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