 3
3 3 3
3 3 3
3 3 3
3 3 3
3 3 3
1 1 2
1 1 2
3 3 3
3 3 3
1 1 2
1 1 2
1 1 2
1 1 2
1 1 2
1 1 2
1 3
2
1 3
2
1 1 2
1 1 2
1 1 2
1 1 2

1 1 1 2
1 1 1 2
1 3 3 3
1 3 3 3
1 1 1 2
1 1 1 2
1 1 3
3
1 1 3
3
1 1 3
3
1 1 3
3
1 1 3
3
. 1 1 3 3
>1 1 3 3
1 1 3
3
1 1 3
3
1 1 3
3
3 3 3 3
3 3 3 3
3 3 3 3
3 3 3 3
1 1 1 1 2
1 1 1 2
3 3 3 3
13 3 3 3
13 3 3 3
3 3 3 3
3 3 3 3
1 1 1 2
1 1 1 2
1 1 1 2
3 3 3 3
3 3 3 3
1 1 1 1 2
3 3 3 3
3 3 3 3
3 3 3 3
3 3 3 3
13 3 3 3
3 3 3 3
1 1 1 2
! 1 1 1 2
1 1 1 2
1 1 1 2
1 1 1 2
1 1 1 2
3 3 3 3
1 1 1 2
3 3 3 3
3 3 3 3
1 1 1 2
3 3 3 3
3 3 3 3
1 1 1 2

Anchor­
age

Hono­
lulu

1
1
1
1
1
1 !
1 i
l !
l j
4
4
4
4
4
4
4
4
1
1
4
4
1
4
4
4
4
4
4
4
4
1
4
4
4

4
4
4

See footnotes at end of table.




93

APPENDIX TABLE VII. REVISED CPI WEIGHT DERIVATION PROCEDURES FOR MAJOR EXPENDITURE CATEGORIES, INDIVIDUAL
EXPENDITURE CLASS TOTALS, CERTAINTY ITEMS, AND PROBABILITY ITEM TOTALS * WITHIN EXPENDITURE CLASSES, BY

REGION 2 AND CITY-SIZE STRATUM, FAMILIES OF TWO OR MORE PERSONS, 1960-61 CES—Continued

Expenditure data code, by region-city size stratum
EC No.

Item
No.

Description

West

A B C

51

52

01
01
02
03
01
03

1
1
1
1
1
3
3
Funeral s e r v i c e s . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3

Cigarettes......................................
Alcoholic beverages * - - - - - - - - - - - . . . . . . . . . - - . - . . — - - . .
Beer and a l e . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Whiskey
.
...
Away from h o m e . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Financial and miscellaneous personal expenses * . . . . . . . . .

i Weights for probability item totals within EC's having both certainty
and probability items are derived by same procedures as certainty items;
for these EC's, probability totals are not separately shown.
* S M S A ' S and cities included in each region are as follows:
South
North Central
West
Northeast
A Boston
Chicago
Baltimore
Los Angeles
New YorkCleveland
Washington
San Francisco
Northeast Detroit
N.J.
St. Louis
Philadelphia
Pittsburgh
Dayton
B Buffalo
Atlanta
Denver
Indianapolis
Hartford
Dallas
Seattle
Wichita
Nashville
C Lancaster
Cedar Rapids
Austin
Bakersfield
Portland
ChampaignBaton Rouge
Urbana
Durham
D
Green Bay
Orlando
Kingston
Millville
Southbridge

Nonindex:
Athol
Burlington
Lewistown

Crookston
Devils Lake
Findlay
Logansport
Niles

Florence
McAUen
Mangum
Martinsville
Unipn
. Vicksburg

Klamath Falls
Orem

Cambridge
LaSalle
Manhattan.
Menasha
Owatonna

Cleveland .
Eureka
. Gainesville
Gallup
Griffin
Okmulgee
Reserve
Sebring
• These items were in the original CPI pricing list but were subsequently
dropped; weights have been allocated as follows:
EC 15-01, Board, to items of EC 15-03, Restaurant meals
EC 16-02, Room rent, to EC 16-01, Rent of house or apartment
EC 17-02, Settlement charges, to EC 17-01, Home purchase
EC 40-02, Auto financing charges, moved by EC 36-1, New cars
EC 41-03, Rent of car, to EC 41-02, Taxi fares
EC 52-01, Other financing charges, imputed to All items, EC 1-52
4
Weight derivation procedures for home purchase and financing refer to
CES and CHUS data after special editing.
• Subsequently adjusted by substitution of specially edited data for
purchase of new and used cars and auto financing charges.
• Weight derivation procedures for alcoholic beverages refer to ex­
penditures after adjustment for underreporting.
, 7 The relative importance of EC 52 to the total of EC 1-51 by region
was applied to the city data for EC 1-51.
Explanation of Codes
The code numbers appearing in the table indicate the scope of the data
used in the weight derivation. Most of the average expenditures used were
calculated as pooled averages, i.e., aggregate expenditures reported by all
index families in the specified area divided by the total number of
families in the sample. A slight variation of this procedure was used for
the D-stratum cities. Individual city averages, calculated as indicated
above, were combined as simple averages, i.e., the unweighted sum of the
city averages divided by the number of cities in the specified area. The
evised weights were derived from CES and CHUS data (as reported or as

94



1 1
1 1
1 1
1 1
1 1
3 3
3 3
3 3

Northeast

South

North
Central

D

A B C D A B C D

A B C D

2
3
3
3
3
3
3
3

1
1
1
1
1
3
3
3

3
1
1
1
1
3
3
3

1 1 2
1 1 2
1 3
3
1 3
3
1 3
3
3 3 3
3 3 3
3 3 3

1
1
2
2
2
3
3
3

3 3 3
1 1 3
1 1 3
1 1 3
1 1 3
3 3 3
3 3 3
3 3 3

3 3 3
1 1 2
1 3
3
1 3
3
1 3
3
3 3 3
3 3 3
3 3 3

Anchor­
age
4
1
1
1
1
4
4
4

Hono­
lulu-

3
3
3

adjusted after special editing procedures) for the following combinations
of city data:
Code 1. Average expenditures for index families in individual cities
Code 2. Average expenditures for index families in all cities within a
region-size class—pooled averages for A, B, C, stratum
and averages of city averages in D stratum .
Code 3. Average expenditures for index families in all cities within a
region
Code 4. Average expenditures for index families in all Alaskan cities.
For example, in the South the total expenditures for fresh fruits (EC 7) in
the largest cities (stratum A) were used in each city without adjustment*
but the distribution of this total was based on the average distribution in
all A-stratum cities in the South; in all other southern cities the distri­
bution of EC 7 was based on the average distribution in all southern cities
regardless of size.
Exceptions in Weight Derivation, Specific Expenditure Classes and Items
Item
EC
No.
Region
No.
NE
2
Average of A cities in Northeast for Pittsburgh
South
Average of C cities in South for Baton Rouge
NC
Average of all cities in North Central region for
Wichita
South
Average of all cities in South for Baltimore and
Durham
NC
Only 1961 city data used in Detroit for board.
15
D-city average recalculated excluding Cam­
bridge.
South
Average of C cities in South for Baton Rouge
18
All
City data for specific fuels priced and allocated
21
unpriced fuels to priced fuels
NC
29
Average of all cities in North Central region for
Wichita
South
Average of C cities in South for Baton Rouge
31
NC
Average of all cities in North Central region for
Wichita
NC
Average of A cities in North Central region for
36-40
St. Louis
01
All
D-cities: zero weight for cities without public
41
transit systems; for other D cities, D-city
average for cities having public transporta­
tion, including any reported expenditures for
cities without public transit..
51
NC
Average of C cities in North Central region for
Champaign-Urbana
South
Average of Dallas and Nashville for Atlanta.
Average of C cities in South for Baton
Rouge.
S&W
D-cities; where alcoholic beverages are not
sold by the drink, weights for drinks away
from home were set at zero; for other D
cities, average of D-cities where drinks are
available, including any reported expendi­
tures in cities where sale is prohibited by
law.
Honolulu Total obtained by applying the Western region
average ratio (including Honolulu) of EC
52 to EC 1-51, to the total of EC 1-51 in
Honolulu. Items within distributed by the
Western regional average relative importance
(including Honolulu).

APPENDIX TABLE VIII-A.

DERIVATION OF POPULATION WEIGHTS FOB REVISED CPI
CPI weights

(Percentage distribution)
Stratum and SMSA or city

1960 Census
urban
population

All
consumer
units

Index
consumer
units

j

Total

ASMSA's
New York
Chicago
Los Angeles
Philadelphia
Detroit..
San Francisco
Boston
.
Pittsburgh
St. Louis
Cleveland
Washington
Baltimore

38.918
11.415
5.133
5.305
3.110
2.845
2.105
1.942
1.572
1.452
1.405
1.463
1.171

39.106
11.944 1
4.803 1
5.462
3.000 !
2.663
2.406
2.001
1.565
1.360
1.276
1.459
1.167

40.021
12.577
5.552 I
5.017
2.703
2.895
2.372
1.930
1.565
1.428
1.325
1.255
1.402

40.021
12.577
5.552
5.017
2.703
2.895
2.372
1.930
1.565
1.428
1.325
1.255
1.402

BSMSA's
Northeast
Hartford
Buffalo
.
North Central
Dayton
Indianapolis...
Wichita—
South
Atlanta
Nashville
Dallas
West
Denver
Seattle

25.479
4.454

25.325
4.235

25.471
4.695

25.471

7.088

6.923

6.629

9.200

9.369

9.800

4.737

4.798

4.347

14.651
3.024

14.486
3.166

13.781
3.606

4.183

4.130

3.852

5.745

5.600

5.000

1.699

1.590

1.323

20.952
3.069

21.083
3.259

20.727
3.512

6.288

6.595

6.759

8.048

8.134 1

7.360

3.093 !

2.693

2.677

.066
.332
100.000

.065
.354
100.000

..

CSMSA's
Northeast
Portland, Maine
Lancaster, Pa
.
North Central
Champaign-Urbana, HI
Green Bay, Wis
.
Cedar Rapids, Iowa
South
Durham, N.C
Orlando, Ma
Austin, T e x . —
West
Bakersfield, Calif..
D cities
,
.
Northeast
Southbridge, Mass
Kingston, N . Y . . .
Millville, N.J
North Central
Findlay, Ohio
Logansport, Ind
Niles, Mich
Crookston, Minn
.
Devils Lake, N. Dak
South
Martinsville, Va.„.
Union, S.C
.
Florence, Ala . . .
Vicksburg, Miss
....
Mangum, O k l a . . . . . „ ,
McAUen, Tex
West
Orem, Utah
Klamath Falls, Oreg
Alaska (Anchorage)
Hawaii (Honolulu)
Total




'

I

.
..

......

.

|
1

.068 1
.386
100.000 i

Families
(percent)

11.290
4.969
4.052
2.590
2.679 !
1.919
1.681
1.432
1.319
1.283
.999
1.285

Singles

1.287
.583
.965
.113
.216
.453
.249
.133
.109
.042
.256
.117

2.348
2.347

2.081
2.080

.267
.267

2.210
2.209
2.210

2.017
2.016
2.017

.193
.193
.193

3.267
3.266
3.267

2.978
2.977
2.978

.289
.289
.289

2.174
2.173

1.875
1.874

.299
.299

1.803
1.803

1.563
1.563

.240
.240

1.284
1.284
1.284

1.122
1.122
1.122

.162
.162
.162

1.250
1.250
1.250
1.250

1.156
1.156 1
1.156
1.156

.094
.094
.094
.094

1.323

1.193

.130

1.170
1.171
1.171

1.041 i
1.042 !
1.042

.129
.129
.129

1.352
1.352
1.351
1.352
1.352

1.186
1.186
1.185
1.186
1.186

.166
.166
.166
.166
.166

1.227
1.227
1.227
1.226
1.226
1.227

1.155
1.155
1.155
1.154
1.154
1.155

.072
.072
.072
.072
.072
.072

1.339
1.338
.065
.354
100.000

1.197
1.196
.052
.305
89.432

.142
.142
.013
.049
10.568

13.781

20.727

95

APPENDIX TABLE VIII-B.

POPULATION WEIGHTS FOR B CITIES BEFORE AND AFTER ADDITION OF 6 CITIES

Index
consumer
units

CPI weights
Original

After addition
of 6 Cities

Percent
BSMSA's

25.471

_

Northeast
Hartford
Buffalo

4.695

North Central
Dayton
Indianapolis
Wichita
Cincinnati
Milwaukee
Minneapolis-St. Paul
Kansas City
-

6.629

South
Atlanta
Nashville
Dallas
Houston

West

Denver
Seattle
San Diego

96



9.800

.

1

4.347

2.348
2.347

2.348
2.347

2.210
2.209
2.210

1.096
1.095
1.096
.740
.850
1.042
.710

3.267
3.266
3.267

2.934
2.933
2.934
.999

2.174
2.173

1.838
1.837
.672

APPENDIX TABLE IX.
CONSUMER PRICE INDEX (NEW SERIES) RELATIVE IMPORTANCE OF MAJOR GROUPS, SUBGROUPS AND
SELECTED INDIVIDUAL ITEMS, DECEMBER 1963, AND COMPARISON WITH OLD SERIES X

Groups, subgroups, expenditure
classes,
and priced items l

All items
FoodFood at home
Cereals and bakery products
Cereals
Bakery products
White bread
Other priced items
Meats, poultry, and fish
Meats
Beef and veal
Hamburger
Steak
Other priced items
Pork.-.
Pork chops
Bacon
Other priced items
Other meats
Poultry
Frying chickens
Other priced items
Fish

__

Dairy products
Milk, fresh (grocery)
Milk, fresh (delivered)—
Butter
Other priced items
Fruits and vegetables
Fresh fruits
Apples
Bananas
Oranges
_.
Other priced items
Fresh vegetables
Lettuce
Potatoes
Tomatoes
Other priced itemsProcessed fruits and vegetables
Other food at h o m e . Eggs
Fats and oils
Margarine
Other priced items
Sugar and sweets
Nonalcoholic beverages
_
Coffee, can and b a g - .
Other priced items
Prepared and partially prepared food.
Food away from home
Restaurant meals
Between meal snacks
Housing
Shelter
Rent-Hotels and motels
Homeownership
Purchase ana financing
Home purchase
Mortgage interest—
Taxes and insurance
Real estate taxes
Property insurance
Maintenance and repairs
Commodities
Services
Fuel and utilities
Fuel oil and coal
Fuel oil
Coal
Gas and electricity
Gas
Electricity
Other utilities
Telephone
Water and sewerage

_

Household furnishings and operation
Textile housefurnishings
Furniture
Bedroom suiteLiving room suite
Other priced items

-.

New series
index

Old series
index

Percent of
all items
Dec. 1963

Percent of
all items
Dec. 1963

100.00

100.00

22.43
17.89
2.45
.80
1.65
.60
1.05

28.18
23.11
3.27
.98
2.29
1.68
.61

5.63
4.45
2.21
.57
.80
.84
1.30
.36
.30
.64
.94
.73
.51
.22
.45

6.43
5.21
2.07
.49
.77
.81
2.09
.51
.76
.82
1.05
.66
.66

2.80
.85
.68
.25
1.02
3.02
.76
.17
.15
.20
.24
.94
.16
.24
.14
.40
1.32
3.99
.64
.55
.15
.40
.64
1.01
.40
.61
1.15

3.81
1.19
1.20
.40
1.02
4.46
1.53
.31
.19
.52
.51
1.38
.19
.39
.26
.54
1.55
5.14
1.01

4.54
3.75
.79

5.07
5.07

33.23
20.15
5.50
.38
14.27
9.11
6.28
2.83
2.13
1.72
.41
3.03
.98
2.05
5.26
.73
.67
.06
2.71
1.30
1.41
1.82
1.38
.44

30.71
18.34
6.16
12.18
7.51
5.76
1.75
1.61
1.37
.24
3.06
.96
2.10
4.91
1.21
.55
.66
2.11
1.18
.93

7.82
.61
1.44
.28
.28
.88

7.46
.67
1.55
.41
.47
.67

.56

.60
1.00
1.30
.50
.80
1.03

Groups, subgroups, expenditure
classes,
and priced i t e m s 1

Household furnishings and operation
—continued
Furniture—continued
Floor coverings
Rugs, soft surface
Other priced items
Appliances
Refrigerator
Other priced items
Other housefurnishings
Housekeeping supplies
Housekeeping services
Domestic service
Babysitter
Postage
Other priced items
Apparel and upkeep
Men's and boys' apparel
Men's apparel
Suits, year round
Other priced items
Boys' apparel
Women's and girls' apparel
Women's apparel
Winter coat
Street dresses
Hose, nylon
Other priced items
Girls' apparel
Footwear
Street shoes, men's
Street shoes, women's
Other priced items
Other apparel
Commodities
Services
Dry cleaning
Men's suit
Women's dress
Other priced items
Transportation
Private transportation
Autos and related goods
Auto purchase
New cars
Used cars
Gasoline and motor oil
Gasoline
Motor oil
Auto parts
Automobile services
Auto repairs and maintenance
Other automobile expense
Auto insurance
Registration fees—
Drivers' license
Parking fees
Auto financing charges 4
Public transportation
Local transit
Taxicabs
Train fares
Airplane fares
Intercity bus fares
Health and recreation
Medical care
Drugs and pharmaceuticals
Over-the-counter items
Prescriptions
Professional services
Family doctor, house visit
Family doctor, office visit
Optometric examination and eyeglasses
Dentists' fees
Other priced items
Hospital services 5
Health insurance
Hospital services
Nonhospital services
Overhead
Personal care
Toilet goods
Services
_
-.
Men's haircut
Other priced items

New series
index
Percent of
all items
Dec. 1963

0.48
.34
.14
1.36
.28
1.08
.83
1.55
1.55
.26
.29
.23
.77
10.63
2.86
2.21
.36
1.85
.65
4.08
3.23
.28
.50
.39
2.06
.85
1.51
.26
.26
.99
2.18
.71
1.47
.79
.44
.35
,68
88
64
,02
02
,55
,47
,28
05
.23
.72
.62
.98
.64
.42
.37
.04
.18
.63
.24
.78
.14
.07
.20
.05
19.45
5.70
1.14
.50
.64
2.59
.12
.77
.29
.86
.55
.36
1.61
.66
.71
.24
2.75
1.52
1.23
.51
.72

See footnotes at end of table.




97

APPENDIX TABLE IX. CONSUMER PRICE INDEX (NEW SERIES) RELATIVE IMPORTANCE OP MAJOR GROUPS, SUBGROUPS AND
SELECTED INDIVIDUAL ITEMS, DECEMBER 1963, AND COMPARISON WITH OLD SERIES *—Continued

Groups, subgroups, expenditure
classes,
and priced items l

New series
index

Old series
index

Percent of
all items
Dec. 1963

Percent of
all items
Dec. 1963

Health and recreation—continued
Reading and recreation
Recreation
Recreational goods
TV sets
Other priced items.
Recreational services
Movies (indoor)
Bowling fees
Other priced items
Reading and education
Newspapers
College tuition
Other priced items

5.94
4.36
2.78
.63
2.15
1.58
.38
.36
.84
1.58
.50
.23
.85

5.57
4.39
2.31
.70
1.61
2.08
2.04

Other goods and servicesTobacco products
Cigarettes
Cigars

5.06
1.89
1.74
.15

4.31
2.06
1.94
.12

.04
1 .18
1 .18

*For this comparison, the items priced have been grouped for both
indexes according to the classification of the new series. The basis of
selection of items to be priced and the allocation of weights to the priced
items are not the same for the two indexes. In the old series, important
items were selected and unpriced items having similar price movements were
allocated directly to priced items. For the new series, the most important
items were selected with certainty, and carry their own importance.
Weights are shown separately only for these items. Some of them are
represented by more than one specification but the weights for the indi­
vidual specifications are not shown. The remaining weight of each ex­
penditure class is shared equally by the probability items.
APPENDIX TABLE X.

Groups, subgroups, expenditure
and priced items *

Other goods and services—continued
Alcoholic beverages
Beer
„
Whiskey and wine
Beer, cocktails away from home.
Personal expenses
Funeral services
Bank service charges
Legal services
Miscellaneous6
Special groups
Commodities
Durable
—
Nondurable
Services

New series
index

Old series
index

Percent of
all items
Dec. 1963

Percent of
all items
Dec. 1963

2.64
1.06
.78
.80
.53
.28
.12
.13
*.38

2.25
1.36

65.97
18.78
47.19
34.03

67.73
17.53
50.20
32.27

4

.85

8
Fifty percent of old# series weight for laundry; remaining 50 percent
included in apparel services.
* Includes 50 percent of old series weight for laundry; formerly in­
cluded
in Household operation.
4
Not actually priced; imputed from priced items.
8
Represented by directly priced services in new series; by premium
rates in old series.
• Personal financing charges other than mortgage interest and auto
financing.

REQUIRED NUMBER OF REPORTERS PER REPLICATED SUBSAMPLE BY COMMODITY GROUP (EXCLUDING
FOOD AT HOME AND ITEMS OBTAINED FROM SECONDARY SOURCES) X
Chicago,
Los Angeles,
New York

Item

Food away from home:
Restaurant meals *
Between meal snacks

Other A
strata cities

C&D
strata cities

B strata
cities'

15
5

12
4

12
4

12
4

750-1,100

500-750

200-500

150-450

200
5

200
4

150-300
4

75-176
4

5

4

4

4

5
5
5

4
4
4

4
4
4

4
4
4

Apparel and upkeep:
Men's, women's, and children's
Footwear
Apparel services

5
5
5

4
4
4

4
4
4

4
4
4

Transportation (private):
Auto purchase (new cars).
Gasoline and motor oil
Automobile services
Other automobile expenses

3
6
5
5

2
4
4
4

2
4
4
4

1
4
4
4

Housing (shelter):
Rent 4
_
Hotel, motel rentals
„„
Property tax
Maintenance and repairs -

__

..,„.,

Housing (fuel and utilities):
*Coal and fuel oil.
*Gas
_
♦Telephone
*Water and sewerage rates

.

.

Transportation (public):
Local transit
Taxi cabs

Personal care:
Toilet goods
Personal care services

See footnotes at end of table.

98



_
-

Housing (other):
House furnishings
Housekeeping supplies
Housekeeping services

Medical care:
Drugs and prescriptions
Physician's services
Dental services
Eye care
Hospital services
Lab t e s t s . . . . .
.

„. _
--.-.....

1U

.

.

All 56 cities,
2 samples
combined

*200

•119
• 78
T
247

«60
*76

.

.

.

.......
..

5
5-12
8-10
5
5
5
5
5

1

4
4-9
6
5
4
4

4
4-9
6
5
4
4

4
3
3
5
4
3

4
4

4
4

4
4

APPENDIX TABLE X.

REQUIRED NUMBER OF REPORTERS PER REPLICATED SUBSAMPLE BY COMMODITY GROUP
FOOD AT HOME AND ITEMS OBTAINED FROM SECONDARY SOURCES) *—Continued
Chicago, T'J Other A
Los Angeles,
strata cities
New York

Item

B strata
cities a

(EXCLUDING

C&D
strata cities

All 56 cities,
2 samples
combined

Reading and recreation:
Recreational goods and services.
Newspapers
College textbooks
Music lessons

,«171
'9

Other goods and services:
Tobacco products
Alcoholic beverages
Financial and miscellaneous personal expenses.
4
One-third of each sample in the 5 largest cities and one-half of each
sample in all other cities is priced on different semiannual cycles; total
sample
in non-replicated and replicated cities is the same.
5
Fifty outlets per region—city size based on the relative importance of
SMSA receipts to total receipts (1958 Census data) for hotels and motels in
the region.
• Sample in each city represents universe of specified outlets located in
the7 city and selected pricing areas—no fixed sample size*
Sample size limited.

♦Mail
questionnaire.
1
Required number not always available due to limited universe in smaller
(strata C & D ) cities; cities in which 2 subsamples are priced are indicated
in appendix table X I .
* Honolulu—3 outlets per subsample instead of 4 used for other B stra­
tum cities.
* One-third of each subsample of restaurants is priced in most cities on
different quarterly cycles.

APPENDIX TABLE X I .

SIZE OF INDEPENDENT FOOD STORE SAMPLE BY TYPE OF OUTLET BY SMSA OR CITY

l

Number of Independent Stores
Total

SMSA/city

Chicago. _ _ . - . - . . - - . . . . . - . . .
Detroit
Los A n g e l e s — - . ----New York
--Philadelphia
Baltimore
.
Boston
—
ClevelandPittsburgh
St. L o u i s . - - - San Francisco.----Washington
-

----~.

-

-

--- -

-

-

--_-

1

-

—

-

--

-.__

-

-

--_

---

--, --_

.
.
.

.

-

-

-

-

-

-

-------

-

-

-

-----

--

-----_--

-

-

-

-

-

---_

-

-

.

-

-

-

-

--

-

-

-

-__

-

Two subsamples combined in replicated cities.




-

--

-__

Baton Rouge—
Cedar R a p i d s . - - Champaign—UrbaD a .
Durham
Green Bay
-Lancaster
-_Orlando
.
Portland, Maine _ - — - . .
Anchorage
--------Crookston
Devils Lake
Findlay
Florence
Kingston
Klamath Falls
Logansport
Mangum.-._---.-..
Martinsville
McAllen
Millville
Niles
Orem
Southbridge Union
.
Vicksburg
---.

--

-

Atlanta.--Buffalo
Cincinnati-----------Dallas
Dayton--. - - ----------_
Denver------ ----- Hartford
Honolulu---------Houston.—------Indianapolis
Kansas City
Milwaukee—.- Minneapolis—St. Paul --._--___-_Nashville------.
San D i e g o - - - - - --Seattle
- Wichita
Austin.——

-

-

-

-

-------

.

_.

.__
, r. „

_,

Grocery
stores

Meat
markets %

Produce
markets *

54
50
44
79
55

37
36
33
35
29

13
11
7
30
18

4
3
4
14
8

31
32
34
33
27
31
26

18
20
19
17
20
22
21

6
9
10
10
4
8
4

7
3
5
6
3
1
1

23
33
32
24
26
29
28
28
26
28
27
30
31
24
30
29
24

20
17
18
21
21
20
22
22
22
20
23
16
21
22
22
23
23

2
10
8
2
3
6
5
3
3
3
3
7
7
2
6
5
I

1
6
6
1
2
3
1
3
1
5
1
7
3

12
10
13
14
10
13
11
14
11
13

10
10
10
10
9
9
10
10
9
10

3
4
1
1
1
3

8
5
5
6
8
8
5
6
3
5
5
3
5
5
5
5
8

8
5
5
5
8
8
5
5
3
5
5
3
5
5
5
5
8

2
1

2

3

1

1

_
«.
—
_3
1
2

_
_
«.
«.
..
_
_
_
«.
_
_
—
«.
—
_
_

* Including a few chain outlets.

99

APPENDIX TABLE XII.

CITIES AND PRICING SCHEDULE FOR THE REVISED CONSUMER PRICE INDEX
Pricing schedule '
Other items

City l and size stratum
Food*

Schedule *

Samples
M

A. Standard Metropolitan Statistical Areas of 1,400,000 or more in 1960:
*Baltimore, Md
*Boston, M a s s . . .
♦Chicago-Northwestern Indiana 4
♦Cleveland, Ohio
♦Detroit, Mich
-.
♦Los Angeles—Long Beach, Calif
♦New York—Northeastern New Jersey *_ .
♦Philadelphia, Pa

__
_

1A,2B—
1A,2B__
1A,1B,2A,2B~
1A,1B,2A,2B—

i
_ _. ._

-

_.

..

-

♦Kansas City, Mo
♦Milwaukee, Wis
-♦Mmneapolis-St. Paul, Minn
Nashville, Tenn
*San Diego, Calif
.

..

1A,2B
1A,2B
1A,2B—
2
1A,2B

_

_

1A.2B
1A,2B

C. Standard Metropolitan Statistical Areas of 50,000 to 249,999 in 1960:
Austin, Tex
Rakersfield, Calif
,.._
Baton Rouge, La—_ .
Cedar Rapids, Iowa
Champaign—Urbana, 111- - _

_

Martinsville, V a .
McAllen, Tex
Millville, N.J
Niles, Mich
Orem, Utah
Southbridge, Mass
Union, S.C
Vioksburg, M i s s . .

-

_

.,

.. - ,r -.- .

-

-- .
, ,„ .,...,
_

..

„. „,..

,_

...
_

--.---_-_-__--_..

_
-

._




_.
_

„

_.._..
_

,„

r

.,„
_

_

._--

1
2
2
1A,2B
1A.2B

1
2
2—
1A,2B
1A,2B
1A.2B
1
1
1A.2B
2

1A,2B
1
1
1A,2B_
2

1,2
1
._ 2
1
1

1,2
_ 1
2
1
1

2
1
»,.,.2
1
2

2

_

1 1
2
1
2

-

2
2
1
2
1

2
2
1
2
1

.

1
-. 2

1
2

_

♦Indicates
areas for which separate indexes are published.
1
The 18 largest Standard Metropolitan Statistical Areas as defined for
the 1960 Census of Population were selected on a certainty basis and
represent themselves only in the population weight patterns. The other
sample selections carry not only their own population weights but also
prorata shares of the population weights of all cities in their region in the
same
population class.
2
Item samples are identified as samples " 1 " and " 2 . " Outlet samples
are identified as samples " A " and " B . " The determination as to the
extent of sampling within an area depended on plans for publishing separate
area indexes and on plans for developing estimates of sampling error and its
components.

100

.,,"

_

D . Urban Places of 2,500 to 49,999 in 1960:
Anchorage, Alaska
-------Crookston, Minn .,„
., ,
-,Devils Lake, N.D
Findlay, Ohio
Florence, A l a - - - - . . - . - - - . .
Kingston, N.Y
TCtaprntfo Fail«, Oreg
Logansport, Ind

|lA,2B

.

-

_

_.,,,

1A,2B
1
1A,2B
1A,2B
2

..-..._--_.

2

3

X
X
X
X
X
X
X
X
X
X
X
X

1A,2B
1A,2B
1A,2B
1A,2B
1A,2B

___
,, . ^
1
.-1

. „..

«

♦Houston, Tex_.
.
Indianapolis, I n d .

Durham, NTC
Green Bay, Wis
Lancaster, Pa . - - . . .
Orlando, Fla
Portland., Maine

1A.2B
1A,2B
.. |lA,lB,2A,2B„ 1A,2B
1A,2B

_

B. Standard Metropolitan Statistical Areas of 250,000 to 1,399,999 in 1960:
♦Atlanta, Ga
_
♦Buffalo, N.Y
_
_
_
♦Cincinnati, Ohio
„ ™« -, -,
. „,
,,.
_ _ „ , ,„ _
♦Dallas, Tex
Dayton, O h i o . . . .
_
-

♦Seattle, Wash
Wichita, Kans

«.

„

♦Pittsburgh, Pa
♦St. Louis, Mo
♦San Francisco-Oakland, Calif
♦Washington, D.C

Denver, Colo
Hartford, Conn

1A,2B
1A.2B
1A,1B,2A,2B„
1A.2B

_

1

X
X
X
X
X
X
X

x
X

x
x
X
X
X
X
X
X

X

x
X

X
X

x
X
X

x
X

X
X
X
X
X
X
X
X

x
X

x
X
X
X
X
X
X

' Foods, fuels, and several other items are priced every month in all
cities. Prices of a few items are collected semiannually or annually in all
cities. Prices of other goods and services are obtained on the schedule
indicated:
M « Every month.
1 » January, April, July, and October.
2=February, May, August, and November.
3 = March, June, September, and December.
* Standard Consolidated Areas.

EXHIBIT A
Budget Bureau No. 44-R1081.2

US. DEPARTMENT OF LABOR
Bureau of Labor Statistics
Washington 25, D.C.

REPORTS WILL BE
HELD IN CONFIDENCE

BIS 2549
Rev 5-1-60

BLOCK BOUNDARIES:
North

Block N o . .

. Page

COMPREHENSIVE
H O U S I N G U N I T SURVEY
Listing Form

East.
South-

of

Pages

(City and State)
(Suburban Area)

West.
In-BloclcTtattn

City

f{

IDENTIFICATION

ALL HOUSING UNITS

I

ALL

tt

LIVING
QUARTERS

!

TYM OF HOUSING
UNIT**
With sopororo
1. Kitchtn f o d i i . j

m.
SHEET
NO.

STREET
NAME

2. Kitchtn focili-!

LIME
NO.

NO.
OK
IOC.'

instollod
3. No kitthtn
facilities
Without separate
entrance
j
4 . Kitthtn
!
facilities

2

1

Area

j

Page N o .

Block

1

N o t for
field use

3

!

4

|

'

5

(Col.

ALL
VACANT
HOUSING
UNITS

5 , code I o r 2 )
STMICTUtE

WE
1 . Single,
dttochtd
2. Singio,
semi­
detached
3 . Single,
attachtd
4. Multiunit
5. Other
(Sperffy)

OF
UNITS

YEAIMJIIT
10. More
1920
20.19201929
30. 19301939
40. 19401949
After 1949
enter last
2 digits
of year.

CONDITION

|

7

8

9

ALL OCCUPIED HOUSING UNITS
( C o l . 5 , code 1 or 2 a n d n o entry in C o l . 1 0 )

1
1.
2.
3.
4.

'not

2 . Deterio­
rating
3. Dilapidated

1

6

1. For rent

|

occupied
3. for sole
4 . Sold-not
occupied
5. Held for
occasional
use
t . Under construction
7. Other or
unknown

TENURE
Tenant
Owner
Concession
lent free

MCE
1 . White
2. Negro
3 . Other

HUMIEt
OF
PERSONS
IN
UNIT

10

"

12.

13

1

5
| 4
i

5

!

6

\

7

:

8

9
10

11

'

1 2

Vb

u

1

15

i

!

**If column 5, code 1, complete reverse side.

♦Code
Hoor NambeMst FL, 2d FL, etc.

F-Front

B'Basemen*

R-Rear




L-Left

RvRight
Interviewer

Enter'T'
if nophono
ovailtbl.
in unit.
Enter " 2 "
if phone
refused.

1

12

TEIEFKONE
NUMKK

Date.

14

EXPLANATIONS
LINE
NO.

j

COLUMN
NO.

COMMENT

ALL OCCUPIED HOUSING UNITS WITH INSTALLED KITCHEN FACILITIES
NUMtER
OF
MOMS
M
WOT

UNE
NO.

Bathroom Facilities
WATER
MUWT

COMPLETE
IATHOOM

OTHER
IATHIOOM

1. Mono
2. Colt" only

1. No. com*

FACILITIES
PIIVA1E ONLY

PRIVATE
2. Ono only
3. Two or

2. Wash bowl

tooted by:
3. 60s
4. Electricity
5. Cool
e. Oil

SHAKO
4. Shored
only

'5

j

(Col. 3, code 1 and C o l 11, code 1 to 4)

16

17

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
IBBBSSBHB




18

FUEL
1. Mono
2.6os
3. Elec­
tric.
ity
4. let
or

toilot
4. Tober

HEATW6
EQUIPMENT

NEATIN6
FUEL

Alt
COkTMTION.
IN6

1. None
2. 6es
3. Electrieity
4. Cool

5.011

2. Control
3. Other,
installed
4. Other,
not
in.
stalled

2. 6»
3. Elec
trie
ity
4. Cool
5. Oil
t . Other

1. None
2. Control
system
' wit(s)

UUINDIY
EQUIPMENT
1. No washer
or dryer
2. Washer
bet no
dryer
3. Dryer—gas
4. Dryerelec.
5. Coin

6AIA6E
1. None
2. l e w
3. 2 cars
or
more
4. Cor.
port

equipment
only

5. Any two

19

20

21

22

23

OWNER OCCUPIED UNITS

1

24

25

26

CENTtAl
SERVICES
AVAILAIIE
t . Elevator II
2. Switch. I!

( C o l . 1 1 , code 2 )
DATE OF
PURCHASE

PRICE

(Enter
last 2
digift

if 1957
or ofter
in Col.

3. Elov. t 11 year.)
SJ.
4. Other
(Specify
in foot- II
note.)

27

28

».|

ESTIMATES
CURRENT
MARKET
VALUE
(To
SIM)

(To
nearest
$100)

29

30

|
|

EXPLANATIONS
LINE
NO.

COLUMN
NO.

COMMENT

ALL TENANT-OCCUPIED HOUSING UNITS W I T H INSTALLED KITCHEN FACILITIES
1
Equipment Included i n the Rent i n C o l . 44 o r 4 6

1

FURNISH­

REFRIG-

IN­

HEATING

All

UUNDRY

INGS

EMTOt

STALLED

EQUIP-

CONDITION-

EQUIP­

COOK

MBIT

1 . Hon*

UK)

MENT

1

1 . No

|

2 . fort

2. Yot

I

3. M l

STOVE

1 . No

1 . No

1 . No

1 . No

2. Y «

2 . Yot

2. Yot

WATER

ELEC­

GAS

HEAT

GARAGE

T. No

1 . No

1 . No

2 . Yot

2. Y«

2. YK

TRIC­
1 . Nono

ITY-

only

3. Mot

1 . No

.32

33

34

35

36

1 37

annul

MACN.

SERVICES

CODE

INCOME

CODE

1 . Elovator

1-Undor $2,000
2-2,000-2,999

board

3-3,000-3,999

3.»ov.t
S.I.
tOthor
(SpwifV
in footnot..)
S.Nont

38

ANNUAL FAMILY

MACN.

i

2 . Switch­

2. Y «

and
cold

31

RENT

Services Included in Rent i n C o l . 44 o r 4 6

2 . Cold

2. Y«

I

1 ALL OCCUPIED
|
HOUSING
UNITS
|

(Col. 5. code 1 and Col. 1 1 , code 1)

|

39

40

41

42 '

ftR

ftt

4-4,000-4,999

WEEK

S-S.000-S.999
*-0,000-7,4W
7-7.S00-9.999
1-10.000-14,999
9 - 1 5 , 0 0 0 and ovor

43

45

44

46

II

47

2
2
2
2

!

*

2
2

2

2
2
2

|




2
2
2

'

■

2
2

1

U S. GOVERNMENT PRINTING OFFICE . I » * 0

Of—S»»7I<

103

EXHIBIT B
SAMPLING FRAME FOR SELECTION OF THE CPI ITEM SAMPLE
Item
No.

Item
No.

FOOD (267 items)
Food at home (264 items)

CES
schedule No. 1

CES
schedule No. 1

Cereals and bakery products (35 items)
EC-1 Cereals and grain products (19 items)
1.
2.
3.
4.
6.
6.
7
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18,
19.

1.
2.
3.
4.
6.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.

26-C fot)

27-C
28-C
29-C
30-C
31-C
32-C
33-C
34-C

36-C
36-C
39-C
40-C
41-C
41-C
41-C

EC-4 Poultry (4 items)
1.

White flour, all purpose
Whole wheat flour, soybean and other flour
Biscuit or roll mix
[Cake mix and cake flour
Muffin, gingerbread, etc., mix
Pancake and waffle mix
Pie mix
Corn cereals, ready-to-eat
Wheat cereals, ready-to-eat
Rice cereals, ready-to-eat
Bran cereals, reaay-to-eat
Bread crumbs, cracker meal, prepared stuffing
Cornmeal
Cornstarch, rice flour, and other thickening
Grits and hominy
Macaroni, spaghetti, noodles, etc.
Rice
Rolled oats
Wheat cereals, cooked

16-C*
17_C (pt)—.
18-C
19-C
17-C (pt) 20-C
2L-C
22-C
24-C
25-C
26-C (pt) .

(pt).
(pt)_—
.
(pt)
(pt)
(pt).

37-C.
38-C
43-C
42-C (pt)
42-C (pt)
42-C (pt)
44-C
45-C (pt)
45-C (pt). .

2.
3.
4.

12.
3.
4.
5.
6.

48-50-C
51-C
52-C
53-C
54-C
55-C
56-C (pt) 56-C (pt). _.
58-C...
60-C
57-C... .
59-C . __

13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
23,
24

61-62-C
63-C
64-C
65-C
66-C (pt)
67-C
68-C
69-C
70-C
71-C
72-C (pt)
72-C ( p t ) . - -

25.
26.
27
28,
29.
30.
31.
32.
33.
34.
35.
36.
37.
38.

75-C
73-74-C
76-C (pt)
77-C (pt)
77-C (pt)
77-C (pt)
77-C (pt)
77-C (pt)
78-C
79-C
82-C
83-C
84-C
80-C

.

...

. .

1.

Plain rolls, biscuits, or muffins, baked
Rolls, biscuits, or muffins, partially baked
White bread
Whole and cracked wheat bread
Rye bread
French bread
Other bread, other than white, whole wheat,
rye, and French
Soda crackers including saltines
Crackers other than soda crackers and saltines
Cookies
Cake
Pie
Pastry
Doughnuts
Sweet rolls
Coffee cake

2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.

See footnotes at end of table.

104



\ Chicken fryers, broilers, etc., fresh
Chicken parts, fresh
Chicken, parts or whole, frozen
Turkey, fresh or frozen

93-C
94-C
95-C
96-C
97-C
98-C

Fish, whole, fresh or frozen
Fish fillets and steak
Shell fish, fresh or frozen
___ Canned tuna fish
_ Canned salmon
Other canned fish
Dairy products
EC-6 Dairy products (19 items)

13.
14.
15.
16.
17.
18.
19.

Meats, poultry, and fish (48 items)
EC-3 Meats (38 items)

1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.

85-C
87-C
92-C
86-C
88-89-C
90-91-C

EC-5 Fish (6 items)

EC-2 Bakery products (16 items)

Beef and veal (12 items)
Steak
Beef liver
Ground beef
Chuck roast
Rib roast
Rump roast, brisket, etc.
Dried beef
Corned beef
Veal cutlet, steak
Veal stew meat and other veal
Calves' liver
Veal roast
Pork (12 items)
Pork chops, center cut and end cut
Fresh ham, whole and half
Pork sausage, fresh
Pork loin roast
Other fresh pork
Bacon
Ham slices, smoked
Ham, whole and half, smoked
Picnics (shoulder)
Salt pork, bellies, fatback, etc.
Canadian bacon
Butts
Other meats (14 items)
Leg of lamb
Lamb chops, loin and rib
Shoulder lamb
Bologna
Boiled ham
Salami
Meat loaf
Other cold cuts
Frankfurters
Smoked sausage
Canned ham
Pressed ham
Other canned meat
Tongue, heart, kidney, etc.

FOOD (267 items)
Food at home (264 items)

2-C
3-C
4-C
4-C
4-C
4-C
8-C
8-C
8-C

(pt)
(pt)
(pt)
(pt)
(pt)
(pt)
(pt)

1-C...1

5-C
6-7-C
9-C

11-C (pt)
11-C (pt)
11-C (pt)
12-C
13-C(pt)
13-C (pt)
214-C

Fresh milk
a. bought in stores
b. delivered
Skim milk
Buttermilk
Half and half milk
Chocolate milk
Coffee cream
Whipping cream
Ready whipped cream
Evaporated and condensed milk
Malted milk and other prepared milk powders
Powdered milk and powdered cream *
Ice cream, sherbets, and other frozen milk
products
American cheese
Processed American cheese
Solid cheese other than American
Cheese spreads
Cottage cheese
Cream cheese
Butter
Fruits and vegetables (83 items)
EC-7 Fresh fruits (15 items)

12.
3.
4.
5.
6.
7

102-C
103-C
104-C (pt)
104-C (pt)
105-C
107-C
- 109-C
8. 108-C (pt)
9. 108-C (pt)
10. 124-125-C
11. 111-C
12. 110-C
j 1 13. 106-C
14. 112-C (pt)
1 15. 113-C (pt)

Apples
Bananas
Strawberries
Other berries
Grapefruit
Lemons and limes
Oranges
Watermelons
Cantaloupes
Fruit juice, fresh
Pears
Peaches
Grapes
Tangerines
Plums and prunes
EC-8 Fresh vegetables (20 items)

;

1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.

1 17.

141-C
143-C (pt)
144-C.:.
145-C
146-C
147-C
148-C
149-C
150-C
160-C
152-C
154-C
155-C
156-C
158-C
159-C
142-C (pt)
161-C (pt)
157-C...;

Asparagus
Broccoli
Cabbage
Carrots
Cauliflower
Celery
Corn, sweet
Cucumbers
[Lettuce and salad greens
Onions
Potatoes, white
Snap beans, green or wax
Spinach, kale, or other cooking greens
Tomatoes
Turnips and rutabagas
JOther fresh vegetables
Sweet potatoes, yams

SAMPLING FRAME FOR SELECTION OF THE CPI ITEM SAMPLE—Continued
Item
No.

FOOD (267 items)
Food a t home (264 items)

. CES
schedule No. 1

litem
No.

CES
schedule No. 1

E C - 8 Fresh vegetables (20 items)—Con.
18.
19.
20.

151-C
143-C (pt)
153-C

Lima or kidney beans
Brussels sprouts
Peas
EC-9 Processed fruits and vegetables
(48 items)

1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.

118-C (pt)
118-C (pt)
119-C (pt)
119-C (pt)
120-C
121-C
122-C
123-C (pt)
132-C
133-C
134-C
135-C
137-C (pt)
136-C
137-C (pt)
114-C.__
115-117-C

18.

126-C
130-C
131-C (pt)
127-C
128-C
129-C
182-C
183-C
184-C

19.
20.
21.
24.
25.
26.
27.
28.
29.
30.
81.
32.
33.
34.
35.
36.
37.
38.
39.
40.
41.
42.
43.
44.
45.
46.
47.
48.

172-C
173-C
174-C
175-C
176-C
177-C (pt)
177-C (pt).
178-C..J.
179-C
180-C
139-C
138-C
140-C
162-C
163-C
164-C
165-C
166-C
167-C
168-C (pt)
168-C (pt)
169-C
170-C
171-C (pt)
181-C.:._
185-C

Processed fruits:
Apples, can or jar
Apple sauce, can or jar
Fruit cocktail, can or jar
Citrus segments, can or jar
Peaches, can or jar
Pears, can or jar
Pineapple, can or jar
Other canned fruits
Apple juice, can or jar
Grape juice, can or jar
Mixed fruit juices, can or jar
Orange juice, can or jar
Grapefruit juice, can or jar
Pineapple juice, can or jar
Prune juice, can or jar
Strawberries, frozen
Other frozen fruits and berries except straw­
berries

Other food a t home (79 items)
EC-10 Eggs (1 item)
15-C.

Eggs
EC-11 Fats and oils (8 items)

215-C
216-C
217-C
218-C (pt)
218-C (pt)
219-C
220-C
221-C

EC-12 Sugar and sweets (12 items)
J-30-b.
233-C
224-C
225-C
226-C (pt)
226-C (pt)
228-C (pt)
228-C (pt)
229-C
230-C
231-C
227-C
232-C

Lard
Margarine
Shortening except lard, butter, margarine
French dressing
Salad dressing except French
Mayonnaise and other cooked dressing
Peanut butter
Salad and cooking oil

>Candy
Chewing gum
Icing, fudge mixes, other mixes
Jellies and jams
Preserves and apple butter
Pudding mixes
Gelatin mixes
Brown sugar
White sugar
Corn syrup and maple syrup
Honey and molasses
Chocolate syrup and other flavored syrup
EC-13 Beverages, nonalcoholic (8 items)

234-C
235-236-C
237-238-C
239-240-C
241-C
242-C
243-C (pt)
243-C (pt)
244-245-C

Cocoa
Coffee in bags or cans
Instant coffee and coffee substitutes
Tea, bags, leaves, tea concentrates
Cola drinks
Ginger ale and soda
Other carbonated drinks
Noncarbonated drinks, liquid or concentrate
EC-14 Prepared and partially prepared
foods (50 items)

[ Other frozen juice
Lemonade, frozen
Mixed fruit juice, frozen
Orange juice, frozen
Prunes
Raisins
Other dried fruits
Processed Vegetables:
Asparagus, can or jar
Beets, can or jar
Corn, cream style or whole kernel, can or jar
Lima and kidney beans, can or jar
Peas, green, can or jar
Potatoes, white, can or jar
Potatoes, sweet, can or jar
Snap beans, green or wax, can or jar
Tomatoes, can or jar
Other canned vegetables
Tomato juice, can or jar
I Mixed and other vegetable juice, except
l
tomato
Asparagus, frozen
Broccoli, frozen
Brussels sprouts, frozen
Corn, cut, frozen
Green beans, frozen
Lima beans, frozen
Peas and carrots, frozen
Mixed veg. except peas and carrots and suc­
cotash, frozen
Peas, frozen
Spinach, frozen
Other frozen vegetables
> Beans, peas, lentils, etc., dried

FOOD (267 items)
Food a t home (264 items)

187-C
188-C
189-C
190-C (pt)
190-C (pt)
191-C
192-C
246-C
247-C
248-C
249-C
250-C
251-C
252-C
253-C (pt)
253-C (pt)
193-C
200-C
201-C
194-C
195-C
196-C
199-C
198-C
259-C (pt)
254-C
256-C
257-C (pt).
257-C (pt)
258-C
260-261-C.
202-C (pt)
202-C (pt)
202-C (pt)
255-C
203-C
204-C
206-C
207-C (pt)
207-C (pt)
208-C (pt)
208-C (pt)
208-C (pt)
208-C (pt)
209-210-C
212-C
211-C
213-C
197-C

(») —

202-C (pt)
259-C (pt)

Chicken soup, canned
Tomato soup, canned
Vegetable soup, canned
Bean soup, canned
\ Other soup, frozen and canned, (except chicken,
/ tomato, vegetable and bean)
Dried soups
Baby food, cereal
Baby food, puddings
Baby food, SOUP
Baby food, strained and chopped fruit
Baby food, strained and chopped meat
Baby food, strained and chopped mixtures
Baby food, strained and chopped vegetables
Baby food, powdered formulas
Baby food, junior cookies and teething biscuits
Baked beans
Sauerkraut, canned
Spaghetti with sauce or meat balls
Chicken with noodles, chicken a' la king, etc.
Chili con came
Chow mein, chop suey
Enchiladas, tamales, etc.
Corn beef hash
Tomato catsup and chili paste
Baking powder, soda, yeast
Olives
Relishes
Pickles
Salt, spices, seasoning
Nuts
Instant potatoes preparation
Spanish rice and other rice preparations
Macaroni and cheese
Extracts, flavors
Potato chips
Corn chips, popcorn and other snacks
Fish sticks, frozen
Meat pies, frozen
Poultry pies, frozen
Turkey dinner, frozen
Beef dinner, frozen
Fish dinner, frozen
Dinners, frozen except turkey, beef, fish
Frozen desserts
French fried potatoes, frozen
\ Macaroni and cheese, and other frozen prepared
/ dishes
Prepared salads
Dietary formula products
Prepared dishes except Spanish rice, macaroni
and cheese, and instant mashed potatoes
Sauces and gravies except catsup and chili
paste

See footnotes at end of table.




105

SAMPLING FRAME FOE SELECTION OF THE CPI ITEM SAMPLE—Continued
FOOD (267 items)
Food at home (264 items)

CES
schedule No. 1

Item
No.

EC-20 Maintenance and repairs, services
(30 items)—Con.

EC-15 Food away from home
(3 items)
J-24
Q-13 (pt)
J-25 (pt)
J-21
J-26
Q-16

[Board
Restaurant meals
Between meals snacks
E-II-6 (pt).
E-I-10 (pt).
E-II-4
E-II-1 (pt).
E-II-1
1-1- (pt).
(Pt)E - I I1-1
- (Pt).
E - I -11
(Pt)E-II-6
1-6 ((pt).
E-II-6 (pt).

HOUSING (212 items)
Shelter (63 items)
EC-16 Rent (4 items)
C-l(pt)
C-3
SQ-I-2a (pt) *.
C - l (pt)
Q-13 (pt)
SQ-I-2a(pt)..
C-l (pt) J l . . . .
Q-12
SQ-I-2a (pt) —
D-I-0

EC-21 Fuels and utilities (6 items)
■Rent of rooms
G-7
G-8(pt)_
G - l l (pt)
G-8(pt).
G-9
G-10—.
G - l l (pt)

Hotel, motel rentals
Ground rent 8

G-4
G-5
G-l-3...
G-12
G-13
G-16
G-15 (pt)

Purchase
Settlement charges
Mortgage interest contracted and other costs

Solid Fuel;
Coal, coke
Fuel oils
Utilities:
Gas
Electricity
Telephone
Water and sewerage charges

Housefurnishings (111 items)
EC-22 Textile housefurnishings (20 items)

EC-18 Taxes and insurance (2 items)
D-I-3
D-I-8
D-I-10 (pt)
D-i-7..:.:
D-I-10 (pt)

Window panes, replacement
Addition of new room, porch, etc.
Garage construction
General remodeling, siding, structural change
Fence or retaining walls, installation or re­
placement
New walks, patio, installation
Awnings, installation
Central air conditioning
Completing unfinished room
New bathroom construction
Recreation room construction
Termite protection
New lawn development
Landscaping, planting of trees or shrubs

•Rent of house or apartment

Homeownership (6 items)
EC-17 Purchase and financing (3 items)
D-I-13....
D-I-15,16.
D-II-4
D-II-5
D-II-6
D-II-13...
D-II-15...
D-II-22...

HOUSING (212 items)

CES
schedule No. 1

Sheets
Pillow
Pillows

Property tax, and special assessments
Homeowners insurance

Blankets
Maintenance and repairs (44 items)
EC-19 Maintenance and repairs, com­
modities (14 items)

Blankets, electric
Bedspreads, couch covers
Curtains, all fibers
Ready-made draperies, all fibers
Custom-made draperies, all fibers
Slipcovers, ready-made
Slipcovers, custom-made
Bath towels and hand towels
Bath mats and bath sets
Material for curtains, draperies, etc.
Materials for handwork (crochet thread, yarn
for needlepoint, etc.)
Table linens
Shower curtains
Kitchen towels
Comforters and quilts
Tablecloth and table mats, plastic

Outside paint, etc.
Roofing materials
Inside paint, varnish, etc.
Wallpaper
Plumbing materials
Water heater and parts
Furnace parts
Glass, screening, etc.
Lumber
Tile
Other building materials
Plants, shrubs, garden supplies, etc.
Fence, slats, etc.
Electrical materials

Furniture and floor coverings
EC-23 Furniture (31 items)

EC-20 Maintenance and repairs, services
(30 items)
E - I - l (pt'
E-I-2 (pt
E - I - 2 (pt:
E-I-3 (pt:
E - I - 3 (pt:
E - I - 4 (pt:
E-I-6 (pt
E-I-6 (pt
E-I-7 S t
E-I-7 (pt
E-I-8 (ptj
E - I - 8 (ptj
E - I - 8 (pt_.
E-I-9 (pt).
E-I-10 (pt)
E-I-10 (ptj

Outside painting
Roof replacement
Roof and gutter repairs
Fence repair
Outside repairs, except fence
Redecorating
Floor refinishing
Electrical repair
Sink replacement or installation
Other plumbing repairs
Furnace or other heating equipment cleaning
Heating equipment replacement or installation
Heating equipment repairs
Water heater repair or replacement
Storm doors and windows, installation
Screens, installation

See footnotes at end of table.




13.

1-1-1.
1-1-2.
1-1-3 (pt]
1-1-3 (pt
1-1-4 (pt
1-1-4 (pt^
1-1-4 (pt:
1-1-6 (pt:
1-1-5 5>t:
1-2-1
1-2-6 (pt)
1-2-2.....
1-2-5
1-2-3
1-2-5 (pt)
1-2-4
1-2-5 (pt)

Living room suite
Chair, fully upholstered
Cocktail or coffee table
Living room table except cocktail or coffee
Sofa, standard
Sofa, sectional
Sofa, dual purpose
Desk
Occasional chair
•Dining room suite
•Dinette set, wood or metal
Dining room table, wood or metal
•Dining room chair, wood or metal

SAMPLING FRAME FOR SELECTION OF THE CPI ITEM SAMPLE—Continued
Item
No.

CES
schedule No. 1

Item

HOUSING (212 items)

[No.

CES
schedule No. 1

HOUSING (212 items)

Furniture and floor coverings
EC-23 Furniture (31 items)—Con.
14
15.
16.
17
18
19.
20
21.
22.
23
24
25
26
27,
28
29
30,
31.

1-3-1
1-3-2
1-3-3 (pt)
1-3-3 (pt)
1-3-4
1-4
1-5 (pt)
1 1-5 (pt)__
1-5 (pt)
1-5 (pt)
I_42-4
1-42-5 (pt).
I_42-l-3._
1-42-5 (pt)
1-1-5 (pt)

Bedroom suite
Bed
Mattress, except nursery
! Spring
! Dresser, chest, or vanity
jPorch and beach furniture
Unpainted furniture, except kitchen
Card table
Folding chairs
JKitchen cabinet
>Kitchen furniture
Side chair
Bookcase
1
— Cots
Crib, nursery bed
Mattress, nursery
Chest, nursery
] Nursery chairs, play and feed tables, toilet
I seats, play pens, bathinettes, bassinettes,
nursery basket, baby carriages

T—1—5 (n+\

1-^3-5
1-75
1-74
1-77
1-76
1-78
1-79

EC-24 Floor coverings (7 items)
1.
2.
3
4.
5.
6.
7

1-6-7
1-8
1-9
1-14 (pt)
1-10-11
1-12
1-13
1-14 (pt)_
1-15

Rug, soft surface
Scatter rug, soft surface, all fibers
| Stairs and hall rug, soft surface, all fibers
Rug, hard surface
Scatter rug, hard surface
Tile
JRug pads

_

EC-25 Appliances (21 items)
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.

1-32
1-33
1-35 (pt)
1-35 (pt)
1-43
1-44
_
1-46
1-47
1-48
1-84
T_flK

_

(n*\

1-85 (pt)
I-S2
1-38
1-37
1-39
I-49_ - _
1-88 (pt)_
1-88 (pt)
1-34 1-36
1-83

_-

Refrigerator
Home freezer
- Cooking stove, free standing
Cooking stove, ovens built-in
Vacuum cleaner
Waxer, electric
Washing machine
_ Automatic clothes dryer
- Washer-dryer combination
Air conditioner, demountable
Sewing machine, cabinet
Sewing machine, portable
Room heaters
Electric toaster
Electric iron
[Electric coffee makers, frying pans, mixers, hot
[plate and other electrical kitchen equipment
Electrical fan, portable
Electrical fan, built-in
Dishwasher
Garbage disposal unit
Dehumidifier
EC-26 Other housefurnishings (32 items)

1.
2.
3.
4.
5.
6.
7

I-27-__
1-28
1-29
_
1-30
1-31-1
1-31-2
1-31-3-4.

8.
9.

1-40
1-41

10.
11.
12.

1-45-1
1-45-2
1-51

13.
14.
15.
16.
17.
18.

1-80
1-90 (pt)
1-90 (pt)
1-91
1-92
1-93 (pt)

EC-26 Other housefurnishings (32 items)
—Con.

-

Glasses
Dish sets
Dishes (separate pieces)
Serving pieces
Knives, forks, spoons, etc., sterling silver
Knives, forks, spoons, etc., silver plate
Knives, forks, spoons, etc., stainless steel and
other
Cooking utensils
Kitchen knives, forks, spoons, implements,
crockery and glassware
Brooms
Cleaning equipment other than brooms
Laundry boards, tubs, baskets, clothesline,
etc.
Bottles, nipples, sterilizers, bottle warmers
Floor lamp
Table lamp
Fireplace screen and equipment
Clocks
Scissors

19.
20.
21.
22.
23.
24.
25.
26.
27.
28.
1 29.
30.
31.
32.

1 Scales
— Thermos bottles
Lunch kits, etc.
Venetian blinds
| Window shades
Rods, etc.
Typewriters
Hand luggage, trunks, lockers
Lawn mowers
Hand tools
Power tools except lawn mower
Other garden equipment
Miscellaneous hardware and supplies
Fresh flowers and plants for the house

1-93 (pt)
1-93 (pt)
1-93 (pt)
1-94 (pt)
1-94 (pt)
1-94 (pt)
1-86
1-89
H-18
H-19 (pt)
H-19 (pt)
H-19 (pt)
H-20
H-16

Household operation (42 items)
EC-27 Housekeeping supplies (28 items)

1.

26-C

2.
i 3.
4.

14-C
16-C
15-C (pt)
15-C (pt)
! 6. 17-C
i 7. G-17
8. lg_20-C
9. 22-C
_
10. 24-C
11. 25-C
12. 29-C
13. 27-C (pt)
14. 27-C (pt)
I 15. 28-C
16. 30-C
17. 1-87
18. H-15 (pt)
19. 32-C
20. 34-C
21. 35-C
22. 36-C
23. 37-C
24. 38-C
25. 39-C
_
26. H-13 (pt)
1
27.
H-13
(pt)
:
1 | 28. H-13 ( p t ) - - -

_
_

—

___

Liquid household detergents other than laun­
dry
Liquid detergent, laundry
Detergent, solid form
Soap, bars
Soap, flakes or chips, granules or powder
Water softeners and conditioners
Ice
Laundry supplies except soaps and detergents
Air fresheners, air deodorizers
Floor wax
Insect sprays, powders, etc.
Sponges
Furniture polish
Metal polish
Scouring powder
Steel wool and other scouring pads
Electric light bulbs
Candles and matches, etc.
Aluminum foil
Paper napkins
Paper towels
Paper plates, etc.
Shelf paper
Toilet tissue
Wax paper
Stationery
Greeting cards
Pens and pencils and miscellaneous writing
supplies

EC-28 Housekeeping services (14 items)

1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.

H - l (pt) .__
H - l (pt)
H - l (pt)
H - 2 (pt)
H-5a
H-5c
H-5b
H-5d
_
H-6
H - 7 (pt)
H - 7 (pt)
H-8.-H-9-10
H-ll
H-12
G-14
G-15 (pt)

_

Cleaning sent out, rugs
Cleaning sent out, draperies, slipcovers
Cleaning sent out, blankets
1 Laundry sent out, flat work
JDomestic household help and babysitters
JHandyman, gardeners and grass cutters
Day nurseries, child care center service
Reupholstering
Furniture repair
Equipment repair and service charges
Moving expense, other freight and express
charges
Storage charges (except furs and apparel)
Postal charges
— jGarbage and trash collection

Apparel and upkeep (184 items)
Men's and boys' apparel (53 items)
EC-29 Men's apparel (30 items)

1. 1K - I I - 12

2.
3.
4.
5.

3
4
5

Overcoats, heavy
Topcoats, lightweight
Jackets, heavy, all fibers
Jackets, lightweight, all fibers
Sweaters, all fibers

See footnotes at end of table.




107

SAMPLING FBAMB FOR SELECTION OF THE CPI ITEM SAMPLE—Continued
Item
No.

CES
schedule No. 1

APPAREL AND U P K E E P (184 items)

Item
No.

CES
schedule No. 1

Men's and boys' apparel (53 items)
EC-29 Men's apparel (30 items)—Con.
6.
7.

K-II-6
8-13
14_
15.
16
19
20
21
28
27
31 (Pt)
29
30
23 (pt)
31 (pt)
24-25
22
23 (pt)
32
34 (pt)
35 (pt).
33
34 (pt)
35 (pt)
36
38 (pt)
23 (pt)
37
38 (pt)
39-41
49-51
54 (pt)
54 (pt)
53
52._
54 (pt)

8.
9.
10.
11.
12
13.
14.
15.
16.

17.
18.
19.
20.
21.
22.
23.
24.
25.
26.
27.
28.
29.
30.

Raincoats, all fibers
2-pc. or 3-pc. suits, winter and year round
weight, all fibers
Suits, tropical weight, all fibers
_ Sport coats, separate jackets, all fibers
Trousers, slacks, (except work), all fibers
Trousers, work
Overalls, coveralls
Dungarees
Shirts, work
i Shirts, dress, all fibers
J
Shirts, sport, woven and knit, all fibers
Uniforms and special work clothing
[Shorts, walking, Bermuda, etc.
[Undershorts, briefs
[Undershirts
Pajamas, nightshirt
Bathing trunks
►Bathrobes, lounging robes, all fibers
Socks including slipper socks, other hosiery
Hats and caps
Ties
Wallets, belts, etc.
Gloves, work
Gloves, dress, all fibers
Handkerchiefs
EC-30 Boys' apparel (23 items)

1.
2.
3.
4.
5.
6.
7.
8.
9
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.

K-IV-1
2_
9 (pt)
3
9 (pt)
4—
9 (pt)
5
10
11 (pt)
11 (pt)
12—
13—
14
17-181920-23
2629 (pt)
27
28
29
30
35 (pt)
|
31
32
35 (pt)
33
1
34
1
35 (pt)
i
36-38
1
45
7-8—
46
_
47
6.__
9 (pt)

1-2
3
4
5
6
7-1
7-2-7

[Jackets, heavy, all fibers
[Jackets, lightweight, all fibers
Sweaters, all fibers
[Suits, heavyweight
Suits, lightweight
Sport coats
Trousers, slacks, dress, all
fibers
Trousers, (except dress), all fibers
Dungarees, overalls
Shorts
Bathing trunks, play clothing, etc.
[Shirts, dress, all fibers
[Shirts, sports, woven and knit, all fibers
[Undershorts
[■Undershirts
[•Pajamas and bathrobes
Socks, slipper socks, other hosiery
Hats, caps and helmets
Snowsuits, ski suits, ski pants, leggings
Gloves
Accessories
[Raincoats, all fibers

Heavy winter coats
Lightweight coats, topper
! Fur coats, full or % length, fur jackets
Fur scarfs, stoles, muffs
Raincoats, all fibers
Jackets, heavyweight
Jackets, lightweight

See footnotes at end of table.

108



Women's and girls' apparel (70 items)
EC-31 Women's apparel (39 items)—Con.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
24.
25.
26.
27.
28.
29.
30.
31.
32.
33.
34.
35.
36.
S 37.
38.
39.

K-I- 8
9-1—_
9-2-7_
11
12
13
14—
25 (pt)
15
25 (pt)
16
17-117-2-7
18
19-20—
21
23-24
26 (pt)
26 (pt)
27
28
29
_
30-31
32
35 (pt)._
33
35 (pt)
34
35 (pt)
36
39 (pt)

37
38

39 (pt)
46
47
48
49
50 (pt)
50 (pt)

Sweaters, all fibers
Suits, heavyweight
-] Suits, lightweight
1 Street dresses, all fibers
Housedresses, all fibers
Dresses for formal or semiformal wear
jSkirts, jumpers, culottes

\[Blouses, shirts, all fibers
Aprons, smocks, brunch coats, dusters
| Slacks, heavyweight
Slacks, lightweight
__ Dungarees, blue jeans
Shorts, pedal pushers
Bathing suits
._ Uniforms and special work clothing
Slips
Petticoats
__ Garter belts
Girdles, corsets
Brassieres, all fibers
1 Panties, briefs, union suits, other underwear
[•Nightgowns
[Pajamas
1 I Robes, housecoats, and negligees

\[Stockings

-J [Anklets, socks, slipper socks
Hats
_ Gloves
_ — Handbags, purse
Umbrellas
Handkerchiefs and scarfs
Other accessories
EC-32 Girls' apparel (31 items)

[Overcoats, coat sets

Women's and girls' apparel (70 items)
EC-31 Women's apparel (39 items)
1. K - I 2. !
3. i
4.
5.
6.
7.

APPAREL AND U P K E E P (184 items)

Heavy winter coats, coat sets
1- K - I I I - 1 - 2 .
Lightweight coats, toppers
3
2.
4-5
Snowsuits, ski suits, leggings, ski pants
3.
Jackets, lightweight, all fibers
8
4.
7
Jackets,
heavy, all fibers
5.
9
Sweaters, all fibers
6.
1
0
1
7
Suits,
lightweight
and heavyweight
7.
Raincoats, all fibers
8.
6
12
School
ana
similar
dresses
9.
13
. . . Party dresses, all fibers
i 10.
)
Skirts
and
jumpers,
all fibers, pinafores,
14-15—
11.
26 (pt)
f smocks
16
12.
> Blouses, all fibers
26 (pt)_
'
13.
Tee shirts, polo shirts
17
18
I Slacks
14.
Overalls, dungarees, blue jeans
19
15. 1
Shorts
20
16.
21-22
Play suits, sun suits, special playclothes
17.
Bathing suits
18.
23
24-25
Uniforms and special clothes, except play
19.
27 (pt)
Slips
20.
1 Petticoats
27 (pt)
21.
28-29
Garter belts, brassieres
22. I
30
23.
[Panties, briefs
32 (pt)
24.
31
-__ {
[Undershirts
32 (pt)
25.
33
[Nightgowns, pajamas
35 (pt)
26.
34
\>Bathrobes, housecoats
35 (pt)
27.
36-39
Hosiery
28.
Hats
46
_
29.
47
Gloves
30.
48
Purses
31.
1 Accessories
49—
EC-33 Footwear (21 items)
12.
3.

K-II-42
43
44—

Street shoes, men's
Work shoes, safety shoes, men's
men's

SAMPLING FRAME FOR SELECTION OF THE CPI ITEM SAMPLE—Continued
CES
schedule No. 1

APPAREL AND U P K E E P (184 items)

Item
No.

CES
schedule No. 1

EC-33 Footwear (21 items)—Con.

K-I

45
40 (pt)_
40 (pt)_
41-42-.

K-IV 3940
41
K-III40
41-42K - I I 46.
K - I 43.
K-IV 42.
K-III43.
K - I I 47.
K - I 44
K - I 44
K - I V 43.
K-III44.

88:

Athletic and special sport shoes, men's
Shoes, street, dress, women's
Shoes, evening, women's
Casual shoes, sneakers, athletic and special
sport shoes, all fibers, women's
Shoes, leather upper including cowboy boots,
boys*
Sneakers, loafers, other casual shoes, boys'
Athletic and special sport shoes, boys'
Street and dress shoes, girls'
Sneakers, other casual shoes, loafers, athletic
and special sport shoes, all fibers, girls'
Houseslippers, men's
Houseslippers, women's
Houseslippers, boys'
Houseslippers, girls'
Rubbers, galoshes, boots, men's
Galoshes, boots, rubbers, women's
Galoshes, boots, plastic, women's
Galoshes, boots, rubbers, boys'
Galoshes, boots, rubbers, girls'

Other apparel (40 items)
EC-34 Other apparel commodities
(26 items)

KV

1-5—.
6
8
9-11-

12-1415
16-17..
18-1920
25
21
22
KV 7
23
24
26
K I I 55 (pt)
55 (pt)
K I V 48
KI
51 (pt).
51 (pt)
K I I I 50
L 1
6 (pt)
2
6 (pt)
3
5
6 (pt)_—.
9
10
11
_.
12
14-15

Infants' outerwear
Tee shirts, polo shirts, etc., infants'
Dresses, infants'
Rompers, suits, playsuits, sunsuits, overalls,
infants'
Slips, undershirts, vests, underpants, infants'
Waterproof pants, etc., infants'
Diapers
Sleeping garments, robes, wrappers, infants'
►Receiving blankets and layettes
Stockings, socks, infants*
Booties, shoes, infants'

APPAREL AND UPKEEP (184 items)
EC-35 Apparel services (14 items)—Con.

L 27.
28.
31.
29.

Dressmaker or tailor at home or outside
>Alterations, weaving, dyeing and repair
Upkeep and storage of fur
Watch repair, jewelry repair
TRANSPORTATION (34 items)
Private transportation (28 items)
Automobile and related goods (10 items)
EC-36 Auto purchase (2 items)

P-I-16(pt)__.
17c ( p t ) 18 (pt) —
SQ-IV-2a (pt)
P-I-16 (pt) —
17c ( p t ) „
18 ( p t ) . . .
SQ-IV-2a (pt)

Passenger cars, new

Passenger cars, used

EC-37 Gasoline and motor oil (2 items)
P-I-2728-

Gasoline
Motor oil
EC-38 Auto parts, etc. (6 items)

P-I-29
32
33~(ptJL
34
35
36

Antifreeze
New tires and tubes
Used and recapped tires
Batteries
Spark plugs
Other equipment and supplies
Automobile services (18 items)
EC-39 Auto repairs and maintenance
(13 items)

Caps, hoods, bonnets, bibs, mittens, jewelry,
etc., infants'
Jewelry, men's
Watches, men's
Jewelry and watches, boys'
Jewelry, women's
Watches, women's
Jewelry and watches, girls'

Lubrication
Washing
Changing oil filter
Miscellaneous minor repairs and services

Yard goods, 100% wool and wool blends

►Carburetor overhaul

Motor tune-up

Yard goods, cotton and cotton blends

►Replacing spark plugs, points and condenser

►Yard goods, man-made fibers

►Replacing muffler

Yarn
Thread
Patterns
Miscellaneous sewing materials
Other notions

►Clutch and transmission work
Brake adjustment, repair and services
^Front end alignment; steering adjustment;
wheel balancing
►Body work and frame repairs
Miscellaneous tire expense

EC-35 Apparel services (14 items)

EC-40 Other automobile expenses (5 items)
L-23a—.
25 (pt)
23b—.
25 (pt)
24
25 (pt)
17
18
21 (pt)
22 (pt)
19
20
21 (pt)
22 (pt)
H-4
2 (pt)..
2 (pt)..
3
L-26

Shoe repairs, men's and boys*
Shoe repairs, women's and girls'
Shoe shines and cleaning
•Drycleaning, men's and boys' clothing

Dry cleaning, women's and girls' clothing
Launderettes
Laundry, men's shirts
Laundry, except men's shirts and flatwork
Diaper service
H a t cleaning, blocking and repair

P-I-17b
P - I I I - 2 (pt).
P-I-44 (a-d)
U-9a
P-I-17c
47 ( p t ) .
49 ( p t ) .
P - I I I - 2 (pt).
P-I-17a
47 ( p t ) .
49 ( p t ) .
45a
45b
P - I I I - 2 (pt).
P-I-46
47 ( p t ) .
49 ( p t ) .
P - I I I - 1 (pt).

Auto insurance premiums

Auto financing charges

Registration and inspection fees

[Drivers' license fees

See footnotes at end of table.




109

SAMPLING FRAME FOR SELECTION OF THE CPI ITEM SAMPLE—Continued
TRANSPORTATION (34 items)

CES
schedule No. 1

Item
No.

CES
schedule No. 1

EC-40 Other automobile expenses (5 items)
—Con.
P-I-48
47 ( p t ) .
49 ( p t ) .
P - I I I - 2 (pt)

EC-45 Toilet goods (28 items)—Con.
3-C
5-C (pt)_,
5-C(pt)_
6-C
7-C ( p t ) .
7-C ( p t ) .
8-C ( p t ) .
8-C ( p t ) .
8-C ( p t ) .

Parking; garage rent; parking meters

EC-41 Public transportation (6 items)
P-II-1
3
Q - l l (pt)
SQ-IV-2d (pt)
P-II-2
Q - l l (pt)
SQ-IV-2d (pt)
P-III-3—....
Q-8
Q-10 (pt)
SQ-IV-2e (pt).
Q-5
10 (pt)
SQ-IV-2e (pt).

Taxi cabs
Rent of car
•Train fares (intercity)
Airplane and steamship fares

10 (pt)
SQ-IV-2e (pt)

l6"(pt)"™I
SQ-IV-2e (pt)

8-C ( p t ) .
12-C (pt)
12-C (pt)
12-C(
13-Cl
13-C
13-C I
13-C _ . ,
4-C (pt).
4^C ( p t ) . .
J-16d
9-C
10-C (pt).
10-C (pt).
2-C(pt)_.
33-C
11-C

Local transit

^i:::::::::

■Bus, intercity fares
-la
5
lb
lc
ld_.__
2 (pt).
2 (pt).
3
4
7

Medical care (15 items)
EC-42 Drugs and prescriptions (2 items)

Over the counter items and medical appliances
and supplies
►Prescriptions

284...
24_.
26.
27.

Family doctor, office and home visits
Dentists' fees
Physicians' fees for childbirth

1-16.
1-17
1-18 (pt) —
1-18 ( p t ) . . .
1-19
_
1-20 (pt)___
1-21 (pt) —
1-22
1-23-1 "Opt) "1-24
1-23-2 (pt)_
1-25
0-21
0-22 (pt)__

Pediatrician, office and home visits
Surgical fees
[Medical specialist other than pediatricians,
j home and office visits
Examination for eyeglasses and other eye care,
and eyeglasses
^Chiropractors and other practitioners' fees
[Nursing care
Laboratory tests, outside hospital
X-rays (excluding dental and eye) outside
hospital

0-22 (pt)
P - I I I - 6 (pt)_
0-23
0-25
0 - 8 (pt)
0-9

EC-44 Hospital services and health
insurance (2 items)
M-II-1.
2_
3_._.
6—.
25...
M-I- 2-b3-b.
7-b.

Hospital services

Health insurance
Personal care (37 items)
EC-45 Toilet goods (28 items)

1-C
2-C ( p t ) .

Toilet soap
Toothpaste and powder

See footnotes a t end of table.

no



1 Men's haircut and shaves
Haircut, boys'
Haircut, women's
Haircut, girls'
Press andcurl
Permanent waves
Shampoo and wave set
Hair coloring
Manicure
Reading and recreation (53 items)
Recreation (42 items)
EC-47 Recreational goods (29 items)

EC-43 Professional services (11 items)
M-II-16
32-37__
8
10 (pt)_
11
17
18
9
10 (pt)_
20
21
39-41-

Mouthwash and gargles
Shaving preparations
Men's toiletries
Face powder
Skin creams
Lotions, facial and hand
Shampoos
Men's hair tonics
Women's hairdressing, coloring or condi­
tioners
Spray hair fixatives
Deodorants
Toilet water and cologne, bath salts
Talcum and body powder
Lipstick, rouge
Compacts
Nail polish, enamel, remover
Manicure implements
Razor blades
Razors
Electric shavers and shaver repair
Home permanent kits
Hairbrushes, combs
Bobby pins, nets, etc.
Toothbrush
Cleansing tissue
Sanitary supplies
EC-46 Personal care services (9 items)

J
Health and Recreation (115 items)

M-II-43
45
46
47
48 (pt)
44
48 (pt)

HEALTH AND RECREATION
(115 items)

25.

0-19
O-20
0-24
P-III 4 (pt).
5 (pt).
0-13 ( p t ) . _ .
0-17
45-46-C

Durable:
TV sets and TV radio-phonograph combina­
tions
Radio
Phonographs
Tape recorders
Piano, organ
TV repair parts
Radio, phonograph repair parts
Hi-Fi components, kits, and parts
Phonograph records
Musical instruments
I Sheet music, music stands, cases, recording
I
tapes and rolls
Tricycles
Wagons, skates, sleds and other play
equipment
I Bicycles
Mechanical toys
Children's playground goods and equipment
> Sports equipment
Motorcycle or scooter
Boats, boat trailers
Outboard motor
Cameras, still
Cameras, movie
Projectors
Photographic equipment except cameras,
projectors, film
Nondurable:
Nondurable toys
Operating expenses of boats, motor­
cycle or scooter
Film
Pets and supplies except food, purchase
Pet foods

SAMPLING FRAME FOR SELECTION OF THE CPI ITEM SAMPLE—Continued
CES
schedule No. 1

HEALTH AND RECREATION
(115 items)

litem
No.

CES
schedule No. 1

EC-48 Recreational services (13 items)
1-20 (pt)
I-26a
1-26 (pt)
P - I I I - 4 - 6 (pt)0-1
Q-d-18 ( p t ) .
0-2
Q-d-18 (pt)_
0-3
Q-d-1 (pt)_.
0-4_
Q-d-1 (pt)._
0-6
0-7 (pt)_—
0-7(pt)_...
0 - 8 (pt)
0-13 (pt)_-_
1-21 ( p t ) - _ _

TV repair
Rental or repair of musical instruments
Rental of motorcycle or scooter, boats, boat
trailers, outboard motors, bicycles

0-33
0-38
0-42 ( p t ) . - Q-d-21 ( p t ) .
0-39
0-43 (pt)_-Q-d-21 ( p t ) .
Q-d-24(pt)O-40
0-42 (pt)_-_
Q-d-22 ( p t ) .
0-41
0-43 ( p t ) . _ .
Q-d-22 ( p t ) .
Q-d-24 ( p t ) .

0-440-35.

Music lessons, dancing lessons, etc.
Book rentals, library fees
Other goods and services (10 items)
EC-50 tobacco products (3 items)

[Drive-in movies
[Spectator sport
[Concerts, plays, and other admissions
Fees for indoor sports
Fees, golf
Fees, outdoor except golf
Hunting or fishing b* cense
Film developing
Radio, phonograph, etc., repair

Newspapers
Magazines
Hard bound books not school or technical
Pocket editions and other paper books, not
school or technical
Comic books

41-C
J 30 ( p t ) ~
42-C- —
J 30 ( p t ) 43-C-._._
44-C
J 30 (pt)-_

>Tuition and fees, other school levels
l-College and professional books
[School books other than college and proj fessionai
J

•Cigarettes
►Cigars
^Tobacco and other smoking supplies

EC-51 Alcoholic beverages (3 items)
265-C
J-30 ( p t ) 266-270-C
J-30 ( p t ) „
J-25d_—_
J-30 (pt)_.
Q-17

Beer and ale
[Whiskies and other alcoholic beverages
[Beer, cocktails, etc. away from home

EC-52 Financial and Miscellaneous
personal expenses (4 items)

^Tuition and fees, college and professional

1
Unless otherwise indicated, numbers indicate Section and line number of
Schedule
B Survey of Consumer Expenditures, (pt) = p a r t .
3
C following line number denotes Schedule C, the record of expenditures
for the 7 days preceding the date of the interview.




EC-49 Reading and education (11 items)
—Con.

[indoor movies

EC-49 Reading and education (11 items)
O-30_
0-310-340-32.

HEALTH AND RECREATION
(115 items)

R-l
_..
SQ-IV-2g (pt)
R-2
R-3
SQ-IV-2g (pt)
R-6
.__
SQ-IV-2g (pt)
SQ-iV~2g"(p"t)

\? Financing charges (excluding mortgage interest
and auto financing)
Bank service charges
> Funeral service
[Legal expenses

s
No CES number assigned, was included as a supplemental item between
items
13 and 14 in EC-14.
4
SQ refers to Standard Question.
8 To
T o be
b e priced
nriftfid only
o n l v in
i n cities where
w h e n applicable.

in

EXHIBIT C
BLS 2742

THIS REPORT
U.S. DEPARTMENT OF LABOR
City
WILL BE HELD
Bureau of Labor Statistics
Block No.
IN CONFIDENCE
Washington 25, D.C.
From BLS 2549:
Budget Bureau No. 44-6112
SURVEY OF WHERE GOODS ARE BOUGHT
Page
Line
Approval expires: 6-30-62
Please answer the questionnaire in the following way for each item listed:
If you did not buy the item in the past IB months—Check No in column (A).
If you bought the item entirely from a mail order catalog—Check Yes in column (A); check All in column (B).
If you made any purchases of the item from stores other than from a mail order catalog—Check Yes in column (A); check the answer which applies in column
(B); write in the names of the stores in column (C) and the names of the cities in column (D). (If you bought in several stores and spent about the
same amount in each, give the names and locations of up to three stores for each item.)
(A)

(B)

(C)

(D)

In the past 12 months did you buy . . ?

What part of the amount
you spent in the past 12
months for these items
was ordered from a mail
order catalog? (Check
one)

In what store or stores, other than from a
mail order catalog, aid you usually buy
these items in the past 12 months?

In what city or town is this
store located?

MEN'S (16 years and older):
Suits and coats

All
More than Y2
Between H and H
Less than M
None

D
D
D
D
D

1
2
3
4
5

Yes D
12 No O 0

AU
More than *4
Between H and }$
Less than 3€
None

D
D
D
D
D

1
2
3
4
5

Yes D
13 No D O

All
More than }4
Between M and ]4
Less than %
None

D
D
D
D
D

1
2
3
4
5

Yes D
14 No D O

AU
More than M
Between M and K
Less than M
None

D
D
D
D
D

1
2
3
4
5

Yes D
21 No D O

AU
More than Y2
Between }i and %
Less than H
None

D
D
D
D
D

1
2
3
4
5

Yes D
22 No D O

AU
More than )4
Between M and H
Less than H
None

D 1
Q2
D 3
Q4
D 5

Yes □
23 No D 0

AU
More than H
Between M and }4
Less than M
None

D 1
D 2
O 3
Q4
D 5

Yes D
24 No D O

AU
More than %
Between H and %
Less than H
None

D 1
Q2
D 3
D 4
D 5

Yes O
25 No D O

AU
More than Y2
Between % and H
Less than M
None

D
Q
O
D
D

1
2
3
4
5

Yes D
26 No D O

AU
X
More than
A
Between H and H
Less than %
None

D
D
D
D
a

1
2
3
4
5

Yes D
27 No D O

AU
More than K
Between M and %
Less than K
None

D
D
D
D
D

1
2
3
4
5

CHILDREN'S:
Clothing (except shoes) for boys
2 to 16 years old
Yes D
31 No D O

AU
More than %
Between % and H
Less than %
None

D
D
D
D
D

1
2
3
4
5

All
More than ££
Between H and H
Less than M
None

D
D
D
D
D

1
2
3
4
5

Yes Q
11 No D O
Shirts, socks, underwear, hats, and
other furnishings

Work clothes

Shoes

WOMEN'S (16 years and older):
Suits and coats

Dresses

Skirts and blouses

Hats, gloves and accessories

Underwear and nightwear

Stockings

Shoes

Clothing (except shoes) for girls
2 to 16 years old
Yes D
32 No D O

Please complete other side

112



(A)

(B)

(C)

(D)

In the past 12 months did you buy . . ?

What part of the amount
you spent in the past 12
months for these items
was ordered from a mail
order catalog? (Check
one)

In what store or stores, other than from a
mail order catalog, did you usually buy
these items in the past 12 months?

In what city or town is this
store located?

All
More than ^
Between )i and K
Less than %
None

O
O
O
O
D

1
2
3
4
5

All
More than %
Between yi and yi
Less than %
None

D
D
D
D
D

1
2
3
4
5

All
More than ^
Between M and ^
Less than yi
None

O
D
D
O
O

1
2
3
4
5

Yes O
42 No D O

All
More than %
Between \i and y2
Less than }&
None

D
O
O
O
O

1
2
3
4
5

Yes O
43 No D O

All
More than ^
Between M and >£
Less than %
None

D
D
O
O
O

1
2
3
4
5

Yes O
44 No O 0

All
More than %
Between % and %.
Less than \&
None

D
D
O
O
D

1
2
3
4
5

Yes O
45 No D O

All
More than )4
Between H and M
Less than M
None

D
O
D
O
D

1
2
3
4
5

All
More than %
Between \i and )4
Less than %
None

O
O
O
D
D

1
2
3
4
5

Boys' and girls' shoes
Yes O
33 No D O
Clothing for infants less than
2 years old
Yes O
34 No D O
HOUSEFURNISHINGS:
Yard goods for clothing, slipcovers,
draperies, etc.
Yes O
41 No O
Sheets, towels, curtains and
other household textiles

Rugs and carpets

Furniture

Television and radio

Major appliances (such as
washer, refrigerator or stove)
Yes O
46 No D O

Yes O
51 No D O

All
D 1
More than ^$
O 2
Between )4> and % O 3
Less than yi
O 4
None
O 5

Yes O
52 No D O

All
More than H
Between % and %
Less than yi
None

D
D
O
O
D

1
2
3
4
5

Yes D
53 No D O

All
More than )4
Between yi and yi
Less than yi
None

D
D
O
D
D

1
2
3
4
5

Auto repairs

Yes O 5
54 No O 0

xxxxxxxxxxxxxx

New automobile

Yes O 5
55 No D O

xxxxxxxxxxxxxx

Laundry service

Yes D 5
56 No O 0

xxxxxxxxxxxxxx

Dry cleaning

Yes O 5
57 No D O

xxxxxxxxxxxxxx

Automobile insurance

Yes O 5
61 No D O

Health insurance

Yes D 5
62 No D O

OTHER I T E M S AND SERVICES:
Toys and sports equipment

Automobile tires

Auto parts and accessories

What is the name of the insurance company, NOT the agency from which you bought the policy?




113

Bibliography
Description of Consumer Price Index—General
The Consumer Price Index (Revised January 196b): A Short Description. U.S.
Department of Labor, Bureau of Labor Statistics, September 1964. 12 pp.
"Consumer Prices." Reprint of Chapter 10, BLS Bulletin U58, BLS Handbook of
Methods for Surveys and Studies, U.S. Department of Labor, Bureau of Labor
Statistics, 1965. (In revision of BLS Bulletin 1168, 1955.)
The Consumer Price Index: Pricing and Calculation Procedures, unpublished paper by
Doris P. Rothwell, U.S. Department of Labor, Bureau of Labor Statistics, March
1964. 22 pp.
"The Statistical Structure of the Revised CPI," by Sidney A. Jaffe, Monthly Labor
Review, August 1964, pp. 916-924.
Computation of Cost-of-Living Indexes in Developing Countries. U.S. Department of
Labor, Bureau of Labor Statistics. (BLS Report No. 283,1964.)
Sampling Aspects of the Revised CPI, unpublished paper by Marvin Wilkerson, U.S.
Department of Labor, Bureau of Labor Statistics, October 1964. 33 pp.
Measurement of Sampling Error in the Consumer Price Index: First Results, un­
published paper by Marvin Wilkerson, U.S. Department of Labor, Bureau of
Labor Statistics, December 1964. 16 pp.
"The Revised City Sample for the Consumer Price Index/' by Marvin Wilkerson,
Monthly Labor Review, October 1960, pp. 1078-1083. Reprint No. 2352.
Interim Adjustment of Consumers9 Price Index: Correction of New Unit Bias in
Rent Component of Consumers9 Price Index and Relative Importance of Items.
(Bulletin 1039,1952.)
Study of Consumer Expenditures, Incomes and Savings—Methodology of the Survey
of Consumer Expenditures in 1950, by Helen Humes Lamale, Wharton School of
Finance, University of Pennsylvania, Philadelphia, Pa., 1959.
Cost of Living and Retail Prices in the United States (1890-1903). U.S. Bureau of
Labor. (Bulletin 54,1904.)
Cost of Living and Retail Prices of Food, in 18th Annual Report of the Commissioner
of Labor, 1903.
Retail Prices of Food, 1890 to 190k. U.S. Bureau of Labor. (Bulletin 59, 1905.)
"Labor and the War," Monthly Review of the U.S. Bureau of Labor Statistics,
March 1918, pp. 67-76.
"Cost of Living in the District of Columbia," Monthly Labor Review, June 1919, p.
117.
Cost of Living in the United States. U.S. Department of Labor, Bureau of Labor
Statistics. (Bulletin 357, 1924.)
"Changes in Cost of Living from September 15 to November 15, 1940," Monthly
Labor Review, January 1941, pp. 146-149.
"Revision of Index of Cost of Goods Purchased by Wage Earners and Lower Sal­
aried Workers," by Faith M. Williams, Margaret H. Hogg, and Ewan Clague,
Monthly Labor Review, September 1935, pp. 819-837.
114



Retail Prices of Food, 1923-36. U.S. Department of Labor, Bureau of Labor Statis­
tics. (Bulletin 635, 1938.)
"Bureau of Labor Statistics, Cost-of-Living Index in Wartime," Monthly Labor Review, July 1943, pp. 82-95. Reprint No. 1545.
"Revision of the Consumers* Price Index," Monthly Labor Review, July 1950, pp.
129-132. Reprint No. 2003.
"Selection of Cities for Consumer Expenditures Survey, 1950," by Marvin Kogan,
Monthly Labor Review, April 1951, pp. 430-436. Reprint No. 2060.
Consumer Price Index Numbers—Sampling Problems in Prices, by A. Basu, Indian
Labour Journal, Delhi, June 1960, pp. 582-588.
"Sampling Considerations in the Construction of Price Indexes with Particular Ref­
erence to the United States Consumer Price Index," by Philip J. McCarthy, in
Government Price Statistics-Hearings, Subcommittee on Economic Statistics of
the Joint Economic Committee, 87th Cong., 1st sess., Part 1, January 24, 1961,
pp. 197-232.
"Controlled Selection—A Technique of Probability Sampling," by Roe Goodman and
Leslie Kish, Journal of the American Statistical Association, September 1950, pp.
350-372.
Relative Importance of Components
"Relative Importance of CPI Items," by Gloria P. Green, Monthly Labor Review,
November 1965, pp. 1346-1349.
Consumer Price Index (New Series), Relative Importance of Major Groups, Subgroups and Individual Items, December 1963 and Comparison with Old Series.
Mimeographed report. U.S. Department of Labor, Bureau of Labor Statistics,
1964. 16 pp.
Relative Importance of Consumer Price Index Components, December 1962 and
1957-59 Average, and Selected Prior Periods for Major Groups. U.S. Department
of Labor, Bureau of Labor Statistics, 1963. 4 pp.
"Relative Importance of CPI Components," by Chester V. McKenzie, Monthly Labor
Review, November 1961, pp. 1233-1236. Reprint No. 2377.
"Relative Importance of Items in the CPI," by Marsha Froeder and Carlyle P. Stallings, Monthly Labor Review, August 1954, pp. 891-896. Reprint No. 2146.
Use of CPI in Wage Adjustments
"The Use of Price Indexes in Escalator Contracts," by Francis S. Cunningham,
Monthly Labor Review, August 1963, pp. 948-952. Reprint No. 2424.
Escalator Clauses from Selected Collective Bargaining Agreements. U.S. Depart­
ment of Labor, Bureau of Labor Statistics, October 1962. 14 pp.
"The Growth, Status, and Implications of Wage Escalation," by H. M. Douty, Monthly Labor Review, February 1953, pp. 126-129. Reprint No. 2095.
"Wage Escalators and the Adjusted CPI," by Lucy M. Kramer and James Nix,
Monthly Labor Review, May 1951, pp. 509-513. Reprint No. 2034.
"Deferred Increases Due in 1965 and Wage Escalation," by George Ruben, Monthly
Labor Review, December 1964, pp. 1381-1384.
Deferred Wage Increases and Escalator Clauses, 1952-63. U.S. Department of Labor,
Bureau of Labor Statistics, February 1963. (BLS Report No. 235,1963.)
Cost of Living Wage Adjustments in Collective Bargaining. Mimeographed report.
U.S. Department of Labor, Bureau of Labor Statistics, September 1951 (Revised).
21 pp.




The Use of Cost-of-Living Figures in Wage Adjustment. U.S. Department of Labor,
Bureau of Labor Statistics. (Bulletin 369,1925.)
Consumer Price Index—Specific Components
"The Calculation of Average Retail Food Prices," by Doris P. Rothwell, Monthly
Labor Review, January 1965, pp. 61-66.
"Rent Component of the Consumers' Price Index, Part I—Concept and Measure­
ment, Part II—Methodology of Measurement," by Helen Humes and Bruno Schiro,
Monthly Labor Review, December 1948, pp. 631-637, and January 1949, pp. 60-68.
Reprint No. 1947.
"Housing Costs in the Consumer Price Index, Part I—Concept and the Expenditure
Basis, Part II—Pricing Procedures," by Helen Humes Lamale, Monthly Labor
Review, February 1956, pp. 189-196 and April 1956, pp. 442-446. Reprint No.
2188.
"Estimate of New Unit Bias in CPI Rent Index," by Ethel D. Hoover and Bruno
Schiro, Monthly Labor Review, July 1949, pp. 44-49.
"Correction of New Unit Bias in the Rent Component of CPI," by George Johnson
and Bruno Schiro, Monthly Labor Review, April 1951, pp. 437-444.
The Home State and Migration of American College Students, Fall 1958. American
Association of Collegiate Registrars and Admissions Officers, Committee on Re­
search and Services, Ohio University, Athens, Ohio, 1959.
"Introductory Prices of 1966 Automobile Models," by Margaret S. Stotz, Monthly
Labor Review, February 1966.
'
"Compact Cars in the Consumer Price Index," by Olga A. Larsgaard and Louise J.
Mack, Monthly Labor Review, May 1961, pp. 519-523. Reprint No. 2368.
"Automobile Prices in the Consumer Price Index," by Louise J. Mack, Monthly Labor
Review, November 1955, pp. 1269-1273. Reprint No. 2179.
"Health Insurance in the Revised CPI, by James C. Daugherty, Monthly Labor Review, November 1964, pp. 1299-1300.
Factors Affecting the CPI
"Use of Varying Seasonal Weights in Price Index Construction," by Doris P. Roth­
well, Journal of the American Statistical Association, March 1958, pp. 66-77.
Seasonal Factors, Consumer Price Index: Selected Series, June 1953-May 1961. U.S.
Department of Labor, Bureau of Labor Statistics. (Bulletin 1366, 1963.)
"The CPI and Problems of Quality Change," by Ethel D. Hoover, Monthly Labor
Review, November 1961, pp. 1175-1185. Reprint No. 2378.
"Taxes and the Consumers' Price Index," Monthly Labor Review, January 1953, pp.
53-57. Reprint No. 2090.
Inquiries into CPI
Price Statistics of the Federal Government, Review, Appraisal, and Recommendations, National Bureau of Economic Research, Report No. 73, General Series,
New York, 1961.
Government Price Statistics—Hearings, Subcommittee on Economic Statistics of
the Joint Economic Committee, 87th Cong., 1st sess., Part I, Washington, January
24,1961, Part II, May 1, 2, 3, 4, and 5,1961.
Consumers' Price Index. Report of a Special Subcommittee of the Committee on
Education and Labor, House of Representatives, 82d. Cong., 1st. sess., Report No.
2, Washington, 1951.




The Consumers9 Price Index—Report of the Joint Committee on the Economic Report on the Consumers9 Price Index of the United States Bureau of Labor Statistics, 80th Cong., 2d sess., 1949.
Report of the President's Committee on the Cost of Living. Office of Economic Stabi­
lization, 1945.
"An Appraisal of the U.S. Bureau of Labor Statistics Cost-of-Living Index," by a
Special Committee of the American Statistical Association, Journal of the American Statistical Association, December 1943, pp. 387-405.
Staff Report on Employment, Growth and Price Levels—Prepared for Considera­
tion by the Joint Economic Committee, 86th Cong., 1st sess., 1959.
Consumer Price Movements
Prices, 1964. U.S. Department of Labor, Bureau of Labor Statistics, (BLS Report
No. 291, 1965.)
Prices: A Chartbook, 1953-62. U.S. Department of Labor, Bureau of Labor Statis­
tics, (Bulletin 1351, 1963). Supplement, September 1963. (Bulletin 1351-1.)
Consumer Prices in the United States, 1953-58: Price Trends and Indexes. U.S. De­
partment of Labor, Bureau of Labor Statistics. (Bulletin 1256,1959.)
Consumer Prices in the United States, 194*9-52. U.S. Department of Labor, Bureau
of Labor Statistics. (Bulletin 1165, 1954.)
Consumers9 Prices in the United States, 1942-48. U.S. Department of Labor, Bu­
reau of Labor Statistics. (Bulletin 966, 1949.)
Changes in Cost of Living in Large Cities in the United States, 1913-41. U.S. De­
partment of Labor, Bureau of Labor Statistics. (Bulletin 699,1941.)
Historical Statistics of the United States, Colonial Times to 1957. A Statistical Ab­
stract Supplement. U.S. Department of Commerce, Bureau of the Census, 1960.
Continuation to 1962 and Revision. 1963.
Historical Statistics of the United States, 1789-1945. U.S. Department of Com­
merce, Bureau of the Census, 1949. Continuation to 1952 of Historical Statistics
of the United States. 1954.
Average Retail Prices, 1955. U.S. Department of Labor, Bureau of Labor Statistics.
(Bulletin 1197, 1956.)
Average Retail Prices: Collection and Calculation Techniques and Problems. U.S.
Department of Labor, Bureau of Labor Statistics. (Bulletin 1182, 1955.)
Index Series
Consumer Price Index. Historical Series (1957-59 = 100) for U.S. and 23 Large
Cities. Series A, All items; Series B, Food; Series B-l, Food at home; Series B-2,
Cereals and bakery products; Series B-3, Meats, poultry, and fish; Series B-4,
Dairy products; Series B-5, Fruits and vegetables; Series B-6, Other foods at
home; Series B-7, Food away from home; Series C, Housing; Series C-l, Rent;
Series C-2, Gas and electricity; Series C~3,~Fuel oil and coal; Series C-4, Housefurnishings; Series C-6, Shelter; Series C-7, Homeownership; Series C-8, Fuel
and utilities; Series C-9, Household furnishings and operation; Series D-5, Ap­
parel and upkeep; Series D-l, Men's and boys' clothing; Series D-2, Women's
and girls' apparel; Series D-3, Footwear; Series E, Transportation; Series E-l,
Private transportation; Series E-2, Public transportation; Series F, Medical care;
Series G, Personal care; Series H, Reading and recreation; Series I, Other goods
and services; Series J, Health and recreation. U.S. Department of Labor, Bureau
of Labor Statistics.




Consumer Price Index. Historical Series, Special Groups (1957-59 = 100), U.S.
average only. Series A-2, All items less food; Series A-3, All items less shelter;
Series K, All commodities; Series K-l, All Commodities less food; Series K-2,
Durable commodities; Series K-3, New cars; Series K-4, Used cars; Series K-5,
Household durables; Series L, Nondurable commodities; Series L-l, Nondurable
commodities less food; Series L-2, Apparel commodities; Series L-3, Nondurables less food and apparel; Series L-4, Apparel commodities less footwear; Series
M, All services; Series M-l, All services less rent; Series M-2, Household services
less rent; Series M-3, Transportation services; Series M-4, Medical care services;
Series M-5, Other services. U.S. Department of Labor, Bureau of Labor Statistics.
Consumer Price Index. Monthly release and detailed report. U.S. Department of
Labor, Bureau of Labor Statistics.
Retail Prices and Indexes of Fuels and Electricity. Monthly report. U.S. Depart­
ment of Labor, Bureau of Labor Statistics.
Consumer Price Indexes for Selected Items and Groups. Quarterly report. U.S.
Department of Labor, Bureau of Statistics.
Consumer Price Index. Price Indexes for Selected Items and Groups. Annual aver­
ages, 1935-61, quarterly indexes, March 1947-June 1964. (Discontinued June
1964.) U.S. Department of Labor, Bureau of Labor Statistics.
Estimated Retail Food Prices by Cities. Monthly report. U.S. Department of Labor,
Bureau of Labor Statistics.
&

118



U.S. GOVERNMENT PRINTING OFFICE: 1«6—22M73