The full text on this page is automatically extracted from the file linked above and may contain errors and inconsistencies.
L 3l . "S * 1 1 SOUTHWEST MISSOURI STATE COLLEGE LIBRARY 5 - d e p o s it o r y COpv MAY 2 5 1972 BLS Handbook of Methods for Surveys and Studies BULLETIN 1711 U. S. DEPARTMENT OF LABOR J.D. Hodgson, Secretary BUREAU OF LABOR STATISTICS Geoffrey H. Moore, Commissioner 1971 For sale by the Superintendent of Documents, U.S. Government Printing Office Washington, D.C. 20402 - Price $2.00 Stock Number 2901-0659 Preface The Bureau of Labor Statistics is essentially a factfinding agency, the functions of which are the collection, tabulation, evaluation, and publication of a wide range of economic and statistical information. A part of the BLS’ responsibility is to make available to the users of its data information on the scope of its pro grams and the methods employed in the surveys and studies that it conducts. The users of BLS information encompass a broad spectrum of the American society— administrators of businesses, large and small; workers; union officials; academicians; technicians; government policymakers and administrators at Fed eral, State, and local levels; and others o f the general public. The methods employed by the Bureau in collecting, analyzing, and presenting its data are often highly technical and complex. A description of these methods at a level of technical detail which would meet the requirements of all users might bore the technician or bewilder the nontechnical user. The Handbook of Methods is designed to serve the broad middle range of users. The technician may wish to seek more detailed treatment of his field of interest, and the casual user may find simpler explanations more helpful. Sources of addi tional information, some more technical and some more popular, are listed under “ Technical References” at the end of most chapters. The chapters for the Handbook were prepared in the program offices of the Bu reau and published by the Office of Publications, Division of Special Publications, under the direction of Tommy C. Ishee with the assistance of Scott Wirtzman. hi Contents Page Introduction ______________________________ __ ___________________ ___ 1 Chapter: 1. Labor force, employment, and unemployment__________________ 2. Employment, hours, and earnings ____________________________ 3. Job vacancies and labor turnover_________________ 7 17 35 Current Employment Analysis Manpower Structure and Trends 4. 5. 6. 7. Employment of scientific and technical personnel_______________ Occupational outlook _________________________________________ Projections of the labor fo r c e __________________________________ Industry-occupational matrix __________________________________ 43 47 53 55 Prices and living conditions 8. 9. 10. 11. 12. 13. Consumer expenditures and incom e____________________________ Family budgets________________________________________________ Consumer prices ______________________________________________ Wholesale prices __________ Industry-sector indexes _______________________________________ Spot market prices ____________________________________________ 59 69 75 97 113 119 Wages and industrial relations 14. 15. 16. 17. 18. 19. 20. 21. 22. Occupational pay and supplementary benefits __________________ Employee earnings and hours frequency distributions__________ Union wage rates --------Current wage developments ___________________________________ Employer expenditures for employee compensation _____________ Work stoppages___________ _________________ ____ ___ _______— Collective bargaining agreements -------------- ---------------- ------------Union membership __________________________________ __ ______ Annual earnings and employment patterns of private nonagricultural workers _________________ ,------------------------- ---------------23. Measuring collective bargaining settlements ___________________ 24. Wage chronologies and salary trend reports ------------------------------ 123 137 147 153 161 171 181 187 195 201 209 Productivity and technology 25. 26. 27. 28. Output per man-hour: Private sector______ ___________________ 213 Output per man-hour measures: Industries_________________ -__ 219 Technological change _____ ________________________ __ ______ ______________ Construction labor requirements ______________ ___ ____________ 235 227 Occupational safety and health 29. Occupational safety and health statistics ___ ___ _______________ 239 v Contents— Continued P ag * Economic Trends and Labor Conditions 30. Foreign labor conditions, international comparisons, and trade research __ 31. Economic growth studies _____________________________________ Appendix: A. The BLS seasonal factor method _______ B. Industrial classification________________________________________ C. Geographic classification ______________________________________ TI 241 245 247 255 257 BLS Handbook of Methods for Surveys and Studies Introduction The country is hungry for information; everything of a statistical character, or even of a statistical appear ance is taken up with an eagerness that is almost pathetic; the community have not yet learned to be half skeptical and critical enough in respect to such state ments. With these words Gen. Francis A. Walker greeted Carroll Wright in 1873, as Dr. Wright assumed charge of the Massachusetts Bureau of Statistics and Labor. And when as U.S. Commissioner of Labor, he issued his first an nual report in March 1886, Carroll Wright established the policy of explaining his sta tistical methods to his readers and of seeking to avoid misinterpretation of the figures pre sented. For example, he said: In stating the facts as they have been found by the agents of the Bureau, many terms are used which are capable of varied application— some even are of doubt ful meaning when considered metaphysically, but all such terms are used in this report in their common acceptation; as, for instance, the term “ overproduction” is used to indicate that condition of a locality, state, or country when more goods have been produced than are sufficient to meet the ordinary demand . . . .* In the same report there are statements on testing the validity of figures (p. 141), prob lems of nonresponse (p. 90), and restrictions on coverage (headnotes to tables). Warnings as to inadequacies of available information occur frequently. During the 86 years which followed the initial report, the definitions, methods, and limitations of the data published by the Bureau of Labor and its successor, the Bureau of Labor Statistics, have been ex plained again and again. The reason for this is not merely to make the readers “ skeptical,” “ critical,” and aware of the known limitations of the statistics, but also to instruct them in the proper use of the information and to assure them that proper standards have been ob served. Furthermore, whereas one might ex pect to breed a certain amount of doubt about a statistical survey by revealing its lack of perfection, frankness about unavoidable de fects more often has the opposite effect, and public confidence in the work is reinforced in the process. The most grave doubts arise when things crying for explanation are not ex plained. The Committee on Government Statistics and Information Services emphasized 30 years ago that the Central Statistical Board “ should urge on each collecting agency the importance of publication by agencies of frank appraisals of the extent of noncomparability, incompleteness, and inaccuracy which may be inherent in their reports at any given time. This candid policy should enhance and not diminish the scientific prestige of the collecting agency.” 1 2 Full understanding of the statistical series and studies of the Bureau of Labor Statistics is not to be gained solely from detailed descrip tions of them, but also from appraisal of the philosophy and approach of the Bureau and of the manner in which it functions. Background The history of the Federal Bureau of Labor extends back to 1884. Before the creation of the cabinet post of Secretary of Labor, the Bureau for a time was known as the Department of 1 See Industrial Depressions, The First Annual Report of the Commissioner of Labor, March 1886, pp. 15-16. 2 Committee on Government Statistics and Information Services, Government Statistics, April 1937, p. 53. Recommen dation #16 of the Committee states: “ Continued criticism and analysis should be made of (a) statistical definitions, specifi cations, and classification; (b) coverage of supposedly com plete surveys and of samples used for current reporting; (c) timing of periodic surveys and current reports, component items and weighting systems of index numbers; and (f) methods and practices in the presentation of data. . . . Frank appraisals of comparability, completeness, and accuracy should be published.” (pp. 48-49). 1 2 BLS HANDBOOK OF METHODS Labor. From the Bureau’s beginnings in the administration of President Arthur until it became a part of a cabinet department under President Wilson, it accumulated nearly 3 dec ades of experience in collecting, interpreting, and presenting facts crucial to the welfare of workers. Details of early Bureau history and of developments of later years are to be found elsewhere.3 However, in describing the various statistical programs4 in this volume, some of the events which led to the development of particular statistical measures are recounted. Against this historical background emerges the philosophy and posture of the Bureau of Labor Statistics as the impartial observer and in terpreter of trends important to the welfare of workers. Voluntary reporting and the pre serving of the confidential nature of reported data are important characteristics of BLS programs. Voluntary Reporting and Confidentiality In the 86-year history of the Bureau’s opera tion it has asked hundreds of thousands of firms and individuals to provide information closely related to their daily affairs and their personal lives. To some of them who have sup plied the desired information, the Bureau has gone back a second time, a third time, and perhaps dozens of times, for later information on the same subject or for new types of infor mation. The response has been remarkable in its generosity, even when it is remembered that a sustained effort has been made to keep the requests reasonable. In no small measure, the cooperation received is due to the great care taken to avoid identifying the firm or the per son supplying the information. The fact that Bureau employees pledge themselves to protect these data is less important than that they have a deep understanding of the adverse longrun consequences of even a single lapse. They are aware of the greater worth, in terms of pure statistical validity, of the information provided voluntarily as compared with that supplied under legal sanctions. The only inducement employed is to tell the respondent that his contribution is important to the ultimate suc cess of the survey and that he may find the survey results useful in his own pursuits. The policy of not identifying the respondent is im plemented by combining the data reported by the different sources and issuing the findings in summary form. Another assurance given the respondent is that his report will be used for statistical pur poses only. Attempts to “ break” this policy, by organizations or individuals who wanted access to data in our possession and were willing to go to the courts to secure it, have been success fully resisted.5 Another form this problem * takes is the case in which an administrative agency of government seeks court action to compel a company to release its file copy of information provided in confidence to a sta tistical agency.0 While it cannot be proved that these policies result in more reliable statistics, Bureau Com missioners and their staffs over the years have been convinced from their experience that it is so. It is notable that some other Federal agen cies (especially the Bureau of the Census), well-equipped with authority to compel the submittal of certain reports, rarely if ever in voke this power. Rather, they choose to rely upon forms of suasion similar to our own. The Bureau of Labor Statistics, while its functions as a statistical agency are prescribed by law,7 has always relied upon voluntary cooperation of respondents in collecting information. 3 See, for example, the Secretary of Labor’s First Annual Report, 1913, for history 1884-1913; and U.S. Department of Labor, The Anvil and the Plow, 1964; pages 4-5, 19-20, 49-51 (1913—30); pp. 63-64, 87-90, 117-119, 136-137, and 155 (1930-48); pp. 172-173, 187, 206-207, and 230-234 (1949-63). 4 For a handy reference to BLS programs, showing their principal characteristics, see U.S. Department of Labor, Bureau of Labor Statistics, Major BLS Programs (issued annually). 5 For example, see Norweigan Nitrogen Company v. United States, 288 U.S. 294; United States v. Kohler, 13 Fed. Rules Serv. 33.333 (E.D., Pa. 1949); Hawes v. Walsh, 277 Fed. 569, the Court of Appeals for the District of Columbia. In all of these cases, the courts sustained the policy of protecting the con fidentiality of information given voluntarily and in confidence to an agency of the Federal Government. 0 See Supreme Court of the United States, St. Regis Paper Company, Petitioner, v. United States, No. 47, October term, 1961. 7 Excerpts from 29 U.S.C. 1, acts of June 27, 1884, ch. 127, 23 Stat. 60; June 13, 1888, ch. 389, § 1, 25 Stat. 182; Feb. 14, 1903, ch. 552, § 4, 32 Stat. 826; Mar. 18, 1904, ch. 716, 33 Stat. 136; Mar. 4, 1913, ch. 141, § 3, 37 Stat. 737. INTRODUCTION BLS Role, Staff, and Organization Among Federal agencies collecting and issu ing statistics, the Bureau of Labor Statistics has been termed a “ general-purpose statistical agency.” 8 The Bureau’s figures are prepared to serve the needs of business, labor, Congress, the general public, and especially the admin istrative and executive agencies, for infor mation on economic and social trends and situations. While the data serve some admin istrative purposes, they are free from the constraints which sometimes result from the close ties normally existing between operations and operational statistics. BLS statistics are often quite specialized, yet they meet general economic and social data requirements. As the needs of users are likely to differ from each other and also over time, no statistic is ideal for all. This makes it important that the characteristics of the measures and their pos sible limitations be well understood. The Bureau plays a larger and more signifi cant role than merely publishing generalpurpose statistics. Its activities frequently in fluence, and sometimes are crucial to, the de termining and shaping of public policy. The Bureau’s experts have the keen understanding of economic and social forces which results from intensive and continuous involvement in factfinding and in the painstaking analysis of data. Staff The Bureau’s work extends beyond the initial collection and processing of data. Over the years, it has developed a staff of professional analysts, trained in the disciplines of eco nomics and other social sciences, to search out the implications of survey findings for the welfare of workers and to present them as cogently and as promptly as possible in written and oral form. How successfully this can be accomplished depends greatly upon the com petence of the analysts and of their supporting personnel. In BLS, analytical and statistical work is performed by economists, statisticians, and s Bureau of the Budget, Statistical Services of the United States Government (revised edition), 1963, p. 7. 3 mathematical-statisticians with the aid of an experienced corps of programers, systems analysts, and other professionals, as well as statistical clerks. For analytical work, eco nomists at even the lowest grade level must meet Civil Service Commission requirements roughly equivalent to a college major in eco nomics. There are comparable requirements for other professionals. The greatest effort is made to locate the best of graduating seniors, Masters, Ph.D.’s, and those with research ex perience, in the colleges, State agencies, busi ness organizations, and labor unions. The Bureau provides training needed for on-the-job skills, as background to special assignments, to keep professionals abreast of changes in their fields, and to aid higher level and execu tive professionals in obtaining the best results from their staffs. In training staff, a special effort is made to impart detailed knowledge of the techniques used in collecting and compiling the statistics, so that maximum application of data results to current problems can be made without a risk of exceeding the limits of their significance. Organization The statistical programs of the Bureau were developed, for the most part, independently of each other, taking on characteristics suited to the requirements of the subject under observa tion. As a result, the Bureau was organized according to subject-matter areas, an arrange ment which has proved efficient and has been continued over the years. Expertise in tech niques, economic analysis, and other staff ac tivities across subject-matter lines were added to provide better utilization of the Bureau’s resources. As the Bureau’s collection activities in creased, regional offices were established in 1943 to provide administration of the field pro grams and staff. Another function of the offices was to disseminate data to local users and to furnish technical advice and assistance to State agencies and other cooperating organizations. An important aspect of the work of the re gional staffs has been the function of explain ing the concepts and techniques which we utilize in compiling our statistics. 4 BLS HANDBOOK OF METHODS Special recognition of the competence of the Bureau in the field of statistics was given by the Secretary of Labor in 1955 when he dele gated to the Commissioner of Labor Statistics the responsibility for continuously reviewing all of the statistical programs of the Depart ment of Labor and of making recommendations for their improvement. Consultation and Advice on Statistical Programs A statistical program too much detached from the uses of its data may fail in its prin cipal mission. To avoid sterility and stagnation, the Bureau continuously invites advice and ideas from users and experts in business, labor, and academic organizations and individual members of the public. Over the years, the advice the Commissioner of Labor Statistics has received on policy and technical matters from responsible parties, relating to the col lection and analysis of our statistics, has usually been sound and therefore very helpful. Of course, decisions on statistical policy have always been the final responsibility of the Commissioner. In order to keep in touch with the current and anticipated needs of business and labor groups and to seek advice on technical prob lems, the Commissioner first established stand ing research advisory committees in 1947. These groups, now called the Business Re search Advisory Council and the Labor Re search Advisory Council, serve in an advisory capacity with respect to technical problems, consult on Bureau programs, and provide per spectives on Bureau programs in relation to needs of their members. The councils accom plish their work in general sessions and also through committees designated to subjectmatter fields on a more specialized basis. Com mittee memberships are augmented by other persons in industry or labor who have special competence although not council members. The councils may take formal action through reso lutions or recommendations on matters re garded as appropriate for such action, but such resolutions are merely advisory. Members of the councils and the subcommittees serve in their individual capacities, not as representa tives of their organizations. The members of the Labor Research Advisory Council are designated by the Commissioner of Labor Statistics under authorization by the Secretary of Labor, from nominations by the Director of Research, AFL-CIO. All research directors of international unions represented in the AFL-CIO, the Railway Labor Executives’ Association, and the railroad operating unions are invited to attend the general meetings of the council. The council provides general direc tion to the advisory activities of trade union research directors in relation to the Bureau. The members of the Business Research Ad visory Council are designated by the Commis sioner under authorization of the Secretary of Labor, after consultation with the National Association of Manufacturers, the U.S. Cham ber of Commerce, and other organizations broadly representative of American business. Members serve in their individual capacities, not as representatives of their companies. The Bureau often seeks the advice of profes sional economists, statisticians, social scien tists, educators, and others, either in their individual capacities or as members of profes sional organizations. This is most likely to oc cur when a conceptual or theoretical question arises which is considered fundamental to the work of the Bureau in a specialized field, and where professional acceptance of the Bureau’s work in that field may be reinforced by the findings of an independent analyst. It is a fundamental objective of the Bureau that its statistical practice be built soundly upon established statistical theory. The objec tive can be realized only if BLS practitioners are trained in statistics and if they keep their knowledge up to date. For this reason the Bu reau encourages their participation in activities of the professional societies, their efforts to improve their education in statistics, and con tinuing contacts with other experts in their disciplines. The efforts of other statistical or ganizations, public and private, are studied unremittingly so that BLS may reap benefits from the experiences of others. INTRODUCTION Standard Definitions Where related statistics cut across program lines or across Government bureaus, the Bu reau of Labor Statistics cooperates to the maximum extent possible in the Office of Man agement and Budget’s (formerly, the Bureau of the Budget) effort to obtain adherence to stand ard definitions of terms for maximum com parability. The use of the definition of estab lishment is a case in point. The business establishment has been found to be the most satisfactory source of data for most industrial statistics because it is the first level of business organization for which com plete records— i.e., production, employment, purchases, sales, wages, inventories, etc.— are generally maintained. The establishment is the primary unit of organization in the business economy and is the first integrated level or combination of employees devoted to the pro duction of a related group of products or services. In BLS programs in which it is ap 5 plicable, the standardized definition of the establishment is utilized.9 The Bureau also follows the Office of Man agement and Budget’s definition of “ production and related workers,” 1 1and uses the “ standard 0 payroll period.” 1 The reader is referred to 1 appendixes B and C for descriptions of the standards followed with respect to industrial classification and geographic classification. 9 See appendix B. 1 Standard Definitions of Types of Workers, Bureau of the 0 Budget, November 7, 1944. “Production and related workers are defined to include working foremen and all nonsupervisory workers (including leadmen and trainees) engaged in fabricating, processing, assembling, inspection, receiving, storage, handling, packing, warehousing, shipping, mainte nance, repair, janitorial, watchman services, product develop ment, auxiliary production for plant’s own use (e.g., powerplant) and record-keeping and other services closely associated with the above production operations. Excluded are supervisory employees (above the working foremen level) and their clerical staffs.” 1 Standard Definition of Payroll Periods for Employment 1 Reports, Bureau of the Budget, March 28, 1952. "In order to maintain a coordinated system of employment reports and to reduce the reporting burden on respondents, requests made to employing establishments for statistical information from their payroll records on the number of employees, payrolls, hours worked, or related items, should refer to the payroll period containing the 12th of the month.” Current Employment Analysis Chapter 1. Labor Force, Employment, and Unemployment Background Each month, the Bureau analyzes and pub lishes statistics on the labor force, featuring information on employment, unemployment, and nonparticipation, classified by a variety of demographic, social, and economic charac teristics. These statistics are derived from the Current Population Survey, (CPS), which is conducted by the Bureau of the Census. The survey is based on a probability sample of households, representative of the civilian noninstitutional population of the United States. Concepts of the labor force, employment, and unemployment were introduced in the lat ter stages of the depression of the 1930’s, chiefly to arrive at more objective measure ments of unemployment and employment than were previously available. Before the 1930’s, aside from attempts in some of the decennial censuses, no direct measurements were made of the number of jobless persons. The develop ment of mass unemployment in the early 1930’s increased the need for statistics, and widely conflicting estimates based on a variety of in direct techniques began to appear. Dissatisfied with these methods, many research groups, as well as State and municipal governments, be gan experimenting with direct surveys of the population or samples of the population. In these surveys, an attempt was made to classify the population as employed, unemployed, or out of the labor force, by means of a series of questions addressed to each individual. In most of the surveys, the unemployed were defined as those who were not working but were “ willing and able to work.” This concept, however, did not meet the standards of objectivity that many technicians felt were necessary to measure either the level of unemployment at a point in time or changes over periods of time. The cri terion “ willing and able to work,” when applied in specific situations, appeared to be too intan gible and too dependent upon the interpretation and attitude of the persons being interviewed. Out of this experimentation, a set of concepts was developed in the late 1930’s which sought to meet these various criticisms. According to the new concepts, the classification of an in dividual depends principally on his actual ac tivity within a designated time period, i.e., was he working, looking for work, or engaged in other activities? These concepts were adopted for the national sample survey of households initiated by the Works Progress Administra tion (W PA) in 1940. Originally termed the Monthly Report on Unemployment when the WPA was responsible for the collection of labor force statistics, the household survey was transferred to the Bu reau of the Census late in 1943. Its name was changed at that time to the Monthly Report on the Labor Force (M RLF). The survey title was changed once more— in 1948— to the pres ent “ Current Population Survey” in order to reflect more accurately its expanding role as a source for a wide variety of demographic and economic characteristics of the population. In 1959, responsibility for analyzing and publish ing the CPS labor force data was transferred to the BLS, although the Census Bureau has continued to collect and tabulate the statistics. Description o f Survey The CPS provides statistics on the civilian noninstitutional population 16 years of age and over. Figures on the Armed Forces (obtained monthly from the Department of Defense) are added to the CPS estimates to derive estimates of the total labor force and the total nonin- 7 8 BLS HANDBOOK OF METHODS stitutional population. Persons under 16 years of age are excluded from coverage in the sur vey because child labor laws, compulsory school attendance, and general social custom prevent most of these children in the United States from working. The institutional population, which is excluded from coverage, consists of inmates of penal and mental institutions, tuber culosis sanitariums, and homes for the aged, infirm, and needy.1 The CPS is collected each month from a probability sample of approximately 47,000 households. Participation in the survey is on a purely voluntary basis. Respondents are as sured that all information obtained is com pletely confidential and is used only for the purpose of estimating national totals. The time period covered in the monthly sur vey is a calendar week. A calendar week was selected as the survey reference period because the period used must be short enough so that the data obtained is “ current” and the time reference not so short that the occurrence of holidays or other accidental events causes ex tremely erratic fluctuations in the information obtained. A calendar week fulfills these condi tions as well as being a convenient and easily defined period of time. Since July 1955 the week containing the 12th day of the month has been used. The actual survey is conducted dur ing the following week, which is the week con taining the 19th day of the month. Concepts The criteria used in classifying persons on the basis of their labor force activity are as follows: Employment. Employed persons comprise (1) all those who, during the survey week, did any work at all as paid employees, or in their own business, profession, or farm, or who worked at least 15 hours as unpaid workers in a familyoperated enterprise and (2) all those who were not working but who had jobs or businesses 1 For a fuller explanation of the Current Population Survey, see Concepts and Methods Used in Manpower Statistics from the Current Population Survey, U.S. Department of Com merce and U.S. Department of Labor (1967), Report 313. from which they were temporarily absent because of illness, bad weather, vacation, labormanagement dispute, or various personal rea sons. Each employed person is counted only once. Those who held more than one job are counted in the job at which they worked the greatest number of hours during the survey week. Excluded from the employed group are persons whose only activity consisted of work around their own home (such as housework, painting, repairing, etc.) or volunteer work for religious, charitable, and similar organizations. Unemployment. Unemployed persons include those who did not work at all during the survey week, were looking for work, and were avail able for work in the reference period. Those who had made efforts to find work within the preceding 4-week period— such as by register ing at a public or private employment agency, writing letters of application, canvassing for work, etc.— and who, during the survey week, were awaiting the results of these efforts, are considered to be looking for work. Also in cluded as unemployed are those who did not work at all during the survey week and (a) were waiting to be called back to a job from which they had been laid off, (b) were waiting to report to a new wage or salary job scheduled to start within the following 30 days (and were not in school during the survey week), or (c) would have been looking for work ex cept that they were temporarily ill. Duration of unemployment represents the length of time (through the current survey week) during which persons classified as un employed had been continuously looking for work. For persons on layoff, duration of un employment represents the number of full weeks since the termination of their most re cent employment. A period of 2 weeks or more during which a person was employed or ceased looking for work is considered to break the continuity of the present period of seeking work. Average duration is an arithmetic mean computed from a distribution by single weeks of unemployment. Labor Force. The civilian labor force comprises the total of all civilians classified as employed LABOR FORCE, EMPLOYMENT, AND UNEMPLOYMENT and unemployed. The total labor force includes members of the Armed Forces stationed either in the United States or abroad. Not in. Labor Force. All civilians who are not classified as employed, unemployed, or in the Armed Forces are defined as “ not in the labor force.” These persons are further classified as “ engaged in own housework,” “ in school,” “ unable to work” because of long-term physical or mental illness, and “ other.” The “ other” group includes retired persons, individuals re ported as too old or temporarily unable to work and the voluntarily idle. It also includes sea sonal workers for whom the survey week fell in an “ off” season and who were not reported as looking for work and persons who did not look for work because they believed either that no jobs were available in the area or that no jobs were available for which they could qualify. Persons doing only incidental unpaid family work (less than 15 hours in the specified week) also are classified as not in the labor force. The category “ not in labor force— in school” includes persons attending school during the survey week who had new jobs to which they Were scheduled to report within 30 days. All persons— whether or not attending school— who had new jobs not scheduled to begin until after 30 days (and who were not working or looking for work) are classified as not in the labor force. For persons not in the labor force, questions are asked about previous work experience, in tentions to seek work, desire for a job at the time of interview, and reasons for not looking for work. The questions for persons not in the labor force are asked only in those households that are new entrants to the sample and in those that are reentering the sample after 8 months’ absence. (See Sampling.) Sampling The Survey Design The CPS sample is located in 461 sample areas comprising 923 counties and independent cities with coverage in every State and the District of Columbia. In all, about 55,000 hous 9 ing units and other living quarters are desig nated for the sample at any time, of which about 47,000, containing about 105,000 persons 16 years and over, are occupied by households eligible for interview. The remainder are units that are vacant, converted to nonresidential use, containing persons who reside elsewhere, or ineligible for other reasons. Of the occupied units eligible for enumeration, about 3 to 5 percent are not interviewed in a given month because the residents are not found at home after repeated calls, are temporarily absent, or are unavailable for other reasons.2 Selection of Sample Areas. The entire area of the United States consisting of 3,000 counties and independent cities is divided into 1,931 primary sampling units. With some minor ex ceptions, a primary sampling unit (PSU) con sists of a county or a number of contiguous counties. Each of the 237 standard metropoli tan statistical areas (SMSA’s )3 in existence at the time of the 1970 Census constitutes a separate PSU. By combining counties to form PSU’s, greater heterogeneity is accomplished. Moreover, another important consideration is to have the PSU sufficiently compact in area so that, with a small sample spread throughout, it can be canvassed efficiently without undue travel cost. A typical primary sampling unit, for example, includes both urban and rural residents of both high and low economic levels and encompasses, to the extent feasible, diverse occupations and industries. The PSU’s are grouped into 461 strata. Among these PSU’s, 146 of the largest SMSA’s (including all having over 250,000 inhabitants) and 10 other areas (not SMSA’s) are separate strata representing themselves. In general, however, a stratum consists of a set of PSU’s as much alike as possible in various charac teristics such as geography, population density, rate of growth in the 1960-70 decade, propor tion of Negroes and other minority races, prin cipal industry, and type of agriculture. 2 For a more detailed description of the sample design and other technical phases of the CPS program, see U.S. Depart ment of Commerce, Bureau of the Census, Technical Paper No. 7, “The Current Population Survey— A Report on Meth odology,” (1963). 3 See appendix C. 10 BLS HANDBOOK OF METHODS Except for the 156 areas mentioned, each of which is a complete stratum, the strata are es tablished so that their sizes in terms of 1970 population are approximately equal. From each stratum a single PSU is selected to represent the entire stratum. In the 156 strata in which there is only a single PSU (the 146 SMSA’s and 10 special cases), the single PSU auto matically falls in the sample. When the stratum has more than one PSU, the sample PSU is selected in a random manner in such a way that its probability of selection is proportionate to its 1970 population. For example, within a stratum the chance that a PSU having a popu lation of 50,000 would be selected is twice that for a unit having a population of 25,000. Selection of Sample Households. The sample design calls for a sampling ratio which de pends on the predetermined total sample size. For 1971-73, it is 1 household for every 1,300 households in each stratum. The sampling ratio is modified slightly by month, as the size of the sample is held relatively constant despite the overall growth of the population. The sampling ratio used within each sample PSU depends on the proportion that the population of the sample area was of the stratum population at the time of the 1970 Census. In a sample area which was one-tenth of the stratum, the withinPSU sampling ratio that results is 1 in 130.0 thereby achieving the desired ratio of 1 in 1,300 for the stratum. For each PSU that is a stratum representing only itself, the sampling ratio is 1 in 1,300 regardless of the size of the PSU. With each of the 461 sample PSU’s, the num ber of households to be enumerated each month is determined by the application of the withinPSU sampling ratio rather than through the assignment of a fixed quota. This procedure makes it possible to reflect, on a current basis, population changes within the sample area. Consequently, the sample as a whole properly reflects the changing distribution of the popu lation and avoids the distortion which would result from the application of fixed quotas of households or persons based on the population at an earlier date. Within each designated PSU, several stages of sampling may be used in selecting the units to be enumerated. The first step is the selection of a sample of census enumeration districts (ED ’s), which are administrative units used in the 1970 Census and contain, on the average, about 250 households. These are selected sys tematically from a geographically arranged listing, so that the sample ED’s are spread over the entire PSU. The probability of selection of any one ED is proportionate to its 1970 popu lation. The next step is to select a cluster of approxi mately 4 households to be enumerated within each designated ED. This selection is made wherever possible, from the list of addresses for the ED compiled during the 1970 Census or, if the addresses are incomplete or inade quate, by area sampling methods. The address lists are used in about two-thirds of the cases, primarily in urban areas, and area sampling is applied in the remainder. An effort is made to have all units at one single address included within the same segment. This makes it rela tively easy for the interviewer to cover all units designated for the sample. The list sample is supplemented by a selec tion of the appropriate proportion of units newly constructed in the PSU since the Census date. The addresses of these units are obtained mainly from records of building permits main tained by the offices responsible for issuing permits in that area. A special procedure for updating parts of the census lists also is fol lowed to reflect either units missed in the Census or new construction in areas where there is no adequate system of building per mits. In those enumeration districts where area sampling methods are used— mainly rural areas— the ED’s are subdivided into segments, that is, small land areas having well-defined boundaries and in general an expected “ size” of about 12 housing units or other living quar ters. For each subdivided enumeration district, one segment is designated for the sample; the probability of selection is proportionate to the estimated “ size” of the segment. An inter viewer does not conduct interviews at all hous ing units in the segment, however; she is in stead given a sampling pattern so that, in gen eral, one-third of the units are enumerated. The LABOR FORCE, EMPLOYMENT, AND UNEMPLOYMENT remaining housing units in the segment are then available for future samples. Rotation of Sample. Part of the sample is changed each month. A primary reason for ro tating the sample is to avoid the problems of uncooperativeness which arise when a constant panel is interviewed indefinitely. Another rea son for replacing households is to reduce the cumulative effect of biases in response that sometimes occur when the same persons are interviewed indefinitely. To accomplish this rotation of the sample on a gradual basis, maps and other materials for several samples are prepared simultaneously. For each sample, eight systematic subsamples (rotation groups) of segments are identified. A given rotation group is interviewed for a total of 8 months, divided into two equal periods. It is in the sample for 4 consecutive months one year, leaves the sample during the following 8 months, and then returns for the same 4 calen dar months of the next year. In any 1 month, one-eighth of the sample segments are in their first month of enumeration, another eighth are in their second month, and so on; the last eighth are in for the eighth time, the fourth month of the second period of enumeration. Under this system, 75 percent of the sample segments are common from month to month and 50 percent from year to year. This pro cedure provides a substantial amount of monthto-month and year-to-year overlap in the panel, thus reducing discontinuities in the series of data, without burdening any specific group of households with an unduly long period of in quiry. Collection Methods Each month, during the calendar week con taining the 19th day, interviewers contact some responsible person in each of the sample house holds in the CPS. At the time of the first enumeration of a household, the interviewer prepares a roster of the household members, including their personal characteristics (date of birth, sex, race, marital status, educational 11 attainment, veteran status, etc.) and their re lationship to the household head. This roster is brought up to date at each subsequent inter view to take account of new or departed resi dents, changes in marital status, and similar items. The information on personal charac teristics thus is available each month for iden tification purposes and for cross-classification with economic characteristics of the sample population. At each monthly visit, a questionnaire is completed for each household member 16 years of age and over. The interviewer asks a series of standard questions on economic activity during the preceding week, the calendar week containing the 12th day of the month, called the “ survey week.” The primary purpose of these questions is to classify the sample popu lation into the three basic /economic groups— the employed, the unemployed, and those not in the labor force. Additional questions are asked each month to help clarify the information on employment status. For the employed, information is ob tained on hours worked during the survey week, together with a description of the curren'F job. For those temporarily away from their jobs, the enumerator records their reason for not working during the survey week, whether or not they were paid for their time off, and whether they usually work full or part time. For the unemployed, he records (1) methods used to find work, (2) the reason the unemployed persons had started to look for work, (3) the length of time they had been looking for work, (4) whether they were seek ing full- or part-time work, and (5) a descrip tion of their last full-time civilian job. For those outside the labor force, their principal activity during the survey week— whether keeping house, going to school, or doing some thing else— is recorded. In addition, for all households in the incoming or returning rota tion group, questions on the work history, rea sons for nonparticipation, and job seeking intentions of individuals not in the labor force are asked. The questionnaires containing the informa 12 BLS HANDBOOK OF METHODS tion obtained for each person in the sample are subjected to a field edit by clerks in each of the 12 regional offices of the Census Bureau. The field edit serves to catch omissions, incon sistencies, illegible entries, and errors at the point where correction is still possible. Many of the error corrections made in the field edit prevent delays in further processing of the questionnaires in Washington. After the field edit, the questionnaires are forwarded to the Washington office of the Census Bureau. All of the questionnaires are received in Washington by the end of the week after enumeration. The raw data are trans ferred to computer tape and checked for com pleteness and consistency. The interviewers on the CPS are chiefly part-time workers, although most of the staff at any time consists of persons who have had several years experience on the survey. They are given intensive training when first re cruited and have either direct or home study training each month, before the survey. More over, through editing of their completed questionnaires, repeated observation during enumeration, and a systematic reinterview of part of their assignments by the field super visory staff, the work of the interviewers is kept under control and errors or deficiencies are brought directly to their attention. Estimating Methods To increase the reliability of the labor force statistics derived from the sample, the esti mation procedure uses two stages of ratio estimates and a “ composite estimate” . Achieve ment of this rather complicated procedure is made rapidly and automatically because of the availability of high-speed electronic digital computers. The principal steps involved are as follows. Adjustment for Households Not Interviewed. The weights for all households interviewed are adjusted to the extent needed to account for units occupied by persons eligible for interview but for which no interview was obtained be cause of absence, impassable roads, refusals, or other reasons. This adjustment is made sep arately by groups of PSU’s and, within these, for each color (white or Negro and other races) and residence group of households (urban, rural nonfarm, rural fa rm ). The adjustment is made separately within each pair of rotation groups (the incoming pair, the two continuing pairs, and the outgoing pair). Ratio Estimates. The distribution of the popu lation selected for the sample may differ some what, by chance, from that of the Nation as a whole in such basic characteristics as age, color, sex, and farm-nonfarm residence, among other things. These particular population characteristics are correlated closely with la bor force participation and other principal measurements made from the sample. There fore, some of the sample estimates can be improved substantially when, by appropriate weighting of the original returns, the sample population is brought as closely into agreement as possible with the known distribution of these characteristics in the entire population. Such weighting is accomplished through two stages of ratio estimates: 1. First stage. The first stage of ratio esti mates takes into account differences in the distribution by color and residence of the popu lation estimates from the sample PSU’s and that of the total population in each of the four major regions of the country. However, in dependent distributions of the total population by residence, cross-classified by color, are not available on a current basis. Therefore, using 1970 Census data, estimated population totals by color and residence for a given region are computed from population counts for PSU’s in the CPS sample. Ratios then are computed between these estimates (based on sample PSU’s) and the actual population totals for the region as shown by the 1970 Census. In deriving these ratios, PSU’s that made up en tire strata and were selected with certainty (usually referred to as “ self-representing” PSU’s) are excluded from the computations, since they represent only themselves. In tabula tions of the monthly results from the CPS, the weights for all sample households from nonself-representing PSU’s in a given region are LABOR FORCE, EMPLOYMENT, AND UNEMPLOYMENT multiplied by the population ratio for that region for the appropriate color-residence group. 2. Second stage. The second stage of ratio estimates takes account of current differences between the population distributions of the sample and that of the Nation as a whole by age, color, and sex. Independent estimates of the entire population, by these characteristics, are prepared each month. They are calculated by carrying forward the most recent census data (1970) to take account of subsequent aging of the population, mortality, and migra tion between the United States and other coun tries. The CPS sample returns (taking into account the weights determined after the first stage of ratio estimates) in effect are used to determine only the distribution within a given age-color-sex group by employment status and various other characteristics. In developing statistics, these sample distributions are multi plied by the ratio of the independent population estimate to the sample estimate for the appro priate age-color-sex group. Composite Estimate. The last step in the prep aration of estimates makes use of a composite estimate. In this procedure, a weighted average of two estimates is obtained for the current month for any particular item. The first esti mate is the result of the two stages of ratio estimates described above. The second estimate consists of the composite estimate for the pre ceding month to which has been added an estimate of the change in each item between the preceding month and the present month, based upon that part of the sample which is common to both months (75 percent). Although the weights for the two components of such a com posite estimate do not necessarily have to be equal, in this instance the weights used for combining these two estimates are each onehalf. Equal weights in this case satisfy the condition that for virtually all items there will be some gain in reliability over the estimation procedure after the first two stages of ratio estimates. The composite estimate results in a reduction in the sampling error beyond that which is achieved after the two stages of ratio estimates 13 described; for some items the reduction is sub stantial. The resultant gains in reliability are greatest in estimates of month-to-month change, although gains also are obtained for estimates of level in a given month, change from year to year, or change over other inter vals of time. Presentation and Uses The CPS provides a large amount of detail on the economic and social characteristics of the population of the United States. It is the source of monthly estimates of total employ ment, both farm and nonfarm; of nonfarm selfemployed persons, domestics, and unpaid help ers in nonfarm family enterprises; and of total unemployment, whether or not covered by un employment insurance. It is a comprehensive source of information on the personal charac teristics such as age, sex, race, educational attainment, and marital status of the total la bor force and of the employed, the unemployed, and those not in the labor force. It provides distributions of workers by the numbers of hours worked, as distinguished from aggregate or average hours for an in dustry, permitting separate analyses of parttime workers, workers on overtime, etc. The survey is a comprehensive current source of information on the occupation of workers, whether teachers, stenographers, carpenters, laborers, etc. It also provides limited statistics on the industries in which they work. Information is available from the survey not only for persons in the current labor force but also for those who are outside the labor force, some of whom may be considered to be a “ labor reserve.” The characteristics of such persons — whether married women with or without young children, disabled persons, students, older retired workers, etc.— can be determined. Also, through special inquiries, it is possible to obtain information on their skills and past work experience, if any. Each month, a significant amount of basic information about the labor force is analyzed and published in Employment and Earnings. The tables in this report provide information 14 BLS HANDBOOK OF METHODS on the labor force, employment, and unemploy ment by a number of characteristics, such as age, sex, color, marital status, industry, and occupation. Approximately 150 of the most im portant estimates from the CPS are presented each month on a seasonally adjusted basis. These estimates are adjusted by the BLS Sea sonal Factor Method, which is described in the appendix on seasonal adjustment. The CPS is used also for a program of spe cial inquiries to obtain detailed information from particular segments, or for particular characteristics of the population and labor force. Approximately 8 to 10 such special sur veys are made each year. The inquiries are repeated annually in the same month for some topics, including the earnings and total incomes of individuals and families (published by the Census Bureau), the extent of work experience of the population during the calendar year, the extent of overtime work at premium pay, usual weekly earnings of wage and salary workers, and the prevalence of multiple jobholding. They also include marital and family charac teristics of workers, the employment of schoolage youth, the employment of recent high school graduates, the educational attainment of workers, and the employment situation in poverty areas of our major cities. Surveys have been made periodically on subjects such as job mobility, and length of time on current job. In addition, surveys are published in very great detail on the characteristics of the unemployed and persons not in the labor force. Generally, the persons who provide informa tion for the monthly CPS questions also answer the supplemental questions. Occasionally, the kind of information sought in the special sur vey requires the respondent to be the person about whom the questions are asked. Information obtained through the supple mental questions is combined with data in the regular schedule to provide tabulations of all the desired personal and economic charac teristics of the persons in the special survey. Reports on these special surveys are first pub lished in the Monthly Labor Review. Reprints of the articles, together with technical notes and additional tables, are published as Special Labor Force Reports. Limitations Geographic. The CPS is designed to produce reliable National estimates. It is not designed to produce estimates for States and areas. A sample which could produce State estimates as reliable as those now published for the Nation would have to be approximately fifty times as large as the present sample. Sources of Errors in the Survey Estimates. The estimates from the survey are subject to sampling errors, that is, errors arising from the fact that the estimates each month are based on information from a sample rather than the whole population. In addition, as in any survey work, the results are subject to errors made in the field and to errors that occur in the process of compilation. Classification errors in labor force surveys may be particularly large in the case of persons with marginal attachments to the labor force. These errors may be caused by interviewers, respondents, or both, or may arise from faulty questionnaire design. In spite of a continuous quality control program, interviewers may not always ask the questions in the prescribed fashion. To the extent that varying the word ing of the question causes differences in re sponse, errors or lack of uniformity in the statistics may result. Similarly, the data are limited by the adequacy of the information pos sessed by the respondent and the willingness to report accurately. The estimates from the survey are subject to various other types of errors beyond those al ready mentioned. Some of these are: 1. Nonresponse. About 3 to 5 percent of oc cupied units are not interviewed in a typical month because of temporary absence of the occupants, refusals to cooperate, or various other reasons. Although an adjustment is made in weights for interviewed households to ac count for noninterviews, they still represent a possible source of bias. Similarly, for a rela tively few households, some of the information is omitted because of lack of knowledge on the part of the respondent or because the inter viewer forgot to ask certain questions or record the answers. In processing the completed ques LABOR FORCE, EMPLOYMENT, AND UNEMPLOYMENT tionnaires, entries usually are supplied for omitted items on the basis of the distribution in these items for persons of similar charac teristics. 2. Independent Population Estimates. The independent population estimates used in the estimation procedure may also provide a source of error, although on balance their use sub stantially improves the statistical reliability of many of the important figures. (See discus sion under “ Ratio Estimates,” p. 12.) Errors may arise in the independent population esti mates because of underenumeration of certain population groups or errors in age reporting in the last census (which serves as the base for the estimates) or similar problems in the com ponents of population change (mortality, im migration, etc.) since that date. 3. Processing errors. Although there is a quality control program on coding and a close control on all other phases of processing and tabulation of the returns, some processing errors are almost inevitable in a substantial statistical operation of this type. However, the net error arising from processing is probably fairly negligible. Measuring the Accuracy of Results. Modern sampling theory provides methods for estimat ing the range of errors due to sampling where, as in the case of the CPS sample, the prob ability of selection of each member of the popu lation is known. Methods also are available for determining the effect of response variability in the CPS. A measure of sampling variability indicates the range of difference that may be expected because only a sample of the popula tion is surveyed. A measure of response varia bility indicates the range of difference that may be expected as a result of compensating types of errors arising from practices of differ ent interviewers and the replies of respond ents; these would tend to cancel out in an enumeration of a large enough population. In practice, these two sources of error— sampling and response variability, as defined above— are estimated jointly from the results of the survey. The computations, however, do not, in corporate the effect of response bias, that is, any systematic errors of response— for ex ample, those that would occur if, by and large, respondents tended to overstate hours worked. Response biases occur in the same way in a complete census as in a sample, and, in fact, they may be smaller in a well-conducted sample survey because for the relatively small sample it is feasible to pay the price necessary to collect the information more skillfully. Estimates of sampling and response varia bility combined are provided in Employment and Earnings and in other reports based on CPS data, thus permitting the user to take this factor into account in interpreting the data. In general, the smaller figures and small differ ences between figures are subject to relatively large variation and should be interpreted with caution. The availability of the high-speed elec tronic computer makes possible considerably more detailed estimates than were possible earlier. Estimation of response bias is one of the most difficult aspects of survey and census work. Systematic studies on this subject are now an integral part of the CPS,4 but in many instances available techniques are not suffi ciently precise to provide satisfactory estimates of response biases. Considerable experimenta tion is in progress with the aim of developing more precise measurements and improving the overall accuracy of the series.* * For a summary of these studies, see Bureau of the Census, Technical Paper No. 6, “ The Current Population Survey Re interview Program— Some Notes and Discussion” (March 1963). Technical References Number 1. President’s Committee to Appraise Employment and Unemployment Statistics, Measuring Employment and Unemployment (1962). A review of all Federal statistical series on employment and unemployment and a compari son of the sources and uses of each series. The discussion of labor force statistics includes a brief history of their development, an evaluation of current concepts and techniques, and rec- 15 16 BLS HANDBOOK OF METHODS 2. 3. 4. 5. ommendations for further research and improvements, several of which were inaugurated in January 1967. U.S. Department of Commerce, Bureau of the Census, Technical Paper No. 6, “ The Current Population Survey Reinterview Program— Some Notes and Discussion” (1963). A summary of procedures and results of the Current Population Survey Reinterview Pro gram from 1955 through 1961 and some interpretations and comparative results from other studies. U.S. Department of Commerce, Bureau of the Census, Technical Paper No. 7, “ The Current Population Survey— A Report on Methodology” (1963). A brief history of the Current Population Survey (CPS) from its inception (1943) to the present. A detailed description is given for both the sample design and survey procedures. Also included is a detailed discussion of the various modifications in design and procedures and the resultant gain in precision. U.S. Department of Labor, Bureau of Labor Statistics and U.S. Department of Commerce, Bureau of the Census, Concepts and Methods Used in Manpower Statistics from the Current Population Survey, BLS Report 313 and Current Population Reports, Series P-23, No. 22 (1967). A concise description of the methodology in obtaining labor force information from sample households. Labor force concepts and definitions are set forth. The adequacy of labor force data and quality controls are discussed, and major improvements in the Current Population Survey are listed chronologically. U.S. Department of Labor, Bureau of Labor Statistics, How the Government Measures Un employment, Report 312 (1967). A short nontechnical discussion of the concepts and methods used in obtaining labor force statistics from the Current Population Survey. Jo h n E . B regger Chapter 2. Employment, Hours, and Earnings Background The first monthly studies of employment and payrolls by the Bureau of Labor Statistics (BLS) began in October 1915 and covered four manufacturing industries. Before that year, the principal sources of employment data in the United States were the census surveys— the decennial Census of Population, and beginning in 1899, the quinquennial Census of Manufac tures. No regular employment data had been compiled between the Census dates. By November 1916, the BLS program had been expanded to cover 13 manufacturing in dustries, and this number remained unchanged until 1922. The depression of 1921 directed at tention to the importance of current employ ment statistics, and in 1922 Congress granted additional funds to provide for program expan sion. By June 1923, the number of manufactur ing industries covered by the monthly employ ment survey had increased to 52. In 1928, concern over increasing unemployment induced Congress to provide additional appropriations for the program. In the next 4 years, 38 manu facturing and 15 nonmanufacturing industries were added to the list of industries for which the Bureau published monthly information on employment and payrolls. The onset of the Great Depression in 1930 and the deepening economic crisis impelled President Hoover to appoint an Advisory Com mittee on Employment Statistics to study the need for expanded data in this field. The Com mittee made its report in the spring of 1931 with a number of recommendations for exten sion of the program. The most important of these called for the development of series on hours and earnings. For the fiscal year 1932, Congress granted the Bureau a substantial in crease in the appropriation for the program. In January 1933, average hourly earnings and average weekly hours for the first time were published for all manufacturing, for 90 manu facturing industries, and for 14 nonmanufac turing categories. During the Great Depression when mass un employment threatened to become a continuing aspect of American life, there was much contro versy among various authorities concerning the actual number of the unemployed. These discussions pointed up the fact that no reliable measures of either unemployment or employ ment existed. In the early years of the Roose velt administration, the Secretary of Labor frequently referred to the value of the Bureau’s employment estimates as an indirect measure of unemployment. This interest stimulated efforts to develop comprehensive estimates of total wage-and-salary employment in nonagricultural industries, and in 1936, the Bureau first published such a figure. The preparation of these estimates of overall employment totals on a monthly basis was con tingent on the development of benchmark data. It was recognized, even in the 1920’s, that :month-to-month employment trends derived from a sample of establishments might be fairly accurate for short periods, but over long inter vals of time the series would not represent the true movement of employment, unless they were adjusted periodically to reasonably com plete counts of employment, called benchmarks. The first such adjustment was made in 1935, when the Bureau’s employment series in manu facturing were adjusted to totals from the Cen sus of Manufactures for 1923, 1925, 1927, 1929, and 1931. These series were subsequently ad justed to the successive biennial Censuses of Manufacturers, through that of 1939. For non manufacturing industries, benchmarks were developed from various sources, including the Censuses of Business taken at intervals from 1929 on. From 1915 to the beginning of World War II, interest in employment statistics for States and areas was growing constantly. Even before the Bureau of Labor Statistics entered the field in 17 18 BLS HANDBOOK OF METHODS 1915, three States (Massachusetts, New York, and New Jersey) were preparing employment statistics. As early as 1915, New York and Wisconsin had entered into “ co-operative” agreements with the Bureau, whereby sample data collected by the State agency were to be used jointly with the Bureau of Labor Statistics for the preparation of State and national series. By 1928, five other States had entered into such compacts, and another five were added by 1936. Over the years, the amount of published data on employment and payrolls for States and areas underwent a constant expansion. In 1940, estimates of total nonagricultural employment for all 48 States and the District of Columbia were published for the first time. The onset of World War II in 1939, followed by the entry of the United States after the as sault on Pearl Harbor in December 1941, placed additional demands upon the Bureau’s Employ ment Statistics program. The added responsi bilities pointed up the need for greater uni formity among the various programs of estab lishment statistics on employment and related subjects which were being prepared by the BLS, the Bureau of the Census, and the agen cies administering the emerging social security programs. While most improvements had to await the end of the war, several important advances took place during those years. The most far reaching decision was to use as employment benchmarks the data on employ ment collected primarily for administrative purposes by the newly organized social insur ance programs. Tabulations of such materials became available about 1940 from the unem ployment insurance program and they soon be came the preferred sources of benchmark data. They covered several industrial categories not covered by the Census of Manufacturers and Business, respectively, and they were available annually. After 1939, these were taken only at 5-year intervals. As the unemployment insurance program de veloped, the feeling grew that the proper place to estimate State and area employment was in the State agencies rather than in Washington. By 1949, all States had joined the system, and since that year the industry employment sta tistics program has been a fully integrated Federal-State project which provides employ ment, hours, and earnings information on a national, State, and area basis in considerable industrial detail. This cooperative program has as its formal base of authority a Congressional act of July 7, 1930 (ch. 873, 46 Stat. 1019; 29 U.S.C. 2). In 1971 cooperative arrangements were in effect within 44 States and the District of Columbia and with 6 State labor depart ments. Description o f the Survey The Bureau of Labor Statistics cooperates in collecting data each month on employment, hours, and earnings from a sample of establish ments in all nonagricultural activities including government. In 1970, this sample included over 165,000 reporting units. From these data a large number of series on employment, hours, and earnings in considerable industry detail are prepared and published monthly for the United States as a whole, for each of the 50 States and the District of Columbia, and for most of the metropolitan areas. The data in clude series on total employment, production or nonsupervisory worker employment, women employed, average hourly earnings, average weekly hours, and average weekly overtime hours (in manufacturing). For many series, seasonally adjusted data also are published. Concepts An establishment is defined as an economic unit which produces goods or services, such as a factory, mine, or store. It is generally at a single physical location and it is engaged predomi nantly in one type of economic activity. Where a single physical location encompasses two or more distinct and separate activities these are treated as separate establishments, provided that separate payroll records are available and certain other criteria are met.1 In the collection of data on employment, payrolls, and man hours, the BLS usually requests separate re 1 Standard Industrial Classification Manual (Bureau of the Budget, Executive Office of the President, 1967), p. x. EMPLOYMENT, HOURS, AND EARNINGS ports by establishment. However, when a com pany has more than one establishment engaged in the same activity in a geographic area, these establishments may be covered by a combined report. Industry employment statistics published by BLS and the cooperating State agencies repre sent the total number of persons employed either full-time or part-time in nonagricultural establishments during a specified payroll period. In general, data refer to persons who worked during, or received pay for, any part of the pay period that includes the 12th of the month. However, at the national level, data for Federal Government establishments generally refer to civilian personnel who worked on, or received pay for, the last day of the month, plus inter mittent employees who worked any time during the month (e.g., Christmas temporary em ployees of the postal service). Employed persons include both permanent and temporary employees and those who are working either full- or part-time. Workers on an establishment payroll who are on paid sick leave (when pay is received directly from the employer), on paid holiday or paid vacation, or who work during only a part of the specified pay period are counted as employed. Persons on the payroll of more than one establishment dur ing the pay period are counted in each establish ment which reports them, whether the duplica tion is due to turnover or dual jobholding. Persons are considered employed if they receive pay for any part of the specified pay period, but are not considered employed if they receive no pay at all for the pay period. Since pro prietors, the self-employed, and unpaid family workers do not have the status of “ paid em ployees,” they are not included. Domestic work ers in households are excluded from the data for nonagricultural establishments. Government employment statistics refer to civilian em ployees only. The figure which includes all persons who meet these specifications is designated “ all em ployees.” Major categories of employment are differentiated from this overall total, primarily to ensure the expeditious collection of current statistics on hours and earnings; these groups 19 of employees are designated production work ers, construction workers, or nonsupervisory workers, depending upon the industry. In manufacturing industries, data on em ployment, man-hours, and payrolls are collected for production workers. This group, in general, covers those employees, up through the level of working foremen, who are engaged directly in the manufacture of the product of the establish ment. Among the exclusions from this category are persons in executive and managerial posi tions, and persons engaged in activities such as accounting, sales, advertising, routine office work, professional and technical functions, and force account construction.2 Production work ers in mining are defined in a similar manner. A more detailed description of the classes of employees included in the production and non production workers categories in mining and manufacturing is shown on the facsimile of the BLS 790 A schedule on page 31 of this bulletin. In contract construction, the term construc tion workers covers workers, up through the level of working foremen, who are engaged directly on the construction project either at the site or working in shops or yards at jobs ordi narily performed by members of construction trades. Exclusions from this category include executive and managerial personnel, profes sional and technical employees, and routine office workers. Data on the employment, man-hours, and payrolls of nonsupervisory workers are col lected from establishments in the transporta tion, communication, and public utility indus tries, in retail and wholesale trade, in finance, insurance, and real estate, and in most of the service industries. Nonsupervisory workers include most employees except those in top executive and managerial positions. (See fac simile of BLS 790 E, the reporting form for wholesale and retail trade, p. 32.) The series on hours and earnings is based on reports of gross payroll and corresponding paid - Force-account construction is construction work performed by an establishment, primarily engaged in some business other than construction, for its own account and use and by its own employees. 20 BLS HANDBOOK OF METHODS man-hours for production workers.3 To derive * these series, BLS collects the following data: (See facsimile of BLS 790 A on p. 31). 1. The number of full- and part-time produc tion workers who worked during, or received pay for, any part of the pay period including the 12th of the month. 2. Total gross payrolls for production work ers before deductions for old-age and unemploy ment insurance, group insurance, withholding tax, bonds, and union dues. The payroll figures also include pay for overtime, shift premiums, holidays, vacations, and sick leave paid directly by the firm to employees for the pay period reported. They exclude bonuses (unless earned and paid regularly each pay period) or other pay not earned in the pay period concerned (e.g., retrocative pay). Tips and the value of free rent, fuel, meals, or other payment in kind are not included. 3. Total man-hours worked (including over time hours) during the pay period, hours paid for standby or reporting time, and equivalent man-hours for which employees received pay directly from the employer for sick leave, for holidays, vacations, and other leave paid to these employees. Overtime or other premium paid hours are not converted to straight-time equivalent hours. 4. Overtime man-hours for which premiums were paid because the hours worked were in excess of the number of hours of either the straight-time workday or workweek. Saturday and Sunday hours (or 6th and 7th day hours) are included as overtime only if overtime premiums were paid. Holiday hours worked as overtime are not included unless they are paid for at more than the straight-time rate. Hours for which only shift differential, hazard, incen tive, or similar types of premiums were paid are excluded from overtime hours. Overtime hours data are collected only from establishments engaged in manufacturing in dustries. For government organizations and private educational institutions, payrolls col lected relate to all employees. Manhour data are not collected. 3 Unless otherwise indicated, the references in this chapter to production workers apply likewise to the construction worker and nonsupervisory categories. Industrial Glassification All national, State, and area employment, hours, and earnings series data are classified in accordance with the Standard Industrial Classification Manual, Office of Management and Budget, 1967. (See appendix D of this bulletin for a detailed description of this sys tem.) Reporting establishments are classified into significant economic groups on the basis of major product or activity as determined by the establishments’ percent of total sales or receipts for the previous calendar year. This information is collected once each year on an “ Industry Class Supplement” to the monthly report form. (See p. 33 for a facsimile of this form.) All data for an establishment making more than one product or engaging in more than one activity are classified under the indus try of the most important product or activity, based on the percentages reported. Time Period Employment, hours, and earnings are meas ured for the pay period including the 12th of the month, which is standard for all Federal agencies collecting employment data on an establishment basis. Data Sources Sample Data Each month the State agencies cooperating with the Bureau collect data on employment, payrolls, and man-hours from a sample of es tablishments. The respondent extracts these figures from his payroll records. These data are readily available as the employers must maintain such records for a variety of tax and accounting purposes. A response analysis sur vey of the reporting practices of a scientifically selected sample of reporting establishments in manufacturing industries showed that the re ports were made out almost exclusively from their payroll records. The survey also showed EMPLOYMENT, HOURS, AND EARNINGS that while a number of employers did not re port precisely the data requested on the sched ule for all items, these deviations were not all in the same direction. On balance, they tended to offset each other, and the net effects of in correct reporting were quite insignificant.4 Participation in the industry employment statistics program is entirely voluntary on the part of the reporters. However, in many in dustries, particularly in manufacturing, em ployers who have a high percentage of total employment in the industry supply reports regularly, and many have done so over a long period of years. Benchmark Data An employment benchmark is defined as a reasonably complete count of employment. The estimates are adjusted periodically, annually if possible, to new benchmark levels. Since 1939, the basic sources of benchmark information for “ all employees” have been periodic tabulations of employment data by industry and, beginning with 1959, by size of establishment. These are compiled by State employment security agen cies, according to uniform procedures specified by the Manpower Administration of the U.S. Department of Labor, from reports of estab lishments covered under State unemployment insurance laws. The State employment security agencies receive quarterly reports, from each employer subject to the laws, showing total employment in each month of the quarter, and total quarterly wages for all employees.5 If the employer has more than 50 employees and op erates more than one establishment in a State, he is required to make separate reports for each area (e.g., county) in which he operates and for each establishment in different indus4 Young, Dudley E. and Goldstein, Sidney, “ The BLS Em ployment Series and Manufacturing Reporting Practices,” Monthly Labor Review, November 1957, pp. 1367-1371. r The State employment security agencies are required to > submit tabulations of these reports to the Manpower A d ministration each quarter. These tabulations are due in the Washington office of MA by the middle of the sixth month after the end of the quarter of reference. For example, the first quarter tabulation, which provides the basis of the BLS benchmarks, is due on September 15. Review and editing of these tabulations and preparation of national summaries from them requires several months additional work on the part of both BLS and MA before the benchmark is completed. 21 tries. Employment is reported for the pay period of the month including the 12th, and reports are classified industrially according to the Standard Industrial Classification. The State employment security agencies cooperate closely with the Bureau of Labor Statistics in the assignments of industry classifications, so there is a high degree of uniformity in this respect between the benchmark and sample data. In 1970, unemployment insurance data ac counted for three-fourths of the total bench mark. For the group of establishments exempt from State unemployment insurance laws be cause of their small size0 and for certain classes of nonprofit institutions, the data used are those provided by the national old-age in surance program administered by the Social Security Administration of the U.S. Depart ment of Health, Education, and Welfare. For industries not covered by either of the two programs, benchmarks are compiled from a number of special sources. The most im portant of these are the Interstate Commerce Commission (interstate railroads), the Ameri can Hospital Association (private nonprofit hospitals), the Office of Education in the U.S. Department of Health, Education, and Welfare, and the National Catholic Education Associa tion (private schools, colleges, and universi ties), the U.S. Civil Service Commission (Federal Government), and the Governments Division of the Bureau of the Census (State and local government) .* 7 Special efforts are made to classify establish ments into the same industrial groupings for benchmark purposes as they are for monthly reporting. Wherever possible, employment for the standard midmonth pay period for March is used as the benchmark. 1 In 1970 the unemployment insurance laws of 21 States and 1 the District of Columbia covered all employers of 1 or more workers, 3 States covered employers of 3 workers or more, and the remaining 26 States covered employers of 4 workers or more. Beginning in 1972, all State unemployment insurance laws should cover employers of 1 or more workers. 7 For a more detailed description of the benchmark, see Armknecht, Paul A., “BLS Establishment Estimates Revised to March 1970 Benchmark Levels,” Employment and Earnings, September 1971, pp. 18-30. 22 BLS HANDBOOK OF METHODS Collection Methods The primary collection of the current sample data is conducted by State agencies which have cooperative agreements with the BLS. In most States, this is the employment security agency, affiliated with the Manpower Administration, the organization which administers the State unemployment insurance program. In a few cases the State department of labor acts as the agency. The agencies mail schedules to a sam ple of establishments in the States each month. A “ shuttle” schedule is used (BLS form 790 series) ; that is, one which is submitted each month in the calendar year by the respondent, edited by the State agency, and returned to the respondent for use again the following month. The State agency uses the information pro vided on the forms to develop State and area estimates of employment, hours, and earnings, and then forwards the data, either on the schedules themselves or in machine readable form, to the Washington, D.C., office of the Bureau of Labor Statistics, where they are used to prepare estimates at the national level. The shuttle schedule has been used in this program since 1930, but there have been sub stantial changes in its design and in the data collected over the period. All aspects of the schedule, its format, the wording of the re quested items and definitions, and the concepts embodied therein have been subjected to a con tinuing and intensive review over the entire period, not only by the staff of BLS and of the State agencies, but also by other government agencies and by numerous persons in private business and labor organizations. The report forms are not exactly alike for every industry, but most of them request data on total employ ment, number of women employees, number of production workers (in manufacturing and mining), construction workers (in contract construction industries), or nonsupervisory workers (in other nonmanufacturing indus tries), and, for these workers, data on payroll, paid man-hours, and for manufacturing, over time man-hours. The schedule contains detailed instructions and definitions for the reports. There are several variants designed to meet the specific problems of different industries. (See facsimiles of BLS 790 A and, BLS 790 E, pp. 31 and 32.) The technical characateristics of the shuttle schedule are particularly important in main taining continuity and consistency in reporting from month to month. The design exhibits automatically the trend of the reported data during the year covered by the schedule, and therefore, the relationship of the current figure to the data for the previous month. The sched ule also has operational advantages; for ex ample, accuracy and economy are obtained by entering identifying codes and the address of the respondent only once a year. All schedules are carefully edited by the State agencies each month to make sure that the data are correctly reported and that they are consistent within themselves, with the data reported by the establishment in earlier months, and with those reported by other es tablishments in their industry. This editing process is carried out in accordance with a de tailed manual of instructions prepared by the Bureau of Labor Statistics. When the reports are sent to Washington, they are screened by use of an electronic computer to detect process ing errors and reporting errors which may have escaped the first editing. Questionable cases discovered at any stage of the editing process are returned, if necessary, to the re spondent for review and correction.8 Sam pling Sampling is used by BLS in its industry employment statistics program for collecting data in most industries, since full coverage would be prohibitively costly and time consum ing. The sampling plan for the program must: (a) provide for the preparation of reliable monthly estimates of employment, hours of work, and weekly and hourly earnings which can be published promptly and regularly; (b) through a single general system, yield con siderable industry detail for metropolitan areas, States, and the Nation; and (c) be ap8 Mendelssohn, Rudolph C. "Machine Methods in Employ ment Statistics,” Monthly Labor Review, May 1955, pp. 567-569. EMPLOYMENT, HOURS, AND EARNINGS propriate for the existing framework of op erating procedures, administrative practices, resource availability, and other institutional characateristics of the program. In developing the sample design, the universe of establishments was stratified first by in dustry and within each industry by size of establishment in terms of employment, using six standard size classes. Within each industry, an optimum allocation design was obtained by sampling with probability proportionate to average size of establishment within each of the strata. A total size of sample necessary to produce satisfactory estimates of employment had to be distributed among the size-class cells on the basis of average employment per estab lishment in each cell. In practice, this is equiva lent to distributing the predetermined total number of establishments required in the sam ple among the cells on the basis of the ratio of employment in each cell to total employment in the industry. Within each stratum, the sam ple members are selected at random. Under this type of design, large establish ments fall into the sample with certainty. In nearly all industries, establishments with 250 or more employees are included in the sample with certainty and in many industries the cut off is lower. The sizes of the samples for the various industries were determined empirically on the basis of experience and of cost consid erations. In a manufacturing industry in which a high proportion of total employment was concentrated in a relatively few large establish ments, a high percentage of total employment had to be included in the sample. Consequently, the sample design for such industries provides for a complete census of the larger establish ments with only a few chosen from among the smaller establishments, or none at all if the concentration of employment in the larger es tablishments is great enough. On the other hand, in an industry where a large proportion of total employment is in small establishments, the sample design calls for inclusion of all large establishments, and also for a substantial num ber of the smaller establishments. Many in dustries in the trade and service divisions fall into this category. In order to keep the sample to a size which can be handled with available 23 resources, it is necessary to accept samples in these divisions with a smaller proportion of universe employment than is the case for most manufacturing industries. Since individual es tablishments in these nonmanufacturing indus tries generally show less fluctuation from regular cyclical or seasonal patterns than es tablishments in manufacturing industries, these smaller samples (in terms of employ ment) generally produce reliable estimates. This sample design, although aimed pri marily at meeting the needs of the national program, provides a technical framework within which State and area sample designs can be determined. Since the estimates for States and areas generally are not prepared at the same degree of industry detail as the national estimates, the national design usually provides sufficient reports for the preparation of State and area estimates.9 Estimating Procedures Employment The “ all-employee” estimates by industry are based on reasonably complete employment counts or “ benchmarks.” To obtain employment estimates for the individual estimating cells, the following three steps are necessary: 1. A total employment figure (benchmark) for the estimating cell, as of a specified month, is obtained from sources which provide a rea sonably complete count of employment for the cell. 2. For each cell, the ratio of employment in one month to that in the preceding month (i.e., the link relative) is computed for sample es tablishments which reported in both months. 3. Beginning with the benchmark month, the estimate for each month is obtained by multi plying the estimate for the previous month by the link relative for the current month. Application of the estimating procedure in preparing a series is illustrated by the follow ing example: Assume that total employment 9 For the national sample, additional reports needed for State and area samples are added to those required by the national design. 24 BLS HANDBOOK OF METHODS for a given series was 50,000 in July. The re porting sample, composed of 60 establishments, had 25,000 employees in July and 26,000 in August, a 4-percent increase. To derive the August estimate, the change for identical es tablishments reported in the July-August sam ple is applied to the July estimate: 5 0 ,0 0 0 x -6>° - ( o r 1.04) =52,000 25,000 This procedure for estimating current em ployment is known as the benchmark and link-relative technique. It is an efficient tech nique, taking advantage of a reliable complete count of employment and of the high correla tion between levels of employment in successive months in identical establishments. In addition to estimates of total employment by industry, the Bureau publishes data on production, construction, or nonsupervisory worker employment. For this purpose, the sample ratio for the current month of produc tion workers to total employment is used. For example, the 60 sample firms which had 26,000 employees in August, reported an August pro duction-worker figure of 19,500 resulting in a 19,500 ratio of or .750. Using this ratio, pro26,000 duction-worker employment in August is esti mated to be 39,000 (52,000 multiplied by .750= 39,000). A similar ratio method is used to esti mate the number of women employed.1 0 The estimates for each type of series (all em ployees, production workers, and women em ployees) for individual estimating cells are summed to obtain the corresponding totals for broader industry groupings and divisions. Appropriate revisions, based on new bench marks, are introduced into the employment series as required to correct for classification changes and for deviations resulting from the use of sample trends. In general, the bench mark month is March. The employment esti mates which had been published previously for that month are compared with the new bench 1 If permanent changes in the composition of the sample 0 take place, the “ production-workers, all-employee” ratios and the "w om en” ratios calculated from the sample are modified by a wedging technique described in this chapter under "Hours and Earnings.” (See p. 25.) mark data. The amount of adjustment in the published employment information is indicated by this comparison. The all-employee series, for months between the current and the last preceding benchmark, are adjusted by wedging or tapering out the difference between the cur rent benchmark and the estimate for the bench mark month back from the current benchmark to the last previous benchmark. This difference is assumed to have accumulated at a regular rate. The series for months subsequent to the benchmark month are revised by projecting the level of the new benchmark by the trend of the unadjusted series. A comparison of the amounts of the re visions made since 1966, is presented in table 1. Table 1. Nonagricultural employment estimates, by industry division, as a percentage of the benchmark for recent years Industry division 1966 1967 1968 1969 1970 __________ Total Mining ... ____ .... Contract construction ............ Manufacturing ____________ Transportation and public utilities _ ____ ___ _ Wholesale and retail trade .... Finance, insurance, and real estate __ _______ _____ _ Services ___ __ ___ ____________ Government _________________ 99.9 100.5 99.7 99.4 100.0 99.5 101.6 99.5 100.4 101.7 99.5 99.8 99.8 101.5 99.0 99.8 100.0 100.0 100.1 100.1 99.7 100.1 99.8 100.7 100.7 100.3 100.4 100.0 100.0 100.1 99.5 100.3 100.0 100.2 99.8 100.0 99.2 99.2 102.8 100.0 99.1 100.1 100.3 99.6 100.2 Hours and earnings Independent benchmarks are not available for the hours and earnings series. Conse quently, the levels shown are derived from the BLS reporting sample. Since 1959, when benchmark data stratified by employment size became available through the employment security system, estimates of employment, hours, and earnings have been prepared by a cell structure which makes use of size and in some cases regional stratification. Experience in the preparation of current esti mates shows that the six size classes as de scribed under the sampling design can be combined into a maximum of three size classes for the purposes of preparing current estimates of hours and earnings, when stratification by EMPLOYMENT, HOURS, AND EARNINGS size is needed.1 Whenever a new national 1 benchmark becomes available, national esti mates of average weekly hours and average hourly earnings using eight size strata and four regional strata (Northeast, North Cen tral, South, and West) are prepared. These estimates are used as a standard against which the published averages are compared. In some cases, this comparison indicates that some modification of the pattern of stratification is needed. If this is the case, a change is intro duced into the estimating structure at the time of the next benchmark revision. 25 Note is then taken of the published estimate of average hourly earnings for the previous month, say V0. Because the panel of establish ments reporting in the sample is not absolutely fixed from month to month, there may be differ ences between V0 and U0.1 A final figure for 1 2 the current month is obtained by making use of both pieces of information; the estimate is F1==(0.9 Vo + 0.1 U0) - h ( U 1- U 0) The procedure, reflected in this last equation, accepts the advantage of continuity from the use of the matched sample, and at the same time tapers or wedges the published estimate a. Average weekly hours and gross hourly toward the level of the latest sample average. earnings. To obtain average weekly hours for The same procedure is used to adjust the an individual estimating cell, the sum of the production-worker all-employee-ratio and the man-hour totals reported by the plants classi fied in that cell is divided by the total number Table 2. Number of industries for which “Primary” series are published under the BLS Industry Employ of production workers reported for the same ment Statistics Program— employment, hours, and establishments. Similarly, in computing aver earnings, January 1971 age hourly earnings, the reported payroll total is divided by the reported man-hour total. Hours Pro All Average and The first ratio estimates of average hourly duction Women Industry division em earn overtime hours ployees workers 1 ings 2 earnings and average weekly hours are modi 417 362 402 364 150 Total _____ fied at the estimating cell level by a wedging 12 11 11 9 Mining ______ __ Contract technique designed to compensate for changes 11 11 construction .... 278 11 11 277 277 146 Manufacturing .. 278 in the sample arising mainly from the volun Transportation and public tary character of the reporting. 25 16 21 19 utilities ______ 32 36 35 32 Trade __________ Finance, For example, a first estimate of average insurance, and 9 15 15 9 real estate ..... hourly earnings for the current month, Ult is 23 5 22 5 Services ............ 15 4 Government ___ obtained from aggregates from a matched __ 1 1 1 1 Total private ___ Total nonsample of establishments reporting in the cur 1 1 1 agriculture _ _ rent and previous month. Similarly an estimate 1 Production workers in manufacturing and mining; con of average hourly earnings, U0 for the previ , struction workers in contract construction; nonsupervisory workers, all other divisions. ous month is calculated from the same matched - Average hourly earnings, average weekly hours, and sample. Hence, Ui-U0 is a measure of absolute average weekly earnings. change between the 2 months. ratios for women with regard to changes in the composition of the sample between successive 1 These combinations of size classes have been made be 1 cause of operational economies. In particular, the preliminary months. estimates are based on less than full samples, and if the Weekly hours and hourly earnings for indus estimation of preliminary estimates was attempted using the full stratification pattern, there would be a number of cells tries and groups above the basic estimating for which there were no samples. Experiments and tests over cell level are weighted averages of the figures several years indicate that estimates of hours and earnings prepared from the BLS sample using a maximum of three for component cells and industries. The average size strata generally do not differ significantly from those weekly hours for each estimating cell are computed with four size strata or more. 12 If the difference between the estimate and the average multiplied by the corresponding estimate of computed from the sample (V (— U0), is too great, the sample 1 production-worker employment, to derive ag average is accepted once it has been established that the difference is due to a permanent change in the composition of gregate man-hours. Payroll aggregates are the the sample, and the series is regarded as discontinuous at product of aggregate man-hours and average that point. In general, a difference greater than 3 percent is considered as defining a discontinuity or “ break.” hourly earnings. Payroll and man-hour aggre 26 BLS HANDBOOK OF METHODS gates for industry groups and divisions repre sent the sum of aggregates for component industries. Average weekly hours for industry groups are obtained by dividing the man-hour aggre gates by the corresponding production-worker employment estimates. Average hourly earn ings for groups are computed by dividing the payroll aggregates by the man-hour aggre gates. This method is equivalent to weighting weekly hours by estimated universe productionworker employment and hourly earnings by estimated universe man-hours. For all levels, from individual estimating cells to major industry divisions, average weekly earnings are computed by multiplying average hourly earnings by average weekly hours. b. Overtime man-hours. To obtain average weekly overtime hours in manufacturing in dustries, the sum of the overtime man-hours reported is divided by the number of produc tion workers in the same establishments. c. Spendable average weekly earnings. Be fore the majority of workers in lower income brackets were subject to Federal income and social security taxes, gross average weekly earnings were a satisfactory measure of trends in weekly earnings available for spending. After Federal income taxes began to affect the spendable earnings of an appreciable number of workers, a method was developed for ap proximating spendable earnings by deducting these taxes from gross earnings.1 3 The amount of individual income tax liability depends on the number of dependents sup ported by a worker as well as on the level of his gross income. Spendable earnings for work ers by major industry division are computed and published for a worker with no dependents and a worker with three dependents. Gross and spendable weekly earnings also are computed and published in terms of 1967 dollars, to give an approximate measure of changes in “ real” gross and spendable weekly 1 Utter, Carol, “The Splendable Earnings Series: A Technical :5 Note on its Calculation,” Employment and Earnings and Monthly Report on the Labor Force, February 1969, pp. 6-13. earnings, that is, in purchasing power since that base period. This series is computed by dividing the weekly earnings average (in cur rent dollars) by the BLS Consumer Price Index for the same month. d. Average hourly earnings, excluding over time, in manufacturing are computed by divid ing the total production-worker payroll for the industry group by the sum of total productionworker man-hours and one-half of total over time man-hours, which is equivalent to pay rolls divided by straight-time man-hours. This method excludes overtime earnings at 11/2 times the straight-time rates; no further ad justment is made for other premium payment provisions. e. Indexes of aggregate weekly man-hours and payrolls are prepared by dividing the cur rent month’s aggregates by the average for 1967. Reliability of estimates Although the relatively large size of the BLS establishment sample assures a high degree of accuracy, the estimates derived from it may differ from the figures that would be obtained if it were possible to take a complete census using the same schedules and procedures. As discussed previously a link relative technique is used to estimate employment. This requires the use of the previous month’s estimate as the base in computing the current month’s esti mate. Thus, small sampling and response errors may cumulate over several months. To remove this accumulated error, the estimates are adjusted annually to new benchmarks. In addition to taking account of sampling and response errors, the benchmark revision ad justs the estimates for changes in the indus trial classification of individual establishments (resulting from changes in their product which are not reflected in the levels of estimates until the data are adjusted to new benchmarks). In fact, at the more detailed industry levels, par ticularly within manufacturing, changes in classification are the major cause of benchmark adjustments. Another cause of differences, gen erally minor, arises from improvements in the quality of the benchmark data. EMPLOYMENT, HOURS, AND EARNINGS One measure of the reliability of the employ ment estimates for individual industries is the root-mean-square error (RM SE). The measure is the standard deviation adjusted for the bias in estimates. RMSE—/(Standard Deviation)2-/(B ia s )2 If the bias is small, the chances are about 2 out of 3 that an estimate from the sample would differ from its benchmark by less than the rootmean-square error. The chances are about 19 out of 20 that the difference would be less than twice the root-mean-square error. The hours and earnings estimates for cells are not subject to benchmark revisions, al though the broader groupings may be affected slightly by changes in employment weights. The hours and earnings estimated, however, are subject to sampling errors which may be expressed as relative errors of the estimates. (A relative error is a standard error expressed as a percent of the estimate.) Measures of rootmean-square errors for employment estimates and relative errors for hours and earnings esti mates are provided in the “ Technical Note” of Employment and Earnings. Seasonally adjusted series Many economic statistics, including employ ment and average weekly hours, reflect a regularly recurring seasonal movement which can be measured on the basis of past experi ence. By eliminating that part of the change which can be ascribed to usual seasonal varia tion, it is possible to observe the cyclical and other nonseasonal movements in these series. Seasonally adjusted series are published reg ularly for selected employment hours, and earnings series. The seasonal adjustment method used for these series is an adaptation of the standard ratio-to-moving average method, with a pro vision for “ moving” adjustment factors to take account of changing seasonal patterns. A de tailed description of the method is given in appendix A of this bulletin. The seasonally adjusted series on gross aver age weekly hours, average overtime hours and average hourly earnings are computed by ap plying factors directly to the corresponding 27 unadjusted series, but seasonally adjusted em ployment totals for all employees and produc tion workers by industry divisions are obtained by summing the seasonally adjusted data for component industries. Selected seasonally ad justed series also are prepared for aggregate weekly man-hours. Presentation and Uses At the national level, the program produces each month a total of over 2,600 separate pub lished series. Tables 2, 3, and 4 provide a sum mary of the detail which is published currently. Table 2 describes the “ primary” series pro duced by the program, that is, those computed directly from the sample and benchmark data. Table 3 indicates the “ special” series which are obtained from the primary series by ap plication of special adjustments, while table 4 lists the seasonally adjusted series by type and industry division. In addition to the series published on a cur rent monthly basis, a single annual figure for employment in March of each year (based on benchmarks) is published for a number of in dustries for which monthly estimates do not currently meet established standards for pub lication. In 1971, following revision to the 1970 benchmark, data for 239 such industries were published. In April 1971, employment and hours and earnings statistics were available for 50 States, the District of Columbia, and 211 areas. Ap proximately 8,200 employment series and hours and earnings series for about 3,400 industries were published for these States and areas by the State agencies. The employment series usually covered total nonagricultural employ ment, major industry divisions (e.g., contract construction, manufacturing), and major in dustry groups (e.g., textile mill products, transportation equipment, retail trade) for each State and area. Additional industry detail frequently is provided for the larger States and areas, particularly for industries which are locally important in the various jurisdictions. The series on employment, and hours and earnings appear in several BLS publications. 28 BLS HANDBOOK OF METHODS Table 3. Number of industries for which special series are published under the BLS Industry Employment Statistics Program— employment, hours, and earnings, January 1971 Industry division Total private .. Mining, construction and manufacturing Mining _____ ____ Contract construction Manufacturing .. Transportation and public utilities ______ Trade ..... ........... . Finance, insurance and real estate ________ Services _____ Index of Index o f aggreaggregate gate weekly weekly manpayhours rolls Spendable average weekly earnings 1 Gross weekly earnings (1 9 6 7 dollars) 1 Average hourly earnings ( excluding overtim e) 1 1 1 1 1 1 1 24 I 1 1 1 1 1 1 1 1 1 1 1 1 1 23 1 In current and 1967 dollars. The summary data are first published each month in a press release which contains pre liminary national estimates of nonagricultural employment, weekly hours, and gross average weekly and hourly earnings in the preceding month, for major industry categories. The re lease also includes seasonally adjusted data on employment, average weekly hours, and aver age overtime hours. The preliminary estimates are based on tabulations of data for less than the full sample to permit early release of fig ures. This release normally is issued 3 weeks after the week of reference for the data. The press release also includes a brief analysis of current trends in employment, hours, and earn ings, pointing up current developments as com pared with those for the previous month and the same month in the preceding years. National estimates in the detail described in tables 2, 3, and 4 are published in the monthly report Employment and Earnings. This pub lication is issued about 5 weeks after the week of reference. Employment data for total non agricultural employment and for the major industry divisions, as well as hours and earn ings for all manufacturing, are published for States and areas in Employment and Earnings 1 month later than those for the Nation. Special articles analyze long-term economic movements or describe technical developments in the pro gram. Many of the national series are repub lished in the Monthly Labor Review with data shown for each series for the most recent 13 months. Following each benchmark revision, an his torical volume called Employment and Earn ings, United States is published. This provides historical data, monthly and annual averages, from the beginning date of each series, in a few instances as far back as 1909. A companion volume, Employment and Earnings, States and Areas, provides historical data (annual aver ages) on all employees and on productionworker hours and earnings series published by State agencies for States and areas back to the beginning of these series, in some instances to 1939. This volume is published annually. De tailed industry rates are available monthly in releases published by the cooperating State agencies. The data are disseminated also through the publications of many other Federal agencies; e.g., the Department of Commerce, the Board of Governors of the Federal Reserve System and the Council of Economic Advisors repub lish all or part of the data. They are also reg ularly republished in summary form or for specific industries in many trade association journals, the labor press, and in general refer ence works. These series are used by labor unions, busi ness firms, universities, trade associations, private research organizations, and many gov ernment agencies. Research workers in labor unions and industry, as well as others responsi ble for analyzing business conditions, use the trends reflected in these particular statistics as economic indicators. The average weekly hours series are utilized as lead indicators of swings in the business cycle. Labor economists and other social scientists find these series to be an important indicator of the Nation's eco nomic activity, as well as a measure of the well-being of the millions of Americans who depend on salaries and wages. Industrial growth and progress may be assessed by using the employment and hours series in conjunction with other economic data to yield measures of productivity. Analysts study employment trends to detect 29 EMPLOYMENT, HOURS, AND EARNINGS Table 4. Number of seasonally adjusted series published under the BLS Industry Employment Statistics Pro gram— employment, hours, and earnings, January 1971 Seasonally adjusted series Industry division A ll em ployees Total nonagriculture ___________ ______ Total Drivate _____________________ Mining, construction, and manufacturing _______ Mining Contract construction .... ........ .............................. Manufacturing Transportation and public utilities __________ Trade ___________________ _______ Finance, insurance, and real estate ..... Services Government ______ ___ ___ ______ ____ _ __ . . . Pro duction workers 1 1 1 1 1 1 24 24 1 3 1 1 1 Employment Total employment in nonagricultural estab lishments from the “ payroll” survey is not directly comparable with the Bureau’s esti mates of the number of persons employed in nonagricultural industries, obtained from the monthly “ household” survey.1 The “ payroll” 5 1 Nelson, Darrell, “BLS Earnings Statistics for Use in 4 Escalation Agreements,” Employment and Earnings and Monthly Report on the Labor Force, March 1968, pp. 18-21. See chapter 1 for a description of this survey. — Average over time hours _ .... 1 1 1 1 1 1 1 1 1 3 1 1 1 Man hour index .... 1 1 1 1 Limitations Average weekly hours 1 1 changes in industrial structure, and to observe growth and decline proclivities of individual industries. They also are used in the Bureau’s Occupational Outlook program as a basis for projection of future trends. Executives use the employment, earnings, and hours data for guidance in plant location, sales, and purchases. Also, firms negotiating long-term supply or construction contracts often utilize series on average hourly earnings as an aid in reaching an equitable agreement; “ escalation clauses” may be included in the contracts, which permit an increase or a lower ing of the settlement price depending on the movement of average hourly earnings in a selected industry.1 Wide need has been demon 4 strated by both labor and business for industry series on hourly earnings and weekly hours, to provide a basis for labor-management negotia tions. They not only furnish current and his torical information on a given industry but provide comparative data on related industries. Index Average Average weekly hourly of earn employ earn ings ment ings 1 1 1 1 1 24 24 3 — — 1 3 1 1 series excludes unpaid family workers, domes tic servants in private homes, proprietors and other self-employed persons, all of whom are covered by the household survey. Moreover, the “ payroll” series counts a person employed by two or more establishments at each place of employment, while the “ household” survey counts him only once, and classifies him accord ing to his single major activity. Certain per sons on unpaid leave are counted as employed under the “ household” survey, but are not in cluded in the employment count derived from the “ payroll” series. In addition to these differ ences in concept and scope, the surveys employ different collection and estimating techniques. Therefore, although each survey measures changes in employment, direct comparability should not be expected. However, over time, the trends are similar. The household survey places its primary emphasis on the employment status of individuals and also provides a great deal of information on the demographic charac teristics (e.g., sex, age, race) of the labor force. The survey is not well suited to providing de tailed information on the industrial and the geographic distribution of employment. The establishment survey, while providing limited information on personal characteristics of workers, is an excellent vehicle for obtaining these detailed industrial and geographic data, and in addition, it provides hours and earnings information which is directly related to the em ployment figures. The payroll and household surveys therefore may be regarded as comple mentary. Employment estimates derived by the Bu- 30 BLS HANDBOOK OF METHODS reau of the Census from its quinquennial censuses and from the annual sample surveys of manufacturing establishments may differ from BLS employment statistics. The most im portant reason for difference stems from the degree to which multiproduct establishments file separate or combined reports in one survey but not the other, which m^y result in different industrial classification of employment. There is also a significant difference at the more de tailed industry levels, since Census classifies auxiliary units and central and district ad ministrative units on the basis of the most appropriate 2-digit major group, while BLS codes these units to the most appropriate 4-digit industry. For broad categories, how ever, the two surveys do show similar levels and trends. Hours and earnings The workweek information relates to aver age hours paid for, which differ from scheduled hours or hours worked. Average weekly hours reflect the effects of such factors as absentee ism, labor turnover, part-time work, and strikes. The gross average hourly earnings series reflect actual earnings of workers, including premium pay. They differ from wage, rates, which are the amounts stipulated for a given unit of work or time. Gross average hourly earnings do not represent total labor costs per man-hour for the employer, for they exclude retroactive payments and irregular bonuses, various welfare benefits, and the employer’s share of payroll taxes. Earnings for those employees not covered under the production worker and nonsupervisory-employee cate gories are, of course, not reflected in the estimates. The series on spendable weekly earnings measure the net earnings of workers who earn the average gross weekly earnings, have the specified number of dependents, and take the standard deductions for Federal income tax purposes. Spendable earnings reflect deductions only for Federal income and social security taxes (calculated on the basis of total annual liabilities), and thus represent only a rough approximation of disposable earnings.1 They 6 do not take into account payroll deductions for such purposes as State income taxes, union dues, or group insurance, and they do not re flect such factors as total family income or tax deductions above the standard amount. The “ real” earnings data (those expressed in 1967 dollars), resulting from the adjustment of gross and spendable average weekly earn ings by means of the Bureau’s Consumer Price Index, indicate the changes in the purchasing power of money earnings as a result of changes in prices for consumer goods and services. These data cannot be used to measure changes in living standards as a whole, which are affected by other factors such as total family income, the extension and incidence of various social services and benefits, and the duration and extent of employment and unemployment. To approximate straight-time average hourly earnings, gross average hourly earnings are adjusted by eliminating only premium pay for overtime at the rate of time and one-half. Thus, no adjustment is made for other premium pay ment provisions such as holiday work, lateshift work, and premium overtime rates other than at time and one-half. The ultimate goal of the program is to pro vide current estimates of employment, hours, and earnings for all nonagricultural industries in the Nation as a whole, and also for all sig nificant industries in all States and all Stan dard Metropolitan Statistical Areas, as defined by the Office of Management and Budget. While very substantial progress toward this objective has been made over the years, and particularly since the end of World War II, there remain some important areas where the goal is yet to be realized. Efforts constantly are being di rected toward strengthening the sample so that series for employment, hours, and earnings for additional industries may be published, and also toward developing series for additional standard metropolitan areas. 1 For a complete analysis of the difference between spend 6 able and disposable earnings, see Paul Schwab, "T w o Mea sures of Purchasing Power Contrasted,” Monthly Labor Re view, April 1971. BLS Codes Form BLS 790 A S ta te R e p o rt No. In d . L O C A T IO N O F E S T A B L IS H M E N T S C O V E R E D IN TH IS R E P O R T (Num ber of establishments) (C ity) (County) (State) .................................................. .................................................. B e fo re en terin g d ata see exp la n a tion s on oth er side P A Y P E R IO D ALL E M P L O Y E E S The Bureau o f Labor S ta tistics and the State agencies coop erating in its statistical p rogram s will hold all in form a tion furnished by the respondent in strict confidence. P R O D U C T IO N A N D R E L A T E D W O R K E R S YOUR From— Through (Both date incluiive) (1) (2 ) (3) Knter the number of days worked plus paid holidays and paid va cation days for majorit y of p rod u ction w orkers. ( N ea rest <lay) Vi During the entire pay period (4) During the 7 consecutiveday period which in cludes the 12th (5) NUM BER DO NOT USE Include all persons who worked during or received pay for any part of period regardless of type of work performed Knter in these columns the number of production and re lated workers who worked during or received pay for any part of the period reported, the p a y earned (before deductions), and all hours worked or paid for. Include pay and man-hours for overtime, sick leave, holidays, and vacations L/P Both sexes (« ) Women only (8) T o ta l P r o d u c t io n W orker P ayroll N um ber or P r o d u c t io n W orkers (») (7) 1970 D ec. 1971 J an .. T o tal P r o d u c t io n W orker M a n -H o u r s COM M ENTS Enter in column 13 the main factors responsible for significant month-to-month changes in employment, average hours worked (col. 11+col. 9), average hourly earnings (ool. 10+col. 11), etc., as indicated by this report. Examples are: Wage rate increase More bueineet DO NOT USE ( Omit centi) (Omit fractioni) Expl. Code (10) (11) Strike Fire Overtime Weather If any general wage-rate changes (not Individual changes for length of service, merit, or promotion) have occurred slnoe last month’s report note amount of increase or decrease (as + 2% , — M ), the effective date of the change, and the approximate number of production workers affected. (13) (12) S F eb -.. M a n .. A p r... BLS HANDBOOK OF METHODS Y ear AND M onth Knter beginning anti e n d i n g dates of pay period w h i c h includes th e 12th of t h e month M a y .. >r ' June . J u ly .» A ug... 7 S ept.. O c t ... N o v .. D e c ... (Person to be addressed if questions arise regarding this report) (Position) CO Form BLS 790 E BLS Code. ......................1 ........................ . . J .............................. II...................... S ta te | R ep ort N o. i Ind. L O C A T IO N OF E S T A B L IS H M E N T S ) C O V E R E D ( N u m b e r o f e s t a b li s h m e n t s ) OJ to Before entering data see explanations on other side (C ity ) IN T H I S REPORT (C o u n ty ) (S td tn T he Bureau o f Labor S tatistics and the State agencies co o p e r ating in its statistical p rogra m s will hold all in form ation fu r nished by the respondent in strict confidence. PAY Y ear AND M onth K n tei b e g in n in g and e n d i n g dates of pay l»e r io d w h ic h in c lu d e s th e 1 2 th of th e m o n th D u r in g th e e n t ir e p a y p e r io d ( Hoth date: in c lu s iv e ) (2) (1) D u r in g th e 7 c o n s e c u t iv e d a y p e r io d w h ic h in c lu d e s th e 1 2 th (4) T h rou gh (5) v3) N O N S U P E R V IS O R Y ALL E M P L O Y E E S NUM BER DO N O T USE L /P I n c lu d e a ll p e r s o n s w h o w o r k e d d u r i n g o r r e c e iv e d p a y for a n y p a r t o f p e r io d r e g a rd le ss o f t y p e o f w o r k p e r fo r m e d B o t h sexes W o m e n o n ly K ilter in th ese c o lu m n s the* n u m b e r o f n o n s u p e r v is o r y e m p lo y e e s w h o w o r k e d d u r i n g o r r e c e iv e d p ay fo r a n y p a r t o f th e p e r io d r e p o r t e d , th e p a y e a r n e d ( b e f o r e d e d u c t io n s b u t e x c l u d i n g c o m m is s io n s ,) a n d a ll h o u rs w o r k e d o r p a id fo r . I n c lu d e p a y a n d m a n -h o u r s f o r o v e r t im e , s ic k le a v e , h o lid a y s , a n d v a c a t io n s N um ber or N on su per v is o r y E m ployees (6) (8) (?) EM PLOYEES \ON8U PERU ISORY KM PLOY EE P ayroll ( E x clu d in g c o m m is s io n s r e jio r te d in c o l . 1 0 A ) ( O m it c e n t s ) T o ta l X o n s u p e r v is o r y Kmployee M a n -H ours ( O m it f r a c t i o n s ) A m ount of c o m m is s io n s (o m it c e n ts) (10) (11) ( 10A ) (») D O N O T USE C o m m is s io n s o r N o n s u p e r v is o r y E m ployees P e r io d in w h ic h earn ed PR |T hrough 1 ( Hoth d a tes in c lu s iv e ) (1 0 B ) 1 (1 0 C ) H E xpl code (111) (12) F rom - (10*) 1970 $ $ ..................... 1971 M a r .. Apr ...................................... M ay June . J u ly.. Aug < N('pt-. D e c ... YOUR COMMENTS ON CHANGES IN EMPLOYMENT, PAYROLL. OR WAGE RATES Enter in colum n 13 the main factors responsible for significant m onth-to-m onth changes in the report above. Examples are: W a g e r a te increase, more b u s in e s s , fir e , te m p o r a r y s u m m e r h elp , o v e r tim e , s tr ik e , w ea th er . If any G E N E R A L W A G E -R A T E C H A N G E S (not individual changes for length of service, merit, or p rom otion ) have occurred since last m onth’s report, note the am ount of increase or decrease (as + 2 % , — 5^), the effective date of the change, and the approxim ate number o f nonsupervisory em ployees affected. _________________________________________________________________ (13) (13) (13) 1971 Sept. 1971 1971 .................................. O c t...................................... Feb July * R r.............................................. .................................................................... (Person to be addressed if questions arise regarding this report) ........... ................... ......................... ............... .................. A ug......... ............................................................................................... N o v ............................................................... D ec. ............ ....... ......................... (Position) .................... EM P LO YM EN T , HOURS, A N D E A R N IN G S From P E R IO D K ilte r th e n u m b e r o f d a y s w o r k e d p lu s p a id h o l i d a y s a n d p a id v a c a t io n d a y s f o r m a jo r i t y o f n o n s u p e r v is o r y e m p lo y e e s . ( X ca rest \ (l a y ) BLS HANDBOOK OF METHODS 33 BLS 790 Industry Class Supplement Budget Bureau No. 44-R745 Approval expires January 31, 1972 STATEMENT OF PRODUCTS MANUFACTURING Return this form as soon as possi ble in the enclosed envelope which requires no postage. U.S. DEPARTMENT OF LABOR BUREAU OF LABOR STATISTICS WASHINGTON, D.C. 20212 r SAMPLE COPY The Bureau of Labor Statistics and the State agencies cooperating in its statistical programs will hold all information furnished by the respondent in strict confidence. L BLS CODES J State Report No. Ind. Empl. Mo. Yr. L O C A T IO N ________________________________________________________ (City) (County) (State) UI CODE Proposed Ind. Ind. Mo. Yr. This report will be used to insure the proper industrial classification of your regular MONTHLY REPORT ON EMPLOYMENT, PAYROLL, AND HOURS and should cover the entire activity of the same establishment. Classification will be by industry on the basis of the principal product or activity of your establishment during the calendar year 1970. Describe your processes or goods produced in your own words, making the distinctions requested on the list of special characteristics provided on the enclosed sheet. This list is not complete but represents the kind of information which should be reported. P rin cipa l P roducts or P ercent o r T otal S ales V alue or R eceipts D uring 1970 A ctivities D uring 19 7 0 (L is t item s separately) P rin cipa l M aterials U sed ( F or each product listed in colum n ( a ) ) (b) gg M j (a ) 1A. Manufacturing (Specify below) W a s M aterial U sed P roduced I n T h is Est a bu sh m e n t ? (d ) (c ) Yes ? i No ^ 0 % % % % % IB. Nonmanufacturing (Specify below) % , 3. Is the establishment primarily engaged in performing services for other units of the % company? % Yes □ No □ If “ Yes,” indicate nature of activity of this establishment: % □ Central administrative office % □ Research, development, or testing □ £ . ' ----- - J ~_________ L ,. IIS Combined Total 2. Is this establishment part of a multiunit company ? □ Yes □ No 100% 4. Storage (warehouse) □ Other (Specify: power plant, etc.) Space for Your Comments. If “Yes,” enter name and location of controlling company. (Person to be addressed if questions arise regarding this report) ☆ U.S. GOVERNMENT PRINTING OPF1CE: 1970— 0 - 3 9 8 - 1 6 4 (P osition ) 34 EMPLOYMENT, HOURS, AND EARNINGS Technical References Number 1. Armknecht, Paul A . Jr., “ BLS Establishment Estimates Revised to March 1970 Benchmark Levels,” Employment and Earnings, September 1971, pp. 18-30. 2. “ The Calculation and Uses of the Spendable Earnings Series,” Monthly Labor Review, April 1966, pp. 405-409. 3. Mendelssohn, Rudolph C. “ Machine Methods in Employment Statistics,” Monthly Labor Re view, May 1955, pp. 567-569. A description of the use of electronic data-processing equipment in the preparation of employment statistics, with particular reference to screening employers’ reports for errors. 4. President’s Committee to Appraise Employment and Unemployment Statistics, Measuring Em ployment and Unemployment, 1962. A comprehensive review and critique of the methods and concepts used by various Fed eral Government programs providing statistics on employment, unemployment, and the labor force in the United States. 5. Nelson, Darrell R., “ BLS Earnings Statistics for Use in Escalator Agreements,” Employment and Earnings and Monthly Report on the Labor Force, March 1968, pp. 18-21. A discussion of the availability, uses, and limitations of BLS average hourly earnings series in escalation agreements. 6. Schechter, Samuel, “ The 1959 Benchmarks for the BLS Payroll Employment Statistics,” Monthly Labor Review, December 1962, pp. 1385-1392. A detailed description of the sources and construction of BLS employment benchmarks. 7. Schwab, Paul M., “ Two Measures of Purchasing Power Contrasted,” Monthly Labor Review, April 1971, pp. 3-14. An examination of the divergent trends in real net spendable earnings and real per capita disposable income. 8. Utter, Carol M., “ The Spendable Earnings Series: A Technical Note on its Calculation,” Em ployment and Earnings and Monthly Report on the Labor Force, February 1969, pp. 11-21. 9. Weinberg, Edgar, “ BLS Earnings Series as Applied to Price Escalation,” Monthly Labor Review, July 1952, pp. 57-59. A discussion of the use of BLS average hourly earnings series in escalation clauses in contracts. 10. Wymer, John P., “ The Revised and Expanded Program of Current Payroll Employment Sta tistics,” Employment and Earnings, November 1961, pp. iv-vii. A description of the impact of a major benchmark adjustment and of important tech nical innovations on the industry employment statistics series. 11. Young, Dudley E. and Sidney Goldstein, “ The BLS Employment Series and Manufacturing Reporting Practices,” Monthly Labor Review, November 1957, pp. 1367-71. A discussion of the findings in a survey analyzing the response patterns of manufactur ing establishments cooperating in the industry employment statistics program. Pa u l A . A r m k n e c h t , Jr . Chapter 3. Job Vacancies and Labor Turnover Background Job vacancies, as used by the Bureau of La bor Statistics, refers to the number of unfilled wage and salary jobs available to workers at the end of a month. Labor turnover refers to the gross movement of wage and salary work ers into and out of employment status with respect to individual establishments during the month. The current job vacancy-labor turnover pro gram began in January 1969 with the addition of questions about job vacancies to the labor turnover questionnaire. The labor turnover program, however, has been in existence for many years. In January 1926, the Metropolitan Life Insurance Co. began the collection of labor turnover data from a small sample of manu facturing establishments. By February 1927, the sample included 175 establishments with 800,000 employees, which was estimated to be about 8 to 10 percent of total manufacturing employment at the time. The original purpose of this series was to provide personnel man agers with national figures on labor turnover rates for manufacturing industries against which they could measure the experience of their own plants. Between November 1927 and July 1929, the Metropolitan Life Insurance Co. published labor turnover rates for total manu facturing. By the latter date, the company felt the project was sufficiently successful and well established to warrant turning it over to the Bureau of Labor Statistics for further develop ment. A decade later, in December 1939, series on labor turnover rates were being published for 30 manufacturing industries, and the sam ple upon which the rates for all manufacturing were based contained 5,500 establishments and nearly 2,600,000 employees. For a number of years, State employment security agencies affiliated with the Bureau of Employment Security (now the U.S. Train ing and Employment Service) had collected labor turnover information for use in job market analysis and as a guide for the op erations of the State employment services. Cooperative arrangements between these agen cies and the Bureau of Labor Statistics for the joint collection of labor turnover data began with an agreement with Connecticut in 1954. By 1964, the cooperative program had been extended to cover all 50 States and the District of Columbia. Experimental programs to determine the feasibility of collecting job vacancy informa tion were conducted in 1965 and 1966. The 1965 study was carried out in 16 metropolitan areas and the 1966 study was conducted in 14 metropolitan areas and three States. Following the successful completion of these programs, cooperative arrangements for the collection of job vacancy data were set up between the BLS and 48 of the State employment services, through the Manpower Administration. In December 1970, these agencies published about 8,000 labor turnover series in manu facturing and mining industries for State and areas, and about 130 job vacancy area series in manufacturing industries. These rates were based on a sample of approximately 32,000 reports in manufacturing and about 1,200 in mining. Description o f the Survey Labor turnover actions are divided into two broad groups: accessions or additions to em ployment, and separations or terminations of employment. These two broad groups are fur ther divided; accessions into new hires and other accessions, and separations into quits, discharges, layoffs, and other separations. La bor turnover is expressed in the BLS series as a monthly rate per 100 employees. Separate rates are computed for each of the component items. 35 36 BLS HANDBOOK OF METHODS The primary difference between types of separations is whether action is initiated by the employee or employer, i.e., whether it is voluntary on the employee’s part or involun tary. Voluntary actions— quits— are initiated by the employee for an almost unlimited va riety of reasons, financial, personal, or social, (e.g., lack of housing and transportation, poor community facilities, etc.). Involuntary actions either may be initiated by the employer or be beyond the control of both employer and em ployee; these actions may arise from economic causes such as business conditions, physiolog ical reasons such as aging, or performance reasons such as incompetence. Job vacancy data are collected both for total job vacancies and for vacancies which have continued unfilled for 30 days or more. Data on the occupations for which the vacancies exist and the wage rates offered for them are also col lected quarterly. Vacancy data are expressed as a monthly rate equal to job vacancies divided by the sum of vacancies and employment, with the quotient multiplied by 100. Concepts Separations are terminations of employment of persons who have quit or been taken off the rolls for reasons such as layoff, discharge, re tirement, death, military service expected to last more than 30 consecutive calendar days, physical disability, etc. Since January 1959, transfers of employees to other establishments of the same company also have been classified as separations. Quits are terminations of employment initi ated by employees for any reason except re tirement, transfer to another establishment of the same firm, or service in the Armed Forces. Included as quits are persons who failed to report after being hired (if previously counted as accessions), and unauthorized absences which, on the last day of the month, have lasted more than 7 consecutive calendar days. Layoffs are suspensions from pay status (lasting or expected to last more than 7 con secutive calendar days), initiated by the em ployer without prejudice to the worker, for reasons such as lack of orders, model change over, termination of seasonal or temporary employment, inventory-taking, introduction of labor saving devices, plant breakdown, or shortage of materials. Discharges are terminations of employment initiated by the employer for such reasons as incompetence, violation of rules, dishonesty, laziness, absenteeism, insubordination, failure to pass probationary period, etc. Other separations include terminations of employment for military duty lasting or ex pected to last more than 30 days, retirement, death, permanent disability, failure to meet the physical standards required, and transfers of employees to another establishment of the company. Accessions are all permanent and temporary additions to the employment roll, whether of new or rehired employees. Transfers from another establishment of the same company also are counted as accessions (beginning with January 1959). New hires are permanent and temporary additions to the employment roll of persons who have never before been employed by the establishment, and former employees rehired although not specifically recalled by the em ployer. This category excludes transfers from other establishments of the same company and employees returning from military service or unpaid leaves of absence. Other accessions include all additions to the employment roll other than new hires. Job vacancies are defined as vacant jobs which are immediately available for filling, and for which the firm is actively trying to find or recruit workers from outside the firm. “ Actively trying to find or recruit” means that the establishment is engaged in current efforts to fill the job vacancies. Long-term job vacancies are those current vacancies which have continued unfilled for 30 days or more. The reporting establishment is also asked to indicate the number of openings with future starting dates for which the firm is actively trying to recruit workers from outside the firm. Job openings with future starting dates may exist for such reasons as: Job unavailable until expected separation of present incumbent JOB VACANCIES AND LABOR TURNOVER occurs; work will not start until some future date; new branch to be opened in the future; or anticipated increase in business. Industry Classification The classification system used for compiling and publishing rates is that described in the 1967 Standard Industrial Classification Manual issued by the Office of Management and Bud get. (See appendix B of this bulletin for a de tailed description of this system.) Reporting establishments are classified on the basis of major product or activity as de termined by annual sales data for the previous calendar year. Most establishments in the job vacancy-labor turnover sample also report em ployment, hours, and earnings under the Bu reau’s industry employment statistics program, and are assigned the same industry classifica tion in both programs. Further discussion of industry classification in the two programs is given under the heading, Industrial Classifica tion in chapter 2 of this bulletin. Occupational Classification Occupational classifications are made in accordance with those established in the Dic tionary of Occupational Titles, Third Edition, U.S. Department of Labor, 1965. These classi fications are the same as those used in State employment service operations. Data Sources Each month cooperating State employment security agencies collect data on the number of job vacancies and on labor turnover actions from a sample of establishments drawn from a list of those subject to State unemployment insurance programs. In nonmanufacturing, supplemental sources are also used to obtain lists of establishments which are not covered by Unemployment Insurance laws. (See chap ter 2, p. 17 of this bulletin.) The respondent ex tracts the figures largely from his personnel records, though some smaller establishments which do not maintain special personnel rec ords utilize their payroll records in making out 37 the reports. Response analysis surveys, which analyzed the reporting practices of a scien tifically selected sample of the establishments in the job vacancy-labor turnover panel, showed that while some employers did not re port the figures for all items precisely as re quested on the schedule, the effect of these deviations on the published data appeared to be quite insignificant, particularly for the broader classes, such as total accessions, total separations, and total job vacancies. Collection Methods Job vacancy and labor turnover data are collected primarily at the State level by em ployment security agencies from cooperating employers via the medium of a mailed “ shuttle” schedule, U.S. Department of Labor form 1219. (See pp. 38 and 39 for a facsimile of this sched ule.) The same form is returned to the re spondent each month of the year for the entry of current data. The respondent reports the total number of job vacancies, the number of actions for each turnover item during the cal endar month and total employment. These em ployment figures, which are the bases used to compute the rates, represent the number of per sons who worked or received pay for any part of the pay period (usually 1 week) which in cludes the 12th of the month. Information on the occupations for which job vacancies exist and the pay rate being offered for them is collected quarterly on a supplemental schedule, form DL 1219A. The State agency uses the information pro vided on the schedule to develop job vacancy and labor turnover rates for the States and for metropolitan areas, and forwards the data to Washington, where they are used by the Bu reau of Labor Statistics to prepare rates at the national level. Sampling Sampling is used by BLS for collecting data in its job vacancy-labor turnover statistics pro- 38 BLS HANDBOOK OF METHODS D L 1219 Budget Bureau No. 44— R1064. Approval expires January 31, 1972. State U.S. DEPARTMENT OF LABOR Report No. Ind. Area Local Off. MONTHLY REPORT ON JOB OPENINGS AND LABOR TURNOVER B U R EA U O F L A B O R S T A T IST IC S and THE M A N P O W E R A D M IN IS T R A T IO N W A SH IN G T O N , D.C. 20212 Enter the data requested and return in the enclosed envelope as soon as the inform ation is available each month. CHANGE NAME AND MAILING ADDRESS IF INCORRECT— INCLUDE ZIP CODE r “ I The Bureau of Labor Statistics, the Manpower Administra tion, and the State agencies cooperating in their statistical programs will hold all information furnished by the respond ent in strict confidence. L LOCATION ................... J Before entering data see explanations on other side (C ity) (County) II. EMPLOYMENT (one pay period) I. LABOR TURNOVER DURING CALENDAR MONTH YEAR AND MONTH PERIOD COVERED BY LABOR TURN OVER (Cols. 4 through 11) Preferably one calendar month From— (B oth dates (1 ) 1970 Dec. Through inclusive) (2 ) (3) ALL E M P L O Y E E S TOTAL NUMBER S E P A R A T IO N S (during calendar month) Total Separations (Sum o f cols. 5 through 8) (4) (State) A C C E S S I O N S (during calendar month) Quits Discharges Layoffs Other separations ( 5) (6) (7 ) (8 ) Total Accessions (Sum o f cols. 10 and 11) (9) New hires Other Accessions (10) (11) who worked dur ing or received pay for any part of the pay pericd which include^ the 12th o f the month (12) 1971 Jan. Feb. M a r. Apr. M ay June July A u g. Sept. Oct. N ov. Dec. III. JOB OPENINGS for which workers from outside the firm were actively being sought as of close of last business day of most recent month. (If “ NONE,” enter “0,” do not leave blank.) YEAR AND M TH ON C R E T JOB OPENINGS URN OPENINGS W ITH (unoccupied and ready for imme FTR UUE STA TIN R G diate filling) DTS AE Openings included in DO N OT UE S col. 13 continuing unfilled for 30 days or more (14) Expl. Code (16) Number o f current Job Openings (13) (not included in column 13) (15) IV. YOUR COMMENTS Enter main factors responsible for any significant month-tomonth changes in SECTIONS I, II, and III. Examples are: More business Strike Fire Weather Temporary summer help Seasonal Increase (17) 1970 Dec. 1971 Jan. Feb. M a r. A p r. M ay June July A u g. Sept. Oct. N ov. Dec. ( Person to be addressed if questions arise regarding this rep ort) (Position) 39 JOB VACANCIES AND LABOR TURNOVER INSTRUCTIONS FOR COMPLETING THIS FORM I. LABOR TURNOVER 12th of the calendar month for which labor turnover data are reported. PERIO COVERED.—Information on labor turnover, cols. 4 through D 11, is requested for the most recent entire calendar month speci fied in column 1 or, if this is not possible, for a period which most closely covers that calendar month. In either case enter in cols. 2 and 3 the beginning and ending dates for the monthly period for which turnover data are reported. SEPARATIONS (A L E PLO E S) L M YE Column 4. TO TAL SEPARATIONS DU G CALEN RIN DAR MONTH.,- Enter in column 4 the sum of columns 5 thru 8. Column 5. QUITS.— quit is a termination of employment ini A tiated by the employee for any reason except to retire, to trans fer to another establishment of the same firm, or for service in the Armed Forces. Include a person who fails to report after be ing hired (if previously counted as an accession) and an unau thorized absence if on the last day of the month the person has been absent more than 7 consecutive calendar days. Column 6. DISCH ARGE.— discharge is a termination of employ A ment initiated by the employer for such reasons as incompetence, violation of rules, dishonesty, laziness, absenteeism, insubordina tion, failure to pass probationary period, etc. Inability to meet organization’s physical standards should be reported in other separations, col. 8. Column 7. LAYOFFS.— layoff is a suspension from pay status A (lasting or expected to last more than 7 consecutive calendar days without pay) initiated by the employer without prejudice to the worker for such reasons a s: lack or orders, model change over, termination of seasonal or temporary employment, inven tory-taking, introduction of labor saving devices, plant break down, shortage of materials, etc.; include temporarily fur loughed employees and employees placed on unpaid vacations. Column 8. O E SEPARATIONS.— TH R This group should include only terminations of employment for military duty lasting or ex pected to last more than 30 calendar days, retirement, death, permanent disability, failure to meet the physical standards re quired, and transfers of employees to another establishment of the company. N O T E : If you include any other types of separa tions in this column, mention the number and type under Com ments. Employees involved in labor-management disputes should not be counted as separations. ACCESSIONS (A E PLO E S) LL M Y E Column 9. TO TAL ACCESSIONS DU G CALEN RIN DAR MONTH.-An ac cession is any permanent or temporary addition to the employ ment roll whether of new or former employees, or transfers from another establishment of the company. Enter in column 9 the sum of cols. 10 and 11. Column 10. N EW H E New hires are temporary or permanent IR S.— additions to the employment roll of (1) anyone who has never before been employed in this establishment, or (2) former em ployees you did not call back. Former employees you did call back should be included in total accessions and “other acces sions,” but not in “ new hires.” Persons transferred from other establishments of this company should be reported in “ other accessions.” Column 11. O E ACCESSIONS.— TH R Include all additions to the em ployment roll other than new hires. This includes all employees called back to work by the employer from a layoff as defined for col. 7, transfers from other establishments of the company, and former employees returning from military leave or other ab sences without pay who have been counted as separations. Em ployees involved in labor-management disputes should not be counted as accessions when they return to work. II. EMPLOYMENT PERIO COVERED.—Employment information, col. 12, is requested D for one pay period (preferably one week) which includes the ☆ Column 12. A L EM YEE Enter the total number of persons L PLO S.— (both sexes) on the payrolls of the establishment (s) covered in this report who worked full- or part-time or received pay for any part of the pay period (preferably one week). Include salaried officers of corporations, executives and their staffs, and employees engaged in a force-account construction but exclude proprietors, members of unincorporated firms, and unpaid family workers. Include persons on vacations and sick leave for which they received pay directly from your firm for the pay period covered but exclude persons on leave without company pay the entire period and pensioners and members of the Armed Forces carried on the rolls but not working during the pay period covered. N O TE : If the number differs from the “ All Employees” total reported on the Employment, Payroll, and Hours form, explain under Comments. III. JOB OPENINGS PERIO COVERED.— D Job openings information, columns 13, 14 and 15, is requested as of close of last business day (or nearest pos sible day) of the most recent monthly period for which labor turnover data are reported. Column 13. N M U BER O C R E T JOB OPENINGS (VACANCIES).-Enter F URN the number of current job openings in your establishment. IF “ N O N E ,” E NTER “0.” A current job opening is an existing vacant job in your establishment that is immediately available for filling and for which your firm is actively trying to find or recruit some worker from outside your firm (i.e., a “new” worker— not a company employee). Include such openings for all kinds of positions, classifications and employment (full-time, part-time,' permanent, temporary, seasonal), including those outstanding on orders with employ ment agencies and notifications to unions. Exclude jobs to be filled by recall, transfer, promotion, demo tion or return from paid or unpaid leave; jobs unoccupied be cause of labor-management disputes; job openings for which “ new” workers were already hired and scheduled to start work later; and the “ openings with future starting dates” reported in column 15. “ Actively trying to find or recruit” means current efforts to fill the job opening through orders listed with public or private em ployment agencies and school placement offices; notifications to labor unions and professional organizations; “ help wanted” ad vertising (newspaper, posted notice, etc.); recruitment pro grams ; interview and selection of applicants. Column 14. C R E T JOB OPENINGS CONTINUING U F L D FO 30 URN N IL E R DAYS OR M ORE.— Enter the number of current job openings in cluded in the figure reported in column 13 which have continued unfilled for 30 days or more. IF “ NO N E ,” E N TER “ 0.” Column 15. OPENINGS W ITH F T R STARTIN D TE Enter the UUE G A S.— number of openings in your establishment for which your firm is actively trying to find or recruit some worker from outside your firm (i.e., a “new” w orker), but which relate to jobs that r are currently occupied or unavailable for immediate occupancy by “ new” workers for such reasons as: job unavailable until ex pected separation of present incumbent occurs: work will not start until some future date; new branch to be opened in future or anticipated increase in business. IF “ N O N E ,” E N T ER “ 0.” N O T E : OPENINGS W IT H FUTURE STARTING DATES ARE NOT TO BE INCLUDED IN COLUMN 13. IV. COMMENTS Column 17. YOUR COM ENTS.—Enter the main factors responsible M for significant month-to-month changes in Labor Turnover <cols. 4 through 11), Employment (col. 12), and Openings (cols. 13 through 15). Some examples are listed in the heading of col umn 17. U.S. G O V ER N M EN T PRINTING OFFICE: 1970-0-407-201 40 BLS HANDBOOK OF METHODS gram, since full coverage would be prohibi tively costly and time consuming. The sampling plan for the program must: (a) provide the preparation of reliable monthly estimates of job vacancy and labor turnover rates which can be published promptly and regularly; (b) through a single general system, yield consider able industry detail for metropolitan areas, States, and the Nation; and (c) be appropriate for the existing framework of operating pro cedures, administrative practices, resource availability, and other institutional charac teristics of the program. In developing the sample design, the universe of establishments was stratified first by in dustry and within each industry by size of establishment in terms of employment. Within each industry, an optimum allocation design was obtained by sampling with probability proportionate to average size of employment within each of the strata. The total size of sample regarded as necessary to produce satis factory estimates of employment was dis tributed among the size cells on the basis of average employment per establishment in each cell. In practice, this is equivalent to distribut ing the predetermined total number of estab lishments required in the sample among the cells on the basis of the ratio of employment in each cell to total employment in the industry. Within each stratum, the sample members are selected at random. Under this type of design, large establish ments fall into the sample with certainty. Establishments with 250 or more employees are included in the sample with certainty, al though in some cases the cutoff is lower. The sizes of the samples for various industries were determined empirically on the basis of exper ience. The sample design, although aimed primarily at meeting the needs of the national program, provides a technical framework within which State and area sample designs can be de termined. Since, however, the rates for States and areas are not generally prepared at the same degree of industry detail as the national 1 For the national sample, additional reports needed for State and area samples are added to those required by the national design. rates, the national design usually provides sufficient reports for the preparation of State and area rates.1 Estimating Procedures Both job vacancy and labor turnover rates are estimates of ratios. For individual indus tries, turnover rates are computed by dividing the number of turnover actions of each type, as reported by the sample establishments, by the total number of employees reported by those establishments. The result is multiplied by 100. In an industry sample, for example, 623 employees quit between January 1 and 31, while 30,062 employees worked or received pay during the week of January 11-17. The Jan uary quit rate for the industry is : 623 30,062 X 100- 2* 1 Turnover rates for industry groups are com puted by weighting the rates for the component industries by the estimates of total employ ment, prepared by the BLS industry employ ment statistics program. These estimates, which cover the pay period including the 12th of the month, are described in chapter 2 of this bulletin. Rates for “ all manufacturing” and for the durable and nondurable goods sub divisions of manufacturing are weighted by employment in the major industry groups. Computation of job vacancy rates for indus try stratum also involves a weighting process. The number of vacancies reported by the sam ple of establishments is weighted by the esti mates of total employment in that industry stratum. The weighted number of vacancies is then divided by the sum of employment in the stratum plus vacancies; the quotient is then multiplied by 100 to determine the vacancy rate. Rates for major industry groups, for the durable goods and nondurable goods sub divisions, and for total manufacturing are computed by summing the weighted number of vacancies in the component cells of industries or subdivisions, then dividing by the sum of vacancies and employment in that industry, JOB VACANCIES AND LABOR TURNOVER subdivision or division. Again, the quotient is multiplied by 100 to obtain the rate. As of 1971, size stratification was not used in the preparation of job vacancy and labor turnover rates. Tests were underway to de termine the effect on the rates of introducing size stratification. Preliminary results of the tests suggested that size stratification would improve the labor turnover rates but have little effect on the job vacancy rates. Upon comple tion of the tests, size stratification will be in troduced, if warranted. Seasonally Adjusted Series Many economic statistics, including labor turnover rates, reflect a regularly recurring seasonal movement which can be measured on the basis of past experience. By eliminating that part of the change which can be ascribed to usual seasonal variation, it is possible to observe the cyclical and other nonseasonal movements in these series. Seasonally adjusted labor turnover rates are published at the all manufacturing industry level. The seasonal adjustment method used for these series is an adaptation of the standard ratio-to-moving average method, with a pro vision for “ moving” adjustment factors to take account of changing seasonal patterns. A de tailed description of the basic method is given in appendix A of this bulletin. The 2 years of experience with job vacancy data indicate the emergence of some probable seasonal patterns. However, a minimum of 3 years of data are needed to develop estimates of the seasonal factors. Presentation The BLS publishes, on a national basis, monthly series of labor turnover rates for se lected industries. These series are currently published for the manufacturing division, the durable and nondurable goods subdivisions, 21 major industry groups in manufacturing, 191 individual manufacturing industries, and 7 categories in mining and communications. Rates are available for all manufacturing from 41 January 1930 and for telephone and telegraph from 1943. For industry groups and individual industries in the manufacturing and mining divisions, all series begin with January 1958. Rates for certain highly seasonal industries, for example canning and preserving, are not now published separately but are included in the computation of rates for the major manu facturing groups. Before 1958, these industries and the printing, publishing and allied indus tries major group were not included in the rates for all manufacturing. The rates for all manufacturing for years prior to 1958 were revised, however, to reflect the influence of these industries. Monthly rates for total accessions, new hires, total separations, quits, and layoffs are shown for manufacturing and mining industries. Ex cept for the new hire rates, the same items are published for the telephone and telegraph in dustries. On a national basis, the BLS presently pub lishes monthly estimates of total and long-term job vacancies for the manufacturing division, the durable goods and nondurable goods sub divisions, and nine selected industry groups. Rates are available from April 1969 for each of these divisions and industry groups. Preliminary job vacancy rates for the nine selected major industry groups and turnover rates for the 21 major industry groups in manufacturing are published monthly in a BLS press release about a month after the reference month, and in the Monthly Labor Review 3 months after the reference month. Preliminary turnover rates for both detailed industries and broad categories and prelimi nary job vacancy rates for the same categories as are included in the press release are pub lished in Employment and Earnings about 2 months after the month of reference. Both job vacancy and labor turnover rates for all manufacturing for selected States and metropolitan areas are published each month in Employment and Earnings. More detailed information is available in releases issued by the cooperating State agencies. National labor turnover rates (monthly data and annual averages) back to the beginning of each series are published in the annual volume 42 BLS HANDBOOK OF METHODS called Employment and Earnings, United States. New editions of this volume are pub lished annually, following each adjustment of the Bureau’s industry employment statistics series to new benchmark levels. Uses and Limitations The two major causes of change in both job vacancy and labor turnover rates are industrial expansion and contraction. In prosperous times, job vacancy rates, quit rates, and new hires are high because of job availability; in periods of economic recession, high layoff rates are coupled with low vacancy, quit, and acces sion rates. Turnover rates are, therefore, re garded as good economic indicators and are widely used by economic analysts in both gov ernment and private industry. Together with the turnover data and other labor force data, such as unemployment rates, job vacancy data are also expected to provide an indicator as to the condition of the economy. Labor turnover rates by industry are also valuable for personnel planning and analysis. Employers use these rates as a yardstick against which to measure the performance of their plants. For example, they consider low quit rates to be an indication of efficient opera tions and good labor-management relations. A consideration of turnover is essential for sched uling production and for planning the orderly recruitment and maintenance of an adequate manpower supply. Labor turnover rates are also widely used by State employment services to plan and appraise their operations. Job vacancy data should also prove useful to State employment services. By identifying emerging labor shortages, vacancy data will allow for more intelligent planning of training programs and should be useful in counselling the unemployed and new entrants to the labor market. The data may also make interarea re cruitment of workers possible. The use of turnover or job vacancy rates to interpret changes in the BLS monthly employ ment series is limited for the following rea sons: (1) The labor turnover series measures changes during the calendar month, while the employment series measures changes from midmonth to midmonth; and (2) employees on strike are not counted as turnover actions, al though such employees are excluded from the employment estimates if the work stoppage lasts throughout the report period including the 12th of the month. The Bureau publishes annual averages of job vacancy and labor turnover rates, which are computed as the arithmetic means of the 12 monthly rates. These can provide a useful measure if a 1-month rate is not suitable for some purposes, as for example when the rate for a specific month is considered to be unusual or affected strongly by seasonal influences.2 — S h e il a C. W h it e Bureau 2 Because they are liable to misinterpretation, the does not prepare cumulative annual rates of labor turnover. For example, an annual quit rate could be obtained b y divid ing the total number of quits during the year b y average employment during the year. An approximation of this figure can be obtained by cumulating the 12 monthly rates. Suppose the annual rate thus obtained amounted to 50 per 100 em ployees. This might seem to im ply that 50 percent of all employees in January voluntarily left their jobs b y the end of December. However, many jobs in a given establishment are vacated and refilled more than once during the year. The Bureau does not have information on the number o f em ployees who remained with the establishment during the en tire year. Over short periods of time, labor turnover rates probably include relatively little repetitive counting of em ployees w ho have held the same jobs, while over a period of as long as a year there is considerable duplication. Manpower Structure and Trends Chapter 4. Employment of Scientific and Technical Personnel Background The growth in industrial employment of scientists and engineers has been both a symp tom and a driving force in the rapid technolog ical progress of American industry. For a number of years, a major element in this growth has been the need of the Federal Gov ernment for increasingly complex and tech nologically advanced national defense material and space related research and technology. The surveys of scientific and technical personnel provide a means for estimating current man power resources and a basis for evaluating future requirements for scientific and engineer ing personnel. The surveys of scientific and technical per sonnel were initiated by the Bureau of Labor Statistics in 1959, under the sponsorship of the National Science Foundation. They are part of a comprehensive statistical program, coordinated by the Foundation, designed to yield estimates of the employment of scientists and engineers in all sectors of the U.S. eco nomy. The Bureau of Labor Statistics con ducted surveys of scientific and technical personnel in industry annually during the 1960’s (except for 1965) and related surveys of State government agencies less frequently. Since 1962, the Bureau of Labor Statistics has had full responsibility for the conduct and publica tion of these surveys, which continue to be planned in consultation with the National Sci ence Foundation. The last survey of scientific and technical personnel was conducted in 1970. In the future, estimates of scientific and tech nical personnel employed in private industry will be developed as part of the Bureau’s Oc cupational Employment Statistics Program, which will provide employment statistics on all occupational fields. cialty. Engineers and scientists are further distributed according to whether they are en gaged primarily in research and development activities, in management and administration, in technical sales and service, in production and operations, or in “ all other” functions. Data are published separately for all major industry groups, and in finer industry detail for a selected number of major industry groups. Beginning in 1966, data were developed by geographic area. State agencies are surveyed for information on employment of engineers, scientists, tech nicians, economists, statisticians, psychologists, social workers, and health professionals. Em ployment is tabulated by the various govern mental functions in which the workers are engaged. Data also are published by State when appropriate. For each professional occupation, respond ents are asked to report the number of persons whose current positions require knowledge equivalent at least to that acquired through completion of a 4-year college course with an appropriate academic major, regardless of whether they hold a college degree. These sur veys, thus, cover all persons actually working in one of the designated occupations, specifi cally including those who do not hold an ap propriate degree or any degree at all, but specifically excluding persons trained in the occupation but currently employed in positions not requiring the use of such training. Technicians are defined as persons actually engaged in technical work at a level which re quires knowledge of engineering, mathematical, physical and life sciences, comparable to that acquired through technical institutes, junior colleges, or other formal post-high school train ing less extensive than 4-year college training, or through equivalent on-the-job training or experience. Description o f Surveys The surveys of scientific and technical per sonnel in industry gather data on the employ ment of engineers as a group, and on scientists and technicians by major occupational spe Data Sources Sources of occupational data reported by re spondents are personnel records and especially for the small reporting units, personal knowl 43 44 BLS HANDBOOK OF METHODS edge of persons completing the reports. Dis cussions with a number of large respondents indicate that their records typically contain much of the data in the desired form, but that some adjustments by the respondent are often necessary because the occupational classifica tions used in his records differ somewhat from those specified for the surveys. Employment benchmarks for the survey of scientific and technical personnel in industry are derived from employment data tabulated from the first quarter reports of the unemploy ment insurance program. The survey of State government agencies is based on a compilation of data from all State government agencies employing personnel in occupations covered by the survey. Collection Methods Data are collected from respondents pri marily by mail, but personal visits are made to many large employers, and to other respond ents who indicate particular difficulty in com pleting the questionnaires. These visits, which limited resources have prevented from being either numerous or frequent, are carried out by senior staff members. Normally two mailings follow and a sub-sample of residual nonre spondents are contacted further by telephone. The response to these surveys has been very encouraging. Respondents supplying usable in formation have constituted approximately 80 percent of the reports solicited in virtually every year, and have never been below 75 percent. Sampling The basic sample for the industry survey was drawn from lists of establishments reporting to each of the 51 (State and D.C.) employment security agencies for unemployment insurance (UI) purposes, and was supplemented by a list of interstate railroads and related companies supplied by the Interstate Commerce Commis sion. Industry classification of establishments is based on information available to the State agencies. Certain categories of establishments are eliminated from the master list before the sample is selected, either because a separate survey of the given category is being made or because the number of scientific and technical personnel employed are believed to be negligi ble. The categories of organizations omitted are those classified according to the Standard In dustrial Classification system in the following major industry groups: 01 and 02— farms; 80 — medical and other health services (except 807, medical and dental laboratories, which is included); 82— educational services; 84— mu seums, art galleries, and botanical and zoo logical gardens; 86— nonprofit membership organizations; 88— private households; 89— miscellaneous services (except 8911, engineer ing and architectural service, which is in cluded); 91 through 94— government; and 99 — nonclassifiable establishments. Establishments below a specified minimum size, determined separately for each industry group, are also excluded from the listing prior to sampling. Because of the large number of establishments in the small size groups, mini mum size cutoffs are essential to the efficiency of the survey. Since excluded establishments employ very few scientists, engineers, or tech nicians, survey results are affected little by these omissions. Sample numbers are allocated among the various industry-size strata according to the principle of optimum allocation; expected re sponse rates by industry and by size are taken into account to obtain maximum reliability within available resources. The overall sample size is determined so that the variance (two relative standard errors) for the estimate of the total number of scientists and engineers for all industries combined is about 5 percent. In every covered industry, all establishments with 1,000 employees or more are included in the sample. In other industry-size cells, the sam pling ratios range from 1 in 1 to 1 in 100. In general, the larger the establishment and the greater the number of technical personnel used by the industry, the higher is the sampling ratio. All selections are made randomly within the designated strata. EMPLOYMENT OF SCIENTIFIC AND TECHNICAL PERSONNEL Since scientific and engineering employment is concentrated to a significant degree in re search and development laboratories, not sep arately identifiable in the UI universe, the probability sample was supplemented. The essential rule for unbiased supplementation is that the supplementary units must be drawn independently of the probability sample; that is, the chance that a unit is drawn in the prob ability sample must be independent of the chance that the unit is selected as a supplement. The initial supplementation was drawn from a list of industrial research laboratories compiled by the National Academy of Science— National Research Council and from a list of small busi ness concerns interested in performing research and development compiled by the Small Busi ness Administration. Beginning in 1966, when a new sample was drawn, supplementation was achieved by retaining in the mailing list all establishments which reported employment of 20 or more scientists and engineers in surveys based on the previous sample. Establishments selected as supplements are tabulated as a sep arate cell within their industry and size class with a weight of 1, regardless of whether they are also members of the probability sample. (See Estimating Procedures.) The mailing list for State government sur veys is not a sample, but includes all agencies of State governments which could conceivably employ personnel in any of the designated oc cupations. The agencies are identified from in formation in directories and other documents furnished by the States. Estimating Procedures For the survey of scientific and technical personnel in industry, estimates are obtained for probability cells as the ratio of primary item employment to total employment (of the reporting units in the cell), multiplied by a total employment figure in that industry and size class that is adjusted for any supplemental units in that industry and size class to prevent duplicate estimation for supplemental units.1 Estimates for supplemental cells are obtained by summing the primary item employment for 45 the supplemental reports plus an estimate for nonrespondent supplementary units. Estimates for the survey of scientific and technical personnel employed by State govern ment agencies are obtained by summing the reported data. The response rate in this survey is extraordinarily high— 96 to 98 percent— and examination of the nonrespondents shows that the number of scientists, engineers, and other personnel employed by them is negligible. Analysis, Interpretation, and Presentation A report on the findings of each survey is published, usually within 2 years of the refer ence date of the survey. Each report consists of an analytical interpretation of the findings, and is supported by a statistical appendix con taining in tabular form all of the data that can be meaningfully derived from the survey. Uses and Limitations Data from these surveys form the essential statistical base (1) for evaluating the adequacy of scientific and technical manpower resources of the United States in light of current or projected demands and (2) for determining the rate of growth of these resources. They have been used to evaluate the impact of new or en larged Federal programs calling for substantial scientific and technical manpower. These data also provide the bases for projections of future manpower requirements in science and engi neering. For example, data are furnished for occupational guidance counselors and others who provide young people with information on which to base a choice of career. These estimates must be interpreted as ap proximations. All surveys are subject to possi1 Symbolically, P '= - ? £ i . v e ( M, where M is the cell universe total employment, v p ( is the sum of the primary item employment of the cell respondents, t is the sum of total employment of the cell repsondents, and P' is the estimate. M is adjusted to prevent duplicate estimation for supplemental sample re porters. 46 BLS HANDBOOK OF METHODS ble response and processing errors, although these are reduced insofar as possible, through checking procedures and through correspond ence with reporters whose data are internally inconsistent or appear to involve misinterpreta tions of definitions or other instructions. In addition, estimates derived from sample sur veys are limited by sampling error. Technical References Number 1. U.S. Department of Labor, Bureau of Labor Statistics. Scientific and Technical Personnel In Industry, 1961-66 (Bulletin 1609, 1968). 2. -------------- Scientific and Technical Personnel in Industry, 1967 (Bulletin 1674, 1970). 3 . ------------- Employment of Scientific, Professional, and Technical Personnel In State Govern ments, January 196U (Bulletin 1557, 1967). 4 . ------------- “ Scientific and Professional Employment By State Governments” (Reprinted from the August 1969 issue of the Monthly Labor Rview.) 5. -------------- Employment of Scientific and Technical Personnel in State Government Agencies (BLS Bulletin 1412, 1964). 6. National Science Foundation, Employment of Scientific and Technical Personnel inState Gov ernment Agencies, Report on a 1959 Survey. N SF 61-17 (1961). 7. -------------- Reports prepared for the National Science Foundation by the U.S. Department of Labor, Bureau of Labor Statistics: National Science Foundation, Scientific and Technical Personnel in American Industry, Report on a 1959 Survey. N SF 60-62 (1960). 8. _________ National Science Foundation, Scientific and Technical Personnel in Industry, 1960. N SF 61-75 (1961). 9. _________ National Science Foundation, Scientific and Technical Personnel in Industry, 1961. N SF 63-32 (1963). — M ic h a e l F . Cr o w l e y Chapter 5. Occupational Outlook Background The occupational outlook program originally stemmed from a report of the Advisory Com mittee on Education appointed by President Roosevelt. This Advisory Committee recom mended, in 1938, that an occupational outlook service be set up in the Bureau of Labor Sta tistics to make studies and provide information for use of individuals in choosing a career, and for the use of those responsible for the plan ning of education and training programs. In 1941, the Occupational Outlook Service was organized under a specific authorization by the Congress. Although the first, preliminary stud ies were begun in 1941, it was not until after World War II that the occupational outlook staff was able to devote its efforts to the prep aration of occupational reports for use in guid ance. In mid-1946, a manual of occupational outlook information was prepared for use in the Veterans Administration (VA) counseling and rehabilitation program. In response to a resolution by the National Vocational Guidance Association, calling upon the Congress to authorize this type of informa tion for sale, and to requests by other private individuals and groups, the first edition of the Occupational Outlook Handbook was published in 1949. The favorable public response to the Handbook was a major factor in the Bureau’s decision to issue, with the backing of the VA, a revised and enlarged edition, which was re leased in 1951. Following the conclusion of the Korean hostil ities, there was a sharp increase in public recognition of the key role of vocational guid ance in staffing essential occupations and effectively utilizing the Nation’s manpower resources. This resulted in the Congress in 1955 providng for the mainteance of the Occupa tional Outlook Handbook and its related publi cations on a regular, continuing, up-to-date basis. In 1957, the third edition of the Occupa tional Outlook Handbook was published; also in that year, the Occupational Outlook Quar terly was originated as a companion piece to the Handbook. The 1957 Handbook was fol lowed in due course by the 1959, 1961, 1963-64, 1966-67, 1968-69, and 1970-71 editions of the Handbook. Description o f Program Under the occupational outlook program, the Bureau of Labor Statistics conducts research in, and provides information on, future occupa tional and industry manpower requirements and resources. It provides vocational guidance information on expected employment oppor tunities for the use of counselors, educators, and others helping young people in choosing a field of work. It also provides manpower information for local and national training authorities and policymakers for use in de veloping programs of education and training. The results of the research are published in the Occupational Outlook Handbook, the Occupa tional Outlook Quarterly, and special bulletins, reports, and pamphlets. In its 21/2 decades of industry and occupa tional research, the occupational outlook pro gram has systematically accumulated and analyzed considerable manpower information on such topics as employment trends for major industries of the economy and for most major occupations; employment effects of a great many long-term programs of government agen cies, including those for defense, highways, scientific research, space technology, medical care, and education; and changes in industry and occupational requirements. Toward providing an overall framework of future manpower requirements for the economy as a whole, projections are developed for the broad industry and occupational groups, and have been published regularly.1 Every other 1 “ America’s Industrial and Occupational Manpower Re quirements, 1964-75,” prepared for the National Commission on Technology, Automation, and Economic Progress, pub lished in Technology and the American Economy, Appendix Vol. I, February 1966, and The U.S. Economy in 1980, (BLS Bulletin 1673, 1970). 47 48 BLS HANDBOOK OF METHODS year, hundreds of different, detailed occupa tional and industry statements are published in the Occupational Outlook Handbook. In most of these occupational outlook statements, informa tion is provided on: nature of work; places of employment, education and training require ments; employment outlook for about 10 years ahead, including, in most cases, estimates of annual requirements for growth and replace ment needs; and earnings and working condi tions. In presenting outlook statements for industries, information is included on nature and location of each industry and other indus try characteristics, as well as information on the industry’s major occupations. In presenting the employment outlook for an occupation, information is provided not only on the demand for workers but also on the poten tial supply of workers from many sources— schools and other training institutions, trans fers from other occupations, and reentries to the labor force. The balance between supply and demand, in those occupations for which an assessment is possible, gives some indication of the nature of job competition in a specific field facing young people in the years ahead. In addition to the overall and detailed indus try and occupational projections developed for the Handbook and described earlier, special manpower studies are prepared, as part of the occupational outlook program, that provide in formation, narrower in scope and greater in depth, on the changing industrial structure and 2 Employment and Changing Occupational Patterns in the Railroad Industry, 1947-60 BLS Bulletin 1344 (1963); Em ploy ment Requirements and Changing Occupational Structure in Civil Aviation BLS Bulletin 1367 (1964); and Employment Outlook and Changing Occupational Structure in Electronics Manufacturing BLS Bulletin 1363 (1963). 3 Maxine G. Stewart, “A New Look at Manpower Needs in Teaching,” M onthly Labor Review, June 1964, pp. 639-644; Technician Manpower: Requirements, Resources, and Train ing Needs BLS Bulletin 1512 (1966); Joe L. Russell, ‘‘Chang ing Patterns in Employment of Nonwhite Workers,” Monthly Labor Review, May 1966, pp. 503-509; "Job Hopes High for Record College Class,” U.S. Department of Labor, Press Release, June 5, 1966 (USDL 7246), and College Educated Workers, 1968-80, BLS Bulletin 1676 (1970). 4 Allan F. Salt, “Estimated Need for Skilled Workers, 1965-75.” M onthly Labor Review, April 1966, pp. 365-371; Joseph F. Fulton, “Employment Impact of Changing Defense Programs,” Monthly Labor Review, May 1964, pp. 508-516; Bernard Yabroff, “ Trends and Outlook for Government Em ployment,” Monthly Labor Review, March 1965, pp. 285291, and Max Carey, “ The Crafts— Five Million Opportunities,” Occupational Outlook Quarterly, Spring 1971, pp. 2-11. occupational composition of American indus tries, such as railroads, civil aviation, and elec tronics.2 Other special studies provide more technical information and quantitative projec tions of manpower requirements and resources in specific occupations, industries, or for speci fic groups of workers— such as teachers, tech nicians and nonwhite workers— which include consideration of the current and future de mand-supply relationships and their implica tions.3 Still others discuss only manpower requirements trends and projections, especially in those occupational groups where the supply of workers is difficult to estimate, such as skilled workers and workers in defense-related employment.4 Sources o f Data The projections and other manpower infor mation developed in the occupational outlook program utilize a wide variety of data sources, which vary mainly with the particular occupa tion or industry under examination. The fol lowing sections indicate some of the major sources of statistical and other information utilized in the program. The basic statistics on current and past em ployment in occupations and industries have been based mainly on Bureau of Labor Statis tics household data from the Monthly Report on the Labor Force (MRLF) and establishment data from Employment and Earnings. (A single publication starting with the February 1966 issue). Use is made also of the scientific and technical personnel surveys conducted by the Bureau, which contain data on scientists, engineers, and technicians. The decennial Cen sus of Population is utilized for data on most occupations not covered by the limited detail published in the MRLF and Censuses of Busi ness and Manufacturing are used to fill in in dustry detail. Information from the Civil Service Commission is used for data on Federal Government workers. These basic sources of occupational and industry employment statis tics are augmented by data from Federal reg ulatory agencies, such as the Federal Aviation Agency and Interstate Commerce Commission, 49 OCCUPATIONAL OUTLOOK which collect industry and occupational sta tistics. In some cases, employment data are obtained from unions, industries, trade associa tions, and professional societies. In most cases, however, these general statistics serve only as a starting point for the development of the specific, current estimates needed for a par ticular report. In developing analyses of past and projected changes in employment requirements— which will be described later— the outlook program utilizes statistics of output, hours of work, and output per man-hour. The major sources of the statistics used are Bureau of Labor Statistics studies of productivity and technological de velopment, the Federal Reserve Board produc tion indexes, and the U.S. Department of Commerce output data from the Annual Survey of Manufactures and the Census of Manu factures. Industry associations and unions also often provide similar types of data. Estimates of the past and probable future supply of workers utilize entirely different sources of information. U.S. Office of Education data on enrollments and degrees in high school, post-secondary schools, and colleges and uni versities form a major component of the supply estimates. Bureau of Apprenticeship and Train ing statistics on apprenticeship, as well as information on company training programs, provided through company reports and per sonal interviews, provide other inputs into the supply estimates. Special studies of various aspects of the sup ply of workers provide essential information for the development of the estimates and projections of supply. A few examples of these types of statistical source materials are occu pational mobility studies (Bureau of Labor Statistics); Tables of Working Life (Bureau of Labor Statistics); followup studies of col lege graduates (National Science Foundation); and many other specific types of studies, often of a one-time nature. Earnings information, which appears in many of the outlook publica tions, is drawn primarily from wage and earn ings surveys conducted by the Bureau of Labor Statistics, supplemented with additional infor mation on many occupations from Federal reg ulatory agencies. Studies of union wage scales are also used. Information is also drawn from reports by the National Science Foundation, professional societies and other groups. Filling in gaps in the various types of sta tistics used is information obtained from (1) personal interviews with employers or others closely associated with an industry or occupa tion; (2) reports and interviews with profes sional or trade associations and licensing agencies; (3) union publications and officials; and (4) periodicals, trade journals, annual re ports, and so on. Methods o f Analysis The projections of requirements and re sources developed for the occupational outlook program require varying methods of analysis, usually because of differences in the factors affecting a particular occupation or industry, but also because of differences in the amount of data available for analysis. The broad pattern of research, however, is generally the same in all of the detailed, comprehensive occupational and industry studies. The starting point in most studies is an analysis of the factors affecting the demand for workers in the occupation, and an assessment of how these factors may operate in the future. Occupational employment is affected by a host of factors. Technological change is the most often discussed factor affecting occupational employment, but occupational changes are also influenced by other factors, such as growth in population and its changing age distribution, as in the case of teachers. Government policy— relating, for example, to the magnitude of the defense and space programs and to expendi tures for research and development— also plays a major role. Occupational employment is also influenced by institutional factors, such as union-management relationships and practices, as in the case of railroad workers, or by the relative supply of workers in other occupa tions, as for example, the substitution effect resulting from shortages of engineers and their replacement by technicians. Also influencing occupational employment are changes in the 50 BLS HANDBOOK OF METHODS total demand for the employing firm’s product; changes in the level of income and distribution of income among consumers, industry and gov ernment; and changing patterns of consump tion. It is apparent, in view of this multitude of factors, that no one technique can be used successfully to project manpower requirements in all occupation and industries. The growth and decline of each occupation is affected by its own complex of factors. The number of teachers required, for example, is affected by the number of pupils (related to birth rates and trends in the proportion of children at each age who attend school) and by trends in the ratio of teachers to pupils, which depend upon education practices and available financing. Projections of requirements for scientists, engineers, and technicians require considera tion of different factors. These must take into account such factors as the growing utilization of technical personnel, the increasing tech nological complexity of industrial products and processes, changes in levels of expenditures for defense, and growing research and develop ment activities. Requirements for automobile mechanics are related to the number of new automobiles and accessories and the age of automobiles; for radio and TV repairmen, to the number of radios and TV’s sold, and their age and complexity; for policemen, to popula tion and urbanization; for truckdrivers, to im proved equipment and highways, and for competing methods of transportation. For many occupations, the significant factors influencing employment are the prospective levels of demand for the products of the various industries in which the occupation is found, and the effect of these changes in demand on employment in the industries. Among the gen eral factors which must be considered in an analysis for an industry study are expected changes in the total domestic production of the industry’s product or service, competition with other products or services, expected tech nological changes, output per man-hour, and changes in hours of work. More specifically, in projecting the activity or production level of an individual industry, it is necessary first to establish the nature of the demand for an in dustry’s products or services and the relation ship of this industry to the growth of the whole economy. Obviously, an industry producing products directly for consumers will have a dif ferent type of demand function than an indus try which is making raw materials to be used as a component for further manufacturing. In projecting the production of steel, for ex ample, consideration must be given to the expected increase in population and the trend in steel output per capita. The total require ments for steel depend on the requirements projected for each of the principal steel-using industries, such as the automobile, construc tion, electrical appliances, machinery, and con tainers industries; competition to steel from other materials, such as aluminum and plastics; and the import-export balance for steel. In effect, it is necessary to project the output of both domestic and foreign users of steel in order to estimate total steel requirements. Future industry production or activity levels can then be translated into overall manpower requirements by estimating changes in man hours per unit of output for each industry, and by making assumptions as to changes in hours of work. Because of the tremendous amount of re sources necessary to make an extensive study of each industry in the economy, a more global type of analysis has been used to fill the gaps and to provide an overall framework for the occupational and industry projections. In mak ing the analysis for the products of each indus try, the usual starting point is the total demand of the economy for goods and services; this can be apportioned among the requirements for each major product or service (classified by industry). The estimates of production can then be translated into requirements for work ers in total and by occupation. The general approach in the development of this industry framework is to begin with the population and labor force projections developed by the Bureau of the Census and the Bureau of Labor Statistics. Assumptions are made as to the size of the Armed Forces, the level of un employment, annual hours of work, and output per man-hour. Multiple correlations are made which take into account past employment OCCUPATIONAL OUTLOOK trends and relationships, and variables such as unemployment, size of the Armed Forces, gross national product, and population. By this tech nique, preliminary projections of manpower requirements are developed for each industry for which adequate data are available. The results of the multiple regression analy sis are then used as the basis for further judg ment decisions as to the level of manpower requirements in the projected period, par ticularly for those industries for which detailed industry studies have been made. The consider able amount of information on individual in dustries developed in the occupational outlook program over many years and through discus sion with representatives of industry and unions provides essential background in mak ing these judgments. Analysis of trends and projections for the economy as a whole, or for indivdiual industries made by other groups, such as the National Planning Association, Stanford Research Institute, State and local governments, and universities, also contribute to these judgments. The adjusted overall in dustry projection framework is then utilized as a basis for occupational requirements. These rough occupational projections are then an alyzed and adjusted on the basis of the individ ual occupational studies previously described, both the detail and the control totals. (See description of Industry-Occupational Matrix, chapter 7.) In general, it may be said that the projections are based heavily on judgment as to the effect of the demand factors on specific occupations. Projections of changes in manpower require ments by occupation and industry provide only one part of the information on the total number of job openings which will need to be filled in the years ahead. In most occupations, more workers are needed yearly to fill positions left vacant by those who leave the occupation to enter other occupations or because of retire ment or death, than are needed to staff new positions created by growth of the field. In esti mating the total number of openings likely to arise in an occupation, the occupational outlook program analyzes studies of occupational mo bility among selected groups of workers, and tables of working life. 51 These tables of working life, which are sim ilar to the actuarial tables of life expectancy used by insurance companies, provide a basis for assessing future rates of replacement re sulting from deaths and retirements, which are in turn affected by differences in sex and aver age age of the workers in particular occupa tions. Where men comprise the great majority of workers, estimated replacement rates for death and retirement usually average between 1 and 3 percent a year. In occupations in which women predominate, the rate is usually much higher, and allowance must be made for the large numbers of women who leave paid em ployment to get married and assume family responsibilities but who return to paid employ ment after marriage or raising a family. The replacement rate among elementary school teachers, for example, is estimated at 4.8 per cent a year; many of these teachers return to employment at a later date. In appraising the overall employment oppor tunities in an occupation, estimates are also made of the future supply of personnel, at least in those fields in which the supply is identifi able. Statistics on high school and college en rollments and graduations are the chief sources of information on the potential supply of per sonnel in the professions and in occupations requiring extensive formal education. Data on numbers of apprentices and graduates of voca tional and technical training programs provide some limited information on new entrants into skilled trades. However, in many occupations most new entrants are trained informally, through on-the-job training or company train ing programs. It is not enough to know, of course, how many persons are being formally trained for an occupation, since not all those completing formal training or education in a particular field enter that field upon completion of their courses. As a result, special surveys are utilized to provide additional information on the actual net supply of workers from a training program or a field of study. These include studies of em ployment plans of college seniors, job place ments of college graduates, and jobs entered after completion of MDTA and other types of training. Limited data on transfers out of an 52 BLS HANDBOOK OF METHODS occupation or re-entries into an occupation are also utilized, although, in general, data on oc cupational mobility are available for only a few occupations. The estimates of the future demand in an occupation is then related to estimates of the future supply to develop the employment out look in that field and to provide information to policy makers, educators, and others on the implications of these relationships. Presentation The Occupational Outlook Handbook is the major publication of the occupational outlook program. Oriented toward vocational guidance, the Handbook is a basic reference source, pub lished every other year, which includes com prehensive and non-technical job information on approximately 800 occupations and 30 major industries, covering the entire spectrum of white-collar, blue-collar, and service occupa tions. An occupational outlook report series provides reprints of individual statements from the Handbook. The Occupational Outlook Quarterly provides a continuous flow of current occupational and job information between editions of the Hand book, together with the most recent information available on earnings, training requirements, and other related topics. In addition to these two publications, developed mainly for use in vocational guidance, the occupational outlook program conducts technical and detailed stud ies on specific occupations and industries in order to furnish information to manpower ex perts, personnel departments, and others in terested in the more technical aspects of the Nation’s future manpower needs. — N eal H. R o sen th al Chapter 6. Projections of the Labor Force Background and Uses Projections of the future size of the labor force are needed for a variety of planning pur poses. They provide a basis for establishing the amount of employment growth the economy must generate to maintain high levels of em ployment. They serve as the basis for one ap proach in setting goals for a general economic growth rate consistent with full utilization of human resources. Projections help to gain in sight into the characteristics and numbers of workers who will be available for industry, and to see what this implies for education, training, and personnel policies. In addition, labor force projections, together with population projec tions, are used to estimate demand for products, develop marketing plans, and evaluate expan sion programs. The U.S. Department of Labor is particularly concerned with the relationship between the expected labor supply and the need for various skills and training created by our changing technology. Method Projections of the labor force as a whole and of the separate age-sex groups are made for quinquennial dates usually for about 15 years ahead. The schedule for preparing the projec tions has been irregular owing, in part, to the timing of new projections of the population of working age. Labor force projections for 1975, 1980, and 1985 were published in May 1970. Because social and economic factors affect the supply of labor, certain assumptions need to be made about conditions surrounding any set of labor force projections. Generally, projec tions have been made on the basic assumptions that past trends in labor force participation would continue into the future, and that the economy would continue to expand and main tain high levels of employment opportunity consistent with an unemployment rate of about 4 percent. Another usual assumption is that there would be no major war or significant change in the size of the Armed Forces which might substantially alter the previous work patterns of the population. It also presupposes that the trend toward increased school enroll ment beyond the high school level, which has a direct bearing on labor force activity o f young persons, would continue, supported by adequate school facilities, staff, and aid to students. The general approach used in preparing the Bureau’s labor force projections is to project the proportion of the population in each agesex group or subgroup that is expected to be in the labor force, i.e., the labor force partici pation rate at the specified future date, and to apply these rates to the expected population in each group. In making projections for a given age-sex group or its subgroup, the standard procedure is to fit a line or curve to a series of points representing the labor force participation rates for that group for the years since 1947, and to extrapolate the line or curve into the period covered by the projection. The procedure is modified, as appropriate, to discount the tem porary effect of factors judged to be operative for only short periods. The population projections used in project ing the labor force are prepared by the Bureau of the Census on the basis of analyzing past trends in birth rates, death rates, and net im migration and projecting these trends. Since the birth rates pose the most uncertainty in projecting the population, the Bureau of the Census prepares several series of population projections on the basis of varying assumptions with respect to birth rates. The uncertainty of projecting birth rates does not directly affect the level of the labor force projections 15 years ahead, since everyone of working age (16 years and over) at that future date has already been born when the projections are made. However, the birth rates do have a bearing on projec tions of the labor force participation rates of younger married women, because mothers of young children are less likely to work. Because of this indirect effect, it was necessary to select the one series of population projections which 68 54 BLS HANDBOOK OF METHODS seemed most reasonable on the basis of an in dependent evaluation of past trends in birth rates. For recent labor force projections, series “ C” of the population projections published in the Census Bureau’s Current Population Re ports, P-25, No. 381 was chosen. The overall size of the labor force is built up by age and sex, not only because the composi tion is needed for many of the purposes noted earlier, but also because the degree of labor force participation varies among the different age-sex groups, and the historical trends in these rates also vary. Some of the factors which help to explain the behavior of the labor force participation rates and which affect particular groups include school attendance, marital status, birth rates, and the availability of social security benefits and the expansion of private pension plans. The method of projecting the labor force participation rates for the various age-sex groups takes into account the influence of a number of these specific demographic and social factors. For example, projections of the proportion of persons enrolled in schools in the various young ages are used to subdivide the future population of young persons into those who are expected to be in school and those not in school. The population of married women in ages 20 to 44, by age, is grouped by those who are expected to have children of preschool age and those with no children under 5, on the basis of projected trends in fertility and child spac ing. Similarly, projected marital status distri butions of older adult women are used to provide the future numbers in each marital category within each age. For each of these subgroups, the projected labor force participation rates are applied to their respective future populations and the re sulting labor force summed to provide the total labor force for each age-sex group and for all ages. Sources o f Data The source of the basic historical data on labor force participation rates by age and sex used to project the labor force is the monthly statistics on the labor force. These are pub lished by the Bureau of Labor Statistics and are based on the Current Population Survey of the Bureau of the Census. Historical data on labor force activity by various categories within several of the age-sex groups are ob tained from the recurring supplementary labor force surveys also based on the Current Popu lation Survey. These include information from the October surveys of the employment of school-age youth and the March surveys of the marital and family characateristics of workers. The population projections are the latest available projections made by the Bureau of the Census and published in their Current Popula tion Reports, Series P-25. Projections of school enrollment and marital status, by age, are based on published and unpublished data of the Bu reau of the Census. Data used in projecting the proportion of women in each age group who will have children under age 5 years include published and unpublished data on birth rates, by age of mother and order of birth, from the Division of Vital Statistics of the Public Health Service; fertility and marriage data from re ports of the Bureau of the Census, Current Population Reports, Series P-20, and data from the decennial censuses of population. Technical References Number 1. Travis, Sophia C., “ The U.S. Labor Force: Projections to 1985,” Monthly Labor Review, May 1970, pp. 3-12, reprinted as Special Labor Force Report No. 119. 2. Cooper, Sophia, and Denis F. Johnston, “ Labor Force Projections for 1970-80,” Monthly La bor Review, February 1965, pp. 129-140, reprinted as Special Labor Force Report No. 49. 3. Population and Labor Force Projections for the U.S., 1960 to 1975 (BLS Bulletin 1242, 1959). — So p h ia C ooper T r a v is Chapter 7. Industry-Occupational Matrix Background Sources o f Data The Bureau of Labor Statistics has developed a comprehensive set of data on the occupational employment composition of all industry sectors in the economy. Presently, industry-occupa tional matrices are available for 1960, 1967, 1970, 1975, and 1980. These data are set up to form a matrix, or table, of 162 specific occupa tions plus groupings of occupations crossclassified with 116 industries. Thus, the occupa tional pattern of each industry is shown— i.e., the proportion of each occupation to total em ployment in an industry. Looked at another way, the tabulation shows how total employment in an occupation is distributed by industry. Initially, work on the Industry-Occupational Matrix grew out of concern by the Department of Defense for anticipating the economic prob lems that might arise from various defense programs. The first set of tables related to 1950 and were prepared by the Bureau as a part of the interindustry program of the early 1950’s, sponsored by the U.S. Department of the Air Force. That program was terminated in 1953, but the 1950 matrix and its successors continue to provide the basic information for emergency manpower planning, now carried on by the Office of Emergency Preparedness. In recent years, a strong interest has developed in determining manpower needs for other pur poses. The latter have included training new workers, retraining workers displaced by auto mation, and providing information to high school counselors and to students making career decisions. The change has focused increased at tention on the need for estimates of numbers presently employed in specific occupations and the likely future employment requirements by occupation. The Industry-Occupational Matrix provides a systematic approach to developing the desired information. Data for the Industry-Occupational Matrices are brought together from a wide variety of sources. A major source for the development of the 1960 matrix was the Occupation by Indus try report from the 1960 Census of Population. The Current Population Surveys (CPS) are the source for total employment, employment for broad occupational groups, and for a few large, specific occupations.1 Other sources of occupa tional employment data included the Bureau of Labor Statistics annual surveys of occupational wage rates in metropolitan areas and selected industries ;2 regulatory agency statistics on em ployment by occupation in the telephone, rail road, and air transportation industries; U.S. Civil Service Commission statistics on employ ment by occupation in the Federal Government * , statistics on selected professional occupations based on licensing data and membership rec ords of professional societies; and surveys of employers by the Bureau and other agencies to obtain estimates of employment in a limited number of highly important occupations such as scientists, engineers,3 teachers, and police men. Specific estimates from sources other than the Census were incorporated into the cells of the matrix for about 16 million workers, or one-fourth of all those who were employed in 1960. The remaining details in the matrix were derived by forcing 1960 population census esti mates for detailed cells (published in Occupa tion by Industry) into agreement with control totals for occupational groups and industries from sources other than the Census. The occu pational control totals were average annual employment by occupational group taken from the CPS. Most of the industry employment 1See chapter 1. ’ See chapter 14. * See chapter 4. 65 56 BLS HANDBOOK OF METHODS totals were based on BLS estimates of private wage-and-salary workers adjusted to include the self-employed, unpaid family workers, and government workers, and to exclude the sec ondary jobs of dual job holders. Total employ ment in agriculture and private households was based on CPS estimates. The adjustments of the matrix to consistency with CPS estimates of total employment and industry employment estimates, derived as described above, brings the matrix for 1960 into agreement with data used as the basis for the Bureau’s projections of total employment and occupational employ ment by industry. The Bureau’s occupational projections are reflected in and developed in part through matrix techniques. (See section on analysis and uses.) The 1975 matrix 4 was developed by examining a variety of historical statistics on the changing occupational struc ture of industries and data from the 1950 and 1960 censuses and evaluating the factors likely to influence changes in the future such as ex pected new technology, changes in product mix, and the general organization of industries. A similar procedure, together with use of infor mation for the 1960-67 period, was used in preparing the 1980 matrix. The 1960 matrix provided the base for the 1967 and 1970 matrices.5 Where available, oc cupational data from other sources, such as those cited above, were incorporated into the updated matrices as fixed cells. For the re maining cells, first approximations of the oc cupational patterns for 1967 were made by interpolating between the patterns of the 1960 and the 1975 matrices. The resulting patterns (in mining and manufacturing) were then brought into consistency with data on produc tion worker trends available from the Bureau’s Current Employment Statistics program. The patterns were then applied to individual indus try employment controls and summed up to arrive at occupational totals. These occupa tional control totals were then compared to data from the CPS and other sources of infor-* * For a detailed description of the procedures followed in developing the 1975 matrix, see Occupational Employment Patterns, 1960 and 1975 (BLS Bulletin 1599, 1968). •For a more complete discussion of the methods used in developing the 1967 and 1970 matrices, see Occupational Employment Statistics 196^-67, BLS Bulletin 1643. mation. When necessary, certain occupations (except for fixed cells) were then forced on a prorated basis to predetermined occupational control levels. This iterative forcing procedure was repeated until the internal matrix cells were consistent with both the industry and the occupational controls. A similar procedure was followed in preparing the 1970 matrix with the exception that the 1967 matrix provided the basic data file of occupational ratios used for the iterative forcing procedure. Thus, both the 1967 and the 1970 industry-occupational matrices were consistent with (a) national em ployment by industry, (b) broad occupational employment levels from the CPS, (c) trends in production (and nonproduction) worker em ployment by industry, (d) anticipated trends in occupational structure within industries, and (e) reliable estimates of detailed occupational employment available from the CPS and other sources. Analysis and Uses A basic objective of the project is to have available a comprehensive set of data on in dustry-occupational relationships which can be used in projecting manpower requirements by occupation. Although statistics on employment by occupation are relatively thin, particularly between decennial censuses, there is a great deal of information on total employment in de tailed industries. Each industry utilizes a unique combination of occupational skills, to gether with other factors of production, in its efforts to achieve least cost for its output. Oc cupational patterns may be markedly different from one industry to another. For example, employment in the insurance industry is pri marily of white-collar workers such as insur ance agents, office clerical workers, actuaries, and others. In contrast, the work force in restaurants is largely made up of waiters, waitresses, cooks, and owner-managers. Over periods as short as a decade, the occupational structure of many industries is relatively stable. Consequently, if good information is available on the occupational composition of individual industries for a base period, it can 57 INDUSTRY-OCCUPATIONAL MATRIX be used together with the available statistics on changing employment in each industry to develop estimates of current employment by occupation for later periods. Further, if pro jections of output and employment are avail able by industry, the base period occupational ratios applied to the industry employment pro jections will yield initial estimates of employ ment requirements by occupation for future periods. Although the occupational patterns of many industries are relatively stable over periods of 5 to 10 years, it is clear that occupational pat terns change with the advance of technology and changes in the supply of workers in each occupation. Hence, information on how tech nology and labor supply are changing the oc cupational pattern in each industry is used to modify the initial estimates. This improves the estimates of current employment by occupation and of future employment requirements by oc cupation, developed by applying base period industry-occupational ratios to industry em ployment estimates. Changing technology and other factors which affect skill requirements are constantly being studied in order to esti mate the future occupational structure of each matrix industry. The adjusted occupational patterns are then used, together with projec tions of employment by industry, to prepare estimates of future employment requirements to 1980. — R ic h a r d D em psey Prices and Living Conditions Chapter 8. Consumer Expenditures and Income Background Consumer expenditure surveys* are specialized 1 family living studies in which the primary em phasis is on collecting data relating to family expenditures for goods and services used in dayto-day living. Expenditure surveys of the Bureau of Labor Statistics also include information on the amount and sources of family income, on changes in savings or debts, and on major demo graphic and economic characteristics of family members. The Bureau’s studies of family living conditions rank among its oldest data-collecting functions. The purpose of the first nationwide expenditure survey in 1888-91,2 in line with the legislation creating the Bureau, was to study the worker’s consumption habits and living costs as elements of production costs, with special reference to com petition in foreign trade. It emphasized the worker’s role as a producer, rather than as a con sumer. Purposes and coverage changed in suc cessive surveys, and problems caused by rising prices led to the second survey, during the year 1901. The index of prices of food purchased by workingmen, with weights based on the 1901 data, was used generally as a deflator for workers’ in comes and expenditures for all kinds of goods until World War I. The third major survey, spanning 1917-19, provided weights for comput ing a “cost-of-living” index, now known as the Consumer Price Index (C P I). (See chapter 10.) The next major study, for 1934-36, was made primarily to revise these index weights and cov ered only urban wage and clerical workers. However, in the severe economic depression of the 1930’s, interest in consumer surveys expanded from study of the welfare of selected groups to general economic analysis. Thus, almost simul taneously with its 1934^36 investigation, the Bu reau cooperated with four other Federal agencies in a fifth survey, the Study of Consumer Pur chases, in 1935-36, which undertook to show con sumption of all segments of the population, both urban and rural.3 The Bureau’s sixth major sur vey, for 1950, covered all urban consumers. It provided the basis for revising the Consumer Price Index (CPI) and also supplied abundant material for broader types of economic and market analysis. The remainder of this chapter deals with the 1960-61 survey—the latest in the series 4 describing changes in the consumption habits of the American people. Description of Survey The basic orientation of the Bureau’s most recent survey was to obtain detailed information for re vising the CPI. The increasing need for consumer expenditure and income data for other purposes was taken into account in planning the survey of urban families in 1960 and 1961. Then, in coop eration with the U.S. Department of Agriculture (U SD A ), the 1961 coverage was extended to rural areas. Thus, for the first time since 1941, informa tion on spending habits became available for a cross-section of the entire noninstitutional popula tion in urban, rural nonfarm, and rural farm areas of the United States. Concepts, techniques, and publications for the 1960-61 survey were planned to provide as much continuity and comparability as possible with the Bureau’s 1950 and earlier ex penditure surveys. 1 In this chapter, the Initials CBS are used to refer to the program of consumer expenditure surveys. 1 The State of Massachusetts conducted the first expenditure survey In the United States. This Investigation of living con ditions in 1874-75 undertook to measure the welfare of the workingman’s family before and after migration to the United States. Bor a more detailed account of expenditure surveys in this country, see U.S. Bureau of the Census, Historical Statistics of the United States— Colonial Times to 1957, Chapter Q (1960). *The Bureau also cooperated with the U.S. Department of Agriculture in a smaller scale nationwide survey of urban and rural families In 1941-42 to obtain facts on which to base deci sions for the civilian economy during wartime. In addition, the Bureau conducted a Survey of Prices Paid by Consumers in 1944 among a nationwide sample of urban families; as a byproduct of the reports required for the analysis of prices, certain data on family Income, savings, and expenditures were obtained. « In addition to these 7 major surveys, the Bureau has con ducted a number of expenditure surveys In cities selected for specialized studies. 69 60 CONSUMER EXPENDITURES AND INCOME All data were collected through the voluntary cooperation of families.6 The family, or consumer unit (CU ), referred to (1) a group of people usually living together who pooled their income and drew from a common fund for their major items of expense,6 or (2) a person whose income and expenditures were not pooled with others, whether living alone or in a household. However, never-married children living with parents always were considered as members of the parents’ consumer unit. Information was recorded for the family as com posed in the survey year, including part-year members. Family members were not eligible for periods in the survey year that they lived in mili tary camps, posts, or reservations; in institutions; abroad (except on vacation, etc.) ; or were mem bers of another CU.7 A complete account of family income and out lays was compiled for a calendar year. This ac count included information to determine net changes in the family’s assets and liabilities dur ing the year. The estimated value of goods and services received as gifts or otherwise, without direct expenditures by the family, was requested also. To supplement the annual data, families who prepared meals at home provided a detailed 7-day record of expenditures for food and related items purchased frequently. For selected items of clothing, housefumishings, and food, the record of expenditures was supple mented by information on quantities purchased and prices paid. Characteristics of the housing occupied by homeowners and renters and an inven tory of the major items of housefumishings they owned were recorded. To permit more meaningful analysis of the spending habits of American families, limited demographic information was obtained. This in formation included the sex, age, years of school completed, occupation, race, and marital status of each family member. Data Sources and Collection Methods All data were collected by personal interview. The BLS was responsible for collecting data from all residents o f urban places. The BLS and USDA shared this responsibility in the rural areas of Standard Metropolitan Statistical Areas (SMSA’s)8 and the USDA had sole responsibility for interviewing rural households in nonmetro politan areas.9 Field Organization To reduce the size of the staff to be recruited and trained and to utilize this staff over a longer period, it was decided that the urban survey would cover 2 years, 1960 and 1961. For similar reasons, the surveys for each year were conducted in two “ waves.” As field work was completed in the largest SMSA’s, supervisors were reassigned to smaller places. Interviews for the 1960 and 1961 CES were conducted in the spring and summer of 1961 and 1962, respectively. The supervisory field personnel were recruited by the BLS Regional Offices1 and brought to 0 Washington for 6 weeks of intensive training on the purposes of the survey, survey techniques, and schedule content. These supervisors went to an assigned city where they, in turn, recruited inter viewers, whom they trained for about 8 days. Questionnaires The detailed questionnaires used by the BLS agents in interviewing families in the 1960-61 sur vey had been tested in surveys for 1959 in three cities. They incorporated modifications based on this experience. Three forms were used in the nonfarm surveys. Schedule A was a two-page form to determine the family’s eligibility for the survey and, for families who refused or were unable to participate in the survey, it provided a record of minimum data for the analysis of non response. Schedule B, on which the interviewer entered the complete annual record o f the family’s living arrangements, income, spending, and * See discussion of bureauwide policy on voluntary reporting and confidentiality, under Introduction. * This category Includes children temporarily away from home at school or college. TThe tabulations published in the CBS reports listed in the table at the end of this chapter Include only full-year consumer units, i.e., units with at least one member who was eligible over the entire survey year. In addition, the Bureau obtained sched ules from approximately 400 part-year CU’s, for special analytical research. 8 See appendix B for description of SMSA classification system. * See discussion of sample design, p. 56. “ The following description refers to procedures of the BLS, but USDA procedures were similar. BLS HANDBOOK OF METHODS changes in sayings, consisted of 59 pages and formed the basic framework of the survey. Sched ule C provided 15 pages for a detailed report of the family’s purchases of food, beverages, tobacco, personal care, and household supplies in the week preceding the interview.1 The coding system for 1 summarizing and classifying the data was devised while the 1960-61 schedules were being designed, and tabulating codes were printed on the schedules. The precoded data were then readily transferred to punch cards and magnetic tapes for tabulating. Information in schedule B was grouped in 23 sections, placed in a sequence so as to establish and maintain rapport between the interviewer and the respondent. The detailed checklists of items1 * 1 2 in each section were included not only to facilitate recall, but to provide the specific information needed to determine CPI weights. However, all sections were rarely applicable to a single family. For example, if the family were renters, the sec tions relating to homeownership could be omitted. Families were encouraged to refer to records whenever possible. Reported receipts and disbursements were sum marized and reviewed in the field to determine the completeness, consistency, and balance of the fam ily account. Families were reinterviewed when the field supervisor deemed it necessary to clarify 11A modification of schedule B, providing for detailed reporting of farm receipts, disbursements, changes in farm assets, and value of home-produced food, but otherwise paralleling the non farm schedule, was used by the U.S. Department of Agriculture In rural farm areas. Schedule C was not used In interviewing farm families. u For a discussion of global versus detailed questions and “free-listing” versus “ check-listing,” see report by Helen Humes Lamale, Methodology of the Survey of Consumer Expenditures in I9 6 0 , Philadelphia, Pa., University of Pennsylvania, Wharton School of Finance, 1959, pp. 18-19. 1 Prior to editing the schedules for machine tabulation, the 1 schedules were reviewed in the Washington offices of the BLS and USDA primarily to determine that entries conformed to the survey concepts and methodology. This review was concentrated on sections of the schedule which had proved most difficult in previous surveys, and on unusual situations which required specialized instructions. Occasionally, schedules were returned to the field for clarification or additional information, and some schedules considered complete In the field were rejected in Wash ington. If a schedule met the test of the review and editing instructions with respect to internal completeness and comsistency of expenditures with each other and the family’s reported manner of living, the record was used even though there was a substantial lack of balance between the family’s reported total receipts and disbursements. 1 This procedure involves the probability selection of a sample 4 “ pattern” from a set of patterns purposively established so that, taken as a group, they give each primary sampling unit its proper chance of appearing in the final sample. The selection of the city sample is described in “Technical Note— The Revised City Sample for the Consumer Price Index,” Monthly Labor Review, October 1960, pp. 1078-1083. (BLS Reprint 2352.) 61 ambiguous entries or to complete a record.1 On 8 the average, the interviewer spent 7 to 8 hours with a family in a series of visits arranged at the family’s convenience. Sampling Separate stratified samples were selected for urban areas, rural areas in metropolitan counties, and rural areas in nonmetropolitan counties. A three-stage sample design was used within each area to obtain a sample of consumer units repre sentative of all U.S. consumer units as defined for this survey. In developing urban sampling plans, continued representativeness of the sample for measuring national changes in consumer prices was of para mount importance. Consideration of probable re sources led to setting 66 as the maximum number of cities for the CES sample. Tests o f the ef fectiveness of some of the more obvious modes of stratification indicated that no elaborate stratifi cation was justifiable for so small a sample of cities. In general, classifications by geographic region and size of city seemed to be most effective, especially since an important objective in selecting specific cities was to achieve good geographic dis persion. For this purpose, the BLS utilized the “ controlled selection” procedure.1 * The primary 4 sampling unit was the Standard Metropolitan Statistical Area (SMSA) in the metropolitan seg ment of the United States and the individual urban place in nonmetropolitan areas. Sample Design for Three Urbanizations In the first stage of the design for the urban sample, all SMSA’s and nonmetropolitan urban places were classified by population size and region. A sample of 66 places listed in the table was selected to represent all urban places in the 50 States. All of the 12 largest areas in the United States automatically were included. For New York and Chicago, the Standard Consolidated Areas, rather than the constituent SMSA’s, were used as primary sampling units. However, in the collection and analysis of the data, the New YorkNortheastern New Jersey Standard Consolidated Area was divided into two subareas—New York, N.Y. and Northeastern New Jersey. All of these largest areas were surveyed in both years with data 62 CONSUMER EXPENDITURES AND INCOME S u r v e y o p C o n s u m e r E x p e n d i t u r e s , 1960-61 (S u m m a r y o f S a m p l e S iz e a n d A v a i l a b i l i t y o p D a t a f o r U r b a n a n d R u r a l A r e a s , b y G e o g r a p h ic R e g io n , P o p u l a t io n S t r a t u m , a n d SMSA o r O t h e r U r b a n P l a c e ) Number of— Region, population stratum, SMSA, or other urban place Assignment addresses Total urban and rural—United States............................................................ Northeast........................... ....................................................................... North Central............................................................................................. South.................................................................................................. ........ West............................................................................................................ Publications Survey year Usable schedules 1960 1961 (*) (*) (*) (*) (*) (*) (*) (*) (*) (*) BLS report number Supplements 237-93 237-89 237-90 237r91 237-92 1,2,3 1,2,3 1,2,3 1,2,3 1,2,3 «17,283 4,127 5,187 5,100 12,869 113,728 3,228 4,092 4,180 12,228 Rural farm—United States___________________________________________ ................. ....................... Northeast North Central .. .. ... ... . South West.. ........ ...... .... ... _ ... 2,581 '214 980 1,181 '206 1,967 145 742 925 155 (*) (•) (*) (*) (*) Northeast_______________________________________________________ North Central___________________________________________________ South___________________________________________________________ West ... ............ _ ............ . 2,497 ' 453 702 999 343 2,285 '406 628 948 303 (*) (*) (*) (*) (*) 237-88 237-84 237-85 237-86 237-87 1,2,3 1,2,3 1,2,3 1,2,3 1,2,3 »12,205 3,460 «9,476 2,677 (*) (*) (*) (*) 237-38 237-34 1,2,3 1,2,3 375 625 500 375 375 268 448 356 313 323 (*) (*) (*) (*) (*) (*) (*) (*) (*) (*) 237-7 and 57 237-4 and 54 237-13 and 63 237-8 and 68 237-11 and 61 *1 *1 »1 »1 »1 250 250 199 176 (*) (*) 237-18 237-68 •1 »1 160 160 135 151 (*) (*) 237-14 237-64 «1 11 65 65 65 65 65 65 52 47 41 60 56 53 (*) (*) (*) (*) (*) (*) 237-24 237-24 237-24 237-74 237-74 237-74 3,505 2,722 (*) 500 375. 375 375 371 294 290 319 (*) (*) (*) (*) 250 250 250 173 180 189 (*) 160 160 160 125 126 130 (*) (*) 65 65 65 65 65 65 65 65 65 65 49 55 65 61 48 43 61 50 45 58 (*) (*) (*) (*) (*) 2,920 2,307 375 375 313 323 250 250 250 198 178 201 (*) W 160 160 160 160 110 106 112 135 (*) (•) Urban—United States...................................................................................... Northeast.................................................................................................... SMSA, population 1,400,000 and over Boston, Mass............................................................................... New York, N .Y ............................................................................ Northeastern New Jersey............................................................. Philadelphia, Pa........................................................................... Pittsburgh, Pa............................................................................... SMSA, population 250,000 to 1,400,000 Hartford, Onnn . ...................... . . _ ........................ SMSA, population 50,000 to 250,000 Portland, Maine Lancaster, Pa . Nonmetropolitan urban place, population 2,500 to 50,000 Athol, Mass Millville, N.J _____ Sonthhridge, Mass . . ______ _ ............ North Central............................................................................................. SMSA, population 1,400,000 and over Chicago, 111................................................................................... Cleveland, Ohio........................................ ...................... .......... Detroit, Mich........ ................__........... ...................................__ St. Louis, Mo................................................................................ SMSA, population 250,000 to 1,400,000 Tndianapnlis, Tnd. _ SMSA, population 50.000 to 250,000 Cedar Rapids, Tnwa. . ..................... Champaign-Urbane, T U GreenBay, W is ..._______________________________________ Nonmetropolitan urban place, population 2,500 to 50,000 Devils Lake, N.Dak ............ .......... .......... Crnnlrstnn, Minn _ T^iganspnrt, Tnd . Manhattan, Nans . Menasha, Wis_. _ ....................... ...... ........... ....... .................... . . . . . South........................................................................................................... SMSA, population 1,400,000 and over Baltimore, Md.............................................................................. Washington, D.C............ .......................................... - ................ SMSA, population 250,000 to 1,400,000 SMSA, population 50,000 to 250,000 Orlando, Fla_______ _______ _ _________ Raton Rouge La .... Durham, N.C................... .............................. ................. .......... See footnotes at end of table. (2) (2 ) (J ) (2 ) (2 ) (») (2 ) (2 ) (2 ) (2 ) (*) (*) (*) (4 ) (4 ) (4 ) (*) 237-35 (*) (*) (*) (*) 237-5 and 65 237-21 and 71 237-1 and 51 237-15 and 65 »1 *1 *1 *1 (*) (*) 237-10 237-60 237-67 11 •1 »1 (*) 237-17 237-23 237-73 »1 *l »1 (») (*) (*) (*) (*) 237-27 237-27 237-27 237-27 237-27 237-77 237-77 237-77 237-77 237-77 (*) (*) 237-36 (*) (*) (*) C) 237-16 and 66 237-3 and 53 *1 *1 <*) 237-6 237-20 237-56 »1 *1 *1 (*) (*) 237-12 237-19 237-62 237-69 »1 »1 *1 *1 1,2,3 (4 ) (<) (4 ) (4 ) (4 ) (4 ) (4 ) (4 ) (4 ) (4 ) 1,2,3 63 BLS HANDBOOK OF METHODS Su r v e y (S u m m a r y of S a m p l e S iz e of C o n s u m e r E x p e n d it u r e s , 1 9 6 0-6 1 — Continued a n d A v a il a b il it y o f l a t io n S t r a t u m , a n d D a t a f o r U r b a n a n d R u r a l A r e a s , b y G e o g r a p h ic o r O t h e r U r b a n P l a c e ) — Continued R e g io n , P o p u SMSA Urban—United States—Continued South—Continued Nonmetropolitan urban place, population 2,500 to 50,000 Cleveland, Tenn___________________ _____________ Griffin, Ga______________________________________ McAllen, Tex____ ______________________________ Reserve, La_____________________________________ Union, S .C ................................................................... Vicksburg, Miss________________________________ Florence, Ala___________________________________ Gainesville, Tex___________________ ____ _________ Mangum, Okla............................................................ Martinsville, Va________________________________ Okmulgee, Okla................................................. ......... Sebring, Fla............................................................ .... Assignment addresses Publications Survey year Number of— Region, population stratum, SMSA, or other urban place 1960 Usable schedules 1961 BLS report number Supplements 65 65 65 65 65 65 65 65 65 65 65 65 43 61 38 64 50 55 54 56 50 55 48 57 12,320 >1,770 (») (*) 237-37 500 375 388 302 (*) (*) (*) (*) 237-22 and 72 237-2 and 52 »1 »1 SMSA, population 250,000 to 1,400,000 Seattle, Wash.................................. Denver, Colo____ ______________ Honolulu, Hawaii.......................... 250 250 250 209 (*) ......... 204 ............ (*) 215 ............ (*) 237-9 237-59 237-78 »1 »1 SMSA, population 50,000 to 250,000 Bakersfield, Calif........................ 160 120 (*) 237-70 U Nonmetropolitan urban place, population 2,500 to 50,000 Anchorage, Alaska....................................................... . Gallup, N.Mex_________________________________ Klamath Falls, Oreg.................................................... Eureka, Calif................................................................ Orem, Utah.................................................................. 275 65 65 65 65 134 58 44 42 54 (s) 237-29 237-26 237-26 237-76 237-76 W est SMSA, population 1,400,000 and over Los Angeles-Long Beach, Calif... San Francisco-Oakland, Calif___ (*) M H <) * (*) (') 8 (*) 237-25 237-25 237-25 237-25 237-25 237-25 237-75 237-75 237-75 237-75 237-75 237-75 1,2,3 «1 (‘ W (* ) (* ) < « > •Asterisk Indicates year of survey, i Includes Anchorage, Alaska, which was surveyed for 1959. * Published by the U.S. Department of Agriculture. 1 Supplements 2 and 3 not to be published; for a limited time photocopies of tables may be obtained from the Bureau of Labor Statistics at a nominal cost. * No supplements available. * Survey for 1959. N ote ; The Bureau also has published reports for the following special-city surveys: Cincinnati, Ohio, 1959 (Report 237-28); Fairbanks, Alaska, 1959 (Report 237-30); Ketchikan, Alaska, 1960 (Report 237-31); Juneau, Alaska, 1960 (Report 237-32); Las Vegas, Nev., 1962 (Report 237-33); Houston, Tex., 1963 (Report 237-83); Kansas City. Kans.-Mo., 1963 (Report 237-79); Mil waukee, Wis., 1963 (Report 237-80); Minneapolis-St. Paul, Minn., 1963 (Report 237-81); and San Diego, Calif., 1963 (Report 237-82). collected from half the sample of consumer units each year. H alf of the remaining sample of smaller SMSA’s and urban places was surveyed each year, i.e., for 1960 and 1961. At the second stage, a sample of housing unit (living quarter) addresses was obtained in each city and in the entire urban part of each SMSA from the Comprehensive Housing Unit Survey (CHUS) conducted by BLS in the fall of 19601 6 or 1961. The BLS housing unit surveys were based on area block samples designed to represent all noninstitutional living quarters (including non transient accommodations in hotels and rooming houses) in the urban area. In the third stage, a subsample of addresses was selected from the housing unit addresses obtained in the CHUS or Census, arranged by a number of characteristics, e.g., household size. The first stage in selecting the rural sample in metropolitan areas utilized all 34 SMSA’s selected for the urban sample. In the second stage, BLS conducted a Rural Housing Unit Survey (RHUS) which consisted of a listing of housing unit ad dresses in a stratified sample of Census Enumera tion Districts (ED ’s) and a subsample of smaller segments or blocks in the designated ED’s. Each housing unit was visited and classified as farm or nonfarm, and as to whether the family included a farm operator.1 In the third stage, subsamples 6 of rural nonfarm and rural farm housing unit addresses were selected from the RHUS listing by applying a ratio based on census data on rural farm and rural nonfarm households in each stratum. u Because of the shortage of time, the BLS did not conduct a CHUS in 1960 In places with population of 2,500 to 50,000. The 1960 sample for these cities was selected from 1960 Census, of Housing and Population listings of living quarter addresses arranged by Enumeration Districts (ED’s). M These classifications, which were on the basis of Census definitions, made it possible for the BLS to refer addresses of all households meeting the Census definition of farm operator or farm resident to the USDA for inclusion in the rural sample. 64 CONSUMER EXPENDITURES AND INCOME In the first stage of the USDA’s design for the rural sample in nonmetropolitan areas, counties were grouped by State Economic Areas1 into 126 7 strata equal in weighted counts of rural farm and rural nonfarm dwellings, as the same sample of counties was to be used for both farm and nonfarm households. For each stratum, one county was chosen at random with a probability proportional to its weighted count. Counties were selected from 41 States. At the second stage, within each sample county, a selection of rural segments was made separately from rural places (100 to 2,500 inhabitants) and the open country. Addresses of all housing units in these segments were listed and classified as farm and nonfarm. Farm operators also were identified. In the third stage, sub samples of nonfarm and farm housing unit addresses were selected from the survey listings. Sample Size The master sample for the total urban and rural population included 17,283 living quarter ad dresses which were assigned to the interviewers.1 8 Usable schedules were tabulated for 13,728 con sumer units. The distribution ,o f assignment addresses and usable schedules by urbanization, geographic region, and for individual metro politan areas or cities in the urban sample is shown in the table. Estimating Procedures To describe the spending and saving of all fami lies in the United States, data from the CES samples have been combined to obtain regional and U.S. levels.1 This information was sum 9 marized for each level of urbanization and for the entire population by using a system of weights based on the 1960 Census of Population. To obtain the weights, adjustments were made in the Census total of persons in the population on April 1,1960, to correct for definitional differences between the Census and the CES universe. The institutional population and on-post military per sonnel, which were not included in the CES, were deducted from the Census population. Since the CES data apply to the full survey year and family size is measured in year-equivalent persons, while the Census data are a count of persons on April 1, 1960, the Census data were adjusted to take account of births, deaths, and net civilian migration dur ing 1960. For the 50 States, the net effect of the adjustments was to lower the population total from 179,325,671 to 177,391,360. The total adjusted population was distributed among the sampling strata in accordance with the distribution of the unadjusted population. The population represented by each surveyed area was divided by the average family size in the area, as determined from the survey, to obtain the total number of families represented by each area. The estimated number of consumer units in the universe for the United States was 55,306,253. The adjusted 1960 population was used as urban weights for both 1960 and 1961. Weights were computed for 67 urban strata,2 including Anchor 0 age, Alaska, which was surveyed for 1959. (See table.) Rural nonfarm weights were computed for 42 strata—34 SMSA’s and a farm operator and nonoperator stratum for each of the four regions. The rural farm sample was designed to be self weighting within regions. Sample averages for the four regions were combined to U.S. levels by the application of weights consistent with those weights used in the urban and rural nonfarm parts of the CES. In applying the weights to the stratum aver ages, to obtain U.S. and/or regional averages, the blow-up factor for each class (e.g., income group, family-size class, etc.) was the number of consumer units in the universe represented by each sample family in a stratum multiplied by the number of families in the sample for that class. The result ing numbers of consumer units became the multi pliers in calculating stratum aggregates which 1 See appendix B. 7 MThe address following each address In the master sample waa picked as an alternate. The alternate address was substituted If the master address could not be located or the unit was vacant, If no one was at home after at least 2 visits, or If the occupant refused or was unable to give the minimum informa tion required for classifying the family. If the interviewer found more than one consumer unit living at an address, each unit was included In the sample. In hotels and apartments, the address given to the interviewer indicated a specific housing unit or units within the structure. 19 The samples were not designed to provide tabulations by State. *° For a single year such as 1961, the city weights dlifered from the 1960-61 weights, In that cities surveyed in that year carried the entire weight for their respective region city-size stratum In the 1961 tabulations. In combining 1960 and 1961, each year’s sample represented approximately half of the ad justed population. The 1959 data for Anchorage were weighted into the combined 1960-61 tabulation for the West and the United States, but not into the tabulations for 1960 or 1961. BLS HANDBOOK OF METHODS were combined to regional and U.S. levels. The regional and U.S. averages were obtained by divid ing the sum of the stratum aggregates by the num ber of consumer units in the universe for the class. Analysis and Presentation Data were tabulated separately for each city and region listed in the table. All dollar values (in come, expenditures, and changes in savings) were shown as averages per family (i.e., consumer unit). As city tabulations were completed, they were examined for reasonableness and internal consist ency, and were compared with tabulations for other cities in the 1960-61 CES sample and with information from independent sources—princi pally the 1960 censuses of population and housing. Similar analytical comparisons were made for the regional and U.S. tabulations at each level o f urbanization. For the urban sample, trends since 1950 were analyzed for cities which were in the CES sample for both periods. Each report con tained brief analytical and interpretative text, plus definitions and statements on methodology. The basic reports (see column 5, table) pre sented averages for major components of family accounts for consumer units classified by five characteristics: Family income after taxes, family size, age o f family head, occupation of the head, and housing tenure. Supplement 1 presented the same information, classified by four additional characteristics: Education of the head, race, fam ily type, and number of full-time earners. At the regional and U.S. levels, families also were classi fied by whether they lived inside or outside metro politan areas; and inside SMSA’s, by central city and other location. Data for eight family characteristics in the above summaries were cross-classified (two vari ables) with each o f other selected characteristics, as follows: 1. Family size with income, age of head, family type and location and size of place. 2. Age of head with income, occupation of head, and tenure. 3. Occupation of head with income, race, and tenure. 4. Education of head with income, and occupation of head. 5. Race with income, and tenure. 6. Number of full-time earners with income. 7. Tenure with income. 8. Family type with income, and occupation of head. 65 The two-variable tables are published as Supple ment 2 to the basic reports. Supplement 3 presents in detail the components of consumer expenditures, income, and changes in savings, which were summarized in the basic re ports and Supplements 1 and 2. To illustrate, the category “ automobile transportation” is broken down into 10 subgroups of expenditures. These detailed tabulations provide data for consumer units cross-classified by family size and income after taxes, and by family size and location of the family’s residence inside or outside SMSA’s. Uses and Limitations From its inception, the 1960-61 CES was planned to serve a variety of purposes. To this end, questionnaires and tabulation plans were circulated among a number of Federal agencies for comment. The BLS also formed the Con sumer Expenditure Advisory Committee, repre senting academic, private research, and marketing users, and consulted with this Committee on a wide range of decisions affecting the CES. To date, the results have been used for only some of the contemplated purposes. As stated earlier, the primary objective o f the 1960-61 CES was to revise the CPI (chapter 10). The Bureau also uses the data to revise and expand its work in deriving standard quantity budgets for selected types o f families (chapter 9). As time and resources permit, the Bureau proposes to draw upon this fund of consumer information to develop a broad program of family living conditions studies.2 1 The U.S. Department of Commerce relies on these family expenditure studies as the sole source of information for revising its benchmark esti mates for a number of components in the household sector of the national accounts. The Internal Revenue Service used the published 1960-61 data to revise its tables of State sales tax payments, for guidelines to taxpayers in filing their Federal in n The Bureau has Initiated a series of special analytical re ports (BLS Report 238-1, et. seq.) based on the 1960-61 CES. A list of reports In this series Is available upon request. The Bureau also prepares methodological monographs for the expenditure surveys which compare national aggregates of In come, expenditures, and savings derived from the CES with aggregates developed for the national Income and product ac counts by the U.S. Department of Commerce and with data from other sources. CONSUMER EXPENDITURES AND INCOME 66 come tax returns. Currently the Department of Health, Education, and Welfare is studying the data in connection with numerous welfare analyses. A number of universities utilize the Bureau’s basic data in a variety of research projects. The potentialities of the survey results in consumer market analysis are evident from reports issued by the National Industrial Conference Board, the Chamber of Commerce of the United States, and a variety of trade publications and business organizations. Data obtained from a sample survey as complex as the survey o f Consumer Expenditures are sub ject to many types of errors. These include sampling, recording, and processing errors, and errors due to the refusal or inability of some fami lies to give the information requested. All data were reviewed, edited, and screened to minimize processing errors. Chance variations due to sampling can be measured statistically, and the BUS has published preliminary rough esti mates of sampling error in the urban sample, in its report for the urban United States. Preparation of more detailed estimates for both the urban and rural segments will depend upon the availability of resources. Each report contains a section alert ing users to the cautions that must be exercised in using averages based on small samples and pro vides either the actual number or the basis for determining the number of families on which the averages in each table are based. Approximately 78 percent of the national sample o f urban and rural families furnished usable schedules and some of the nonrespondents supplied limited infor mation on family characteristics.2 Among the 2 participating families, inaccurate reporting is a source of error, despite continued research in schedule design and intensive training of the inter viewers. Such inaccuracies result from memory errors, misunderstanding of a question or reluc tance to answer it, and incorrect entries by the interviewer. Although the BLS and USDA have accumulated substantial knowledge about such reporting errors and will continue research in this field, these errors cannot be quantified satisfactorily. ” This Information la used In the monographs on methodology (see footnote 21) to evaluate the nature of the sample losses due to nonresponse. It has been the Bureau’s practice not to attempt to introduce in the basic CES tables adjustments for nonresponse or for the underreporting which is common in vir tually all compilations of income statistics. Such adjustments present many difficulties when data are cross-classified as ex tensively as are the CES tabulations. Technical References Num ber 1. Brady, Dorothy S. and Williams, Faith M ., "Advances in the Techniques of Measuring and Esti mating Consumer Expenditures," Journal of Farm Economics, May 1945, pp. 315-344. A review of changes in the direction of family expenditure surveys as the role of the consumer gained importance in economic theory in the mid-1930's. Discusses problems of concepts, definitions, sampling, and data collection that emerged as emphasis shifted from the analysis of expenditures of selected occupational groups to the interrelationships of expenditures, income, and savings of all types of families throughout the United States. 2. Ferber, Robert, "Research on Household Behavior," The American Economic Review, March 1962, pp. 19-63. A survey of the main empirical research since World War II on the determinants of house hold spending and saving behavior. Includes an extensive bibliography. 3. International Labour Office, Family Living Studies— A Symposium, Studies and Reports, New Series No. 63, Geneva, 1961. A collecton of recent case studies selected to draw attention to different types of family living studies and to their methods and problems. The 15 chapters include surveys from coun tries throughout the world. 4. Lamale, Helen H., "Changes in Concepts of Income Adequacy over the Last Century,” The American Economic Review, May 1958, pp. 291-299. Relates changes in criteria for income adequacy to economic and social developments in three broad periods: 1860 to 1900— the "Subsistence" or "Break-Even" concept; 1900 to 1935— the "Living-Wage” concept; since 1935— the "Social" concept. 67 BLS HANDBOOK OF METHODS Technical References—Continued Number 5. Lamale, Helen H. Study of Consumer Expenditures, Incomes and Savings— Methodology of the Survey of Consumer Expenditures in 1950. Philadelphia, Pa., University of Pennsylvania, Wharton School of Finance and Commerce, 1959. A comprehensive statement of the purposes, procedures, and reliability of results of the 1950 survey, with summaries of earlier surveys and research which influenced the 1950 methodology. Includes facsimiles of questionnaires used in BLS expenditure surveys for 1901, 1917-19, and 1950. 6. Stigler, George J., “ The Early History of Empirical Studies of Consumer Behavior,” The Journal of Political Economy, April 1954, pp. 95-113. Examines the relationship of the empirical work on consumer behavior to the theoretical work on income and demand prior to the twentieth century. 7. United Nations Statistical Office, Handbook of Household Surveys: A Practical Guide for Inquiries on Levels of Living, Provisional Edition. (Studies in Methods, Series F. No. 10.) New York, United Nations, 1964. Presentation of standard concepts, definitions, classifications, and tabulations, for household surveys with the aim that their adoption or adaptation would enhance the value of the surveys for national purposes and facilitate international comparison of results. 8. U.S. Bureau of the Census, Historical Statistics of the United States— Colonial Times to 1957, (1960) Chapter G. The same, Continuation to 1962 and Revisions, 1965. A compilation of tabular data from major Government and private statistical reports on consumer income and expenditures in the United States. Text describing each report is keyed to the tabulations. 9. U.S. Department of Labor, How American Buying Habits Change (1959). A popular-style book dovetailing various studies to yield a picture of changes in the con sumption habits of the American people between 1875 and 1950. Includes a chapter of technical comments on consumption statistics and an extensive bibliography. 10. Williams, Faith M. and Zimmerman, Carle C., Studies of Family Living in the United States and Other Countries: An Analysis of Material and Method, U.S. Department of Agriculture, Mis cellaneous Publication No. 223, December 1935. Analytical annotated bibliography of approximately 1,500 studies of family living made in 52 countries. Includes statements on the history and methodology of consumption studies since their inception. A continuation of this bibliography, covering 1946 and later years, is published in Miss Williams chapter, “ International Comparisons of Patterns of Family Consumption,” in Consumer Behavior— Research on Consumer Reactions (Lincoln H. Clark, ed.) New York, Harper and Brothers, 1958. 11. Zimmerman, Carle C., Consumption and Standards of Living, New York, D. Van Nostrand Co., Inc., 1936. A definitive analysis of family budget studies, with pertinent tabular materials, and dis cussion of psychological, social, and economic concepts and theories of consumption. —K athryn R. M urphy Office of Prices and Living Conditions Chapter 9. Family Budgets Background “ Standards of living” refer to the goals of consumers and workers in their consumption of goods and services, use of leisure time, and conditions of work. Standard budgets, also de scribed as family budgets, measure the total costs or amounts of income required to achieve the levels and manner of living implicit in these goals.1 Cost estimates are developed by trans lating the generalized concept of a living stan dard into a list of commodities and services which can be priced. Thus, standard budgets are normative, or benchmark, estimates of living costs. They do not represent the ways in which family incomes should be spent, or the ways average families actually spend their incomes. The first standard budgets prepared by the Bureau of Labor Statistics were developed for the specific purpose of evaluating living condi tions of cotton-mill workers in Fall River, Mass, and in the South in 1908-09. These budgets described two standards of living— a minimum, including only bare necessities; and a fair stan dard, including some allowance for comfort. An other budget defining a standard of health and decency was developed in 1919. In the late 1930’s, BLS cooperated with the Works Prog ress Administration in pricing two budgets: A maintenance budget, described as above the 1 In 1954, at the request of the United Nations Economic and Social Council, a committee of experts from six different countries (including the United States) recommended that the following distinction be maintained between the terms “level” and “standard” of living: The “ level of living” relates to the actual living conditions of a people. The “standard of living” relates to the aspirations or expectations of a people, that is, the living conditions which they seek to attain or regain, or which they regard as fitting and proper for themselves to enjoy. However, laymen and the general public frequently refer to “the high levels of living” actually achieved by the average American worker as the “high standard” of American living. 2 Published sources for these budgets are described in Tech nical Reference No. 3, listed on p. 73. 3 See Technical Reference No. 18. For description of a com panion budget for an elderly couple, see Technical Reference No. 19. 4 See Technical Reference Nos. 7 and 10. 3 See Technical Reference Nos. 11, 14, 15, and 16. minimum subsistence level but approaching a satisfactory American standard of living; and an emergency budget, derived by cutting the maintenance budget for emergency conditions “ with the least harm to the individuals and the social group.” The information available at the time concerning the requirements for nutrition ally adequate diets and healthful housing was incorporated into the definitions of the living standard in each of these budgets. For other components of family living, the “ require ments” were formulated primarily on the basis of the personal judgment of the budget makers.1 2 In 1946, the Bureau compiled the City Work er’s Family Budget for a “modest but adequate” standard of living. The procedures used stan dards of adequacy that reflected the judgments of scientists and experts where these were avail able; for other components, they depended on statistical analyses of consumer choices.3 The same method, with some refinements, was used in 1959, in the interim revisions of the City Worker’s Family Budget and the Budget for a Retired Couple.4 These procedures were used again, with additional refinements, in the mid1960’s to develop the “ intermediate” budgets (initially described as “ moderate” budgets) for a four-person family and a retired couple. Sub sequently, the costs of the intermediate level budgets were scaled downward and upward by a variety of techniques to produce a “ lower” and a “ higher” budget for each family type.5 Procedures for the intermediate budgets of the 1960’s and the scaling techniques are described in the remainder of this chapter. Description of the Budgets All normative estimates of living costs must be based on specific family situations. The con struction of a family budget, therefore, requires a set of assumptions, i.e., specifications, which must be formulated explicitly by the budget maker at the outset. These relate to the age, 69 70 BLS HANDBOOK OF METHODS size, and type of family; the manner of living appropriate for the specified family composition and the locality in which the family resides; and the position of the living standard in rela tion to the actual scale of consumption. Family composition has a significant effect on spending patterns, manner of living, and family needs. The budgets ior a younger, fourperson family, specifies that the family consists of an employed husband, age 38, who has a wife not employed outside the home, and two chil dren, a girl of 8 and a boy of 13. This family type represents a middle stage in the life cycle, and it has been widely used as the unit for other budgets compiled for earlier periods. The family in the budgets for a retired couple consists of a husband and wife, age 65 or over, who are assumed to be self-supporting, in reasonably good health, and able to take care of themselves. This unit, which has a markedly different pat tern of living and needs than the younger fam ily, has been the subject of special concern in national policy formation over the last three decades. Budget quantities and budget com ponent cost estimates for other family types cannot be derived as fractions or multiples of the quantities or cost estimates for food, shel ter, clothing, transportation, etc. for the fourperson family or the retired couple.6 Both types of families were assumed to live in an urban area. Assumptions also were made concerning the living arrangements and tenure of the families; inventories of housefurnishings, household equipment, and clothing; means of transportation; ownership of life insurance; provisions for medical care; savings positions, etc. In making these assumptions, the budget 8 Extensive analyses of consumption data dating back over more than a century have provided a variety of measures of general welfare; e.g., the relative adequacy of diets, the pro portion of income spent for various categories of goods, or the proportion of income saved. These measures, either singly or in combination, have been used as the basis for determin ing scales of equivalent income for families of different size. One such scale is described in Technical Reference No. 12. The scale is based on the assumption that families spending the same proportion of income on food have attained equal levels of living. Although the scale is useful in estimating equivalent costs of goods and services, or net income re quirements after income taxes and occupational expenses, it cannot be applied to individual items or major components of budget costs. 7 For a discussion of the relativity of living standards, see Technical Reference Nos. 5 and 6. 8 For a description of the Bureau’s surveys of consumer e x penditures, see chapter 8. makers were guided by data on the prevalence of ownership of particular types of assets in the urban metropolitan population, and the avail ability of goods and services provided by gov ernments for collective consumption or provided under collective bargaining agreements between employers and unions. All three budgets provide for the mainte nance of physical health and social well-being, and participation in community activities. Within this broad framework, different levels of adequacy were obtained by varying the as sumptions concerning the manner of living and by providing different quantities and qualities of the necessary goods and services. The content of the budgets is based on the manner of living and consumer choices in the 1960’s. The lower budget differs from the inter mediate and higher budgets in several specifica tions: The family lives in rental housing with out air conditioning, (except for a proportion of retired couples who may own their own homes), relies heavily on public transportation, supplemented, where necessary, by the use of an older car, performs more services for itself, and utilizes free recreation facilities in the com munity. Compared with the intermediate budget, the life style in the higher budget, is marked by more homeownership, high levels of new-car ownership, more household appliances and equipment, and more paid-for services. For most items common to all budgets the quanti ties are greater and the qualities higher in the intermediate than in the lower budget, and in the higher than in the intermediate budget.7 Data Sources Budget quantities and pricing specifications were derived from two sources: (1) Scientific or technical judgments concerning the require ments for physical health and social well-being; and (2) analytical studies of the choices of goods and services made by consumers in suc cessive income groups, as reported in the Bu reau’s surveys of consumer expenditures,8 to determine by statistical procedures the income class whose spending pattern would be used as the “ norm” for a specified budget level. Scientific standards for nutritionally ade 71 FAMILY BUDGETS quate diets for individuals in different sex-age groups have been developed by the Food and Nutrition Board of the National Research Coun cil, and translated by the U.S. Department of Agriculture into food plans at different cost levels. These food plans were used as the basis for the food-at-home component of the budgets. Housing standards established by the Ameri can Public Health Association and the U.S. Public Housing Administration were adopted for the budgets. These standards relate to sleep ing space requirements, essential household equipment (including plumbing facilities), ade quate utilities and heat, structural condition, and neighborhood location of the dwelling units. Fuel requirements were derived by analyzing actual fuel purchases of families in the specified types of dwellings in relation to degree-days to provide an adjustment for differences in cli mate. Estimates of electricity and utility serv ices required for the appliances specified for the budgets were furnished by utility companies and associations. The widespread use of insurance to cover the cost of major illness was accepted as a basis for a standard for medical care, and a family mem bership in a group health insurance plan (Medi care for the retired couple) was specified. Quantities of medical care services not covered by insurance were derived from data on utiliza tion rates provided by the 1963-64 U.S. Na tional Health Survey and the 1960-61 Consumer Expenditures Survey. Major medical provisions were specified for the higher budget. No generally accepted scientific standards are available for other components of the budg ets (clothing, housefumishings, transportation, personal care, household operation, reading, recreation, tobacco, education, gifts and contri butions, and miscellaneous expenses). There fore, for most of these components a technique was developed which relied on the choices of con sumers as the basis for a standard. Purchases were examined at successive income levels to determine the income level at which the rate of 8 For a description of the Consumer Price Index, see chapter 10. 1 Technical Reference Nos. 13, 15, and 16 provide additional 0 details on the methods, as well as lists of goods and services priced. Dollar cost estimates are provided in these reports and also in Technical Reference No. 9. increase in quantities purchased began to de cline in relation to the rate of change in income, i.e., the point of maximum elasticity. The aver age numbers and kinds of items purchased at this income level became the quantities and qualities specified for the intermediate level budget. In general, income classes below and above the classes used for the intermediate level were specified as the source of quantities for the lower and higher budgets, respectively. For the transportation component, quantities for the intermediate and higher budgets were based on the average consumption pattern of families of each budget type. For the lower budget, average patterns of renter families were used. Except for the higher budget where costs include a car for all families, automobile ownership was specified in inverse relationship to the availability of mass public transporta tion. Mileage allowances were adjusted by the use of automobiles for work. In determining budget costs, levels of prices paid for items are as important as the numbers of items bought. Items in the intermediate budget were priced in the types of stores and professional and service establishments custom arily patronized by urban families. Prices, pric ing procedures, reporting stores and service establishments, and price calculation methods were those used by the BLS for the Consumer Price Index,8 except that additional quotations 9 were obtained in some cases to calculate aver ages and different qualities were priced in other cases to represent the intermediate budget level. For some items in the lower and higher budgets, special prices were collected directly from stores and establishments. In the main, however, prices for those two levels were esti mates in a variety of ways.1 0 Analysis and Presentation In the methods described, a family budget is the end result of a multitude of decisions by the budget maker, based on standards formu lated by scientists or experts or on analyses of data on consumption patterns from a variety of sources. The budgets are not simply the prod 72 BLS HANDBOOK OF METHODS ucts of a survey of ways families at particular income levels actually spend their money. The judgment of the budget maker is involved in selecting among the family types and manners and levels of living to be represented; in deter mining the most appropriate sources of data to be used in deriving budget quantities; and in interpreting actual family consumption in terms of norms or benchmarks. The appropriateness of the operating assumptions can be evaluated only by the budget users in relation to the pur poses to be served.1 1 Budget estimates may be analyzed in four ways: (1) Costs are compared with income. However, costs for a specific family type should be compared only with average incomes, or in come distributions, for families of the same type. This kind of analysis has been restricted, therefore, by the availability of cost estimates for only two family types. However, family equivalence scales may be used to develop esti mates for comparable benchmark levels for families of other types. (2) Budget costs in one place are compared with costs in another, i.e., the budgets provide a basis for calculating an index of locality differences in living costs. The Bureau has provided this type of analysis in conjunction with its published reports. (3) Costs are compared over time to measure changes in living standards. The sporadic char acter of the Bureau’s family budget research program imposes serious limitations on this type of analysis. Also the judgment factor in developing budgets introduces a serious bias for evaluating changes in the levels and living stan dards of families from decade to decade. (4) Finally, budget estimates of different levels are compared to provide a measure of the aggre gate addition to income required to raise con sumption to particular levels. The development of budgets for three different levels facilitates this type of analysis. Uses and Limitations Family budgets are used in economic research to appraise the economic condition of the popu lation and to evaluate the need for, and the effect of, specific laws and programs. For ex ample, normative living costs are used, to mea sure the extent to which social security or un employment insurance benefits provide income sufficient to purchase the manner and content of living used to define a specified budget level; to estimate aggregate costs of consumer goods as a basis for developing public policies; or to pre pare estimates of the number of families living below the specified budget level. Budgets also provide benchmarks for administrative deter minations, as required by a number of existing laws or policies of social, welfare, and educa tional agencies; e.g., to establish criteria of eligibility for public assistance, public housing, support services for individuals in job develop ment programs, subsidized medical or mental health, guidance services, or college scholar ship aid. In addition to their primary use as tools in evaluating income adequacy, family budgets are used to measure place-to-place differences in liv ing costs, as a basis for family counseling, in wage negotiations, and as an aid in consumer education. Locality indexes based on the BLS budgets reflect differences in costs of established resi dents in a community. Rental costs, for ex ample, are based on the averages for occupied dwellings and are not a valid measure of the costs of vacant units available to new residents. Similarly, the costs of maintaining a home pur chased 7 years ago, while an appropriate mea sure for an established, budget-type family, does not provide information on the relative costs of purchasing homes in current markets. The cost of food reflects not only differences in price levels but also, and more important, differ ences in regional preference patterns in the choice of food to meet nutritional standards.*2 1 The indexes, therefore, are more appropriate as research tools in analyses of the relationship between income and costs of established resi dents in different locations than as measures of differences in costs for families moving from one location to another. u For a discussion of the uses of family budgets, see Tech nical Reference Nos. 2, 4, 8, and 17. 1 For a measure of the effect on food costs of price-level 2 differences versus regional differences in the choice of foods, see Technical Reference No. 1. FAMILY BUDGETS Technical References 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. Brackett, Jean C., “ Intercity Differences in Family Food Budget Costs,” Monthly Labor Re view, October 1963, pp. 1189-1194. An analysis of the effects on food budget cost estimates of using for all cities a single set of weights representing urban U.S. food patterns, or different weights for each city reflecting the food preferences of the region in which the city is located. Brackett, Jean C., “ New BLS Budgets Provide Yardsticks for Measuring Family Living Costs,” Monthly Labor Review, April 1969, pp. 3-16. Provides a nontechnical description of the concepts and procedures used to develop the budgets for a four-person family at three levels of living, a summary of the spring 1967 cost estimates and locality indexes, and a discussion of the appropriate uses of the budgets. Clorety, Joseph A., “ Consumption Statistics: A Technical Comment,” How American Buying Habits Change, Chapter X , 1959, pp. 217-242. Presents a representative cross-section of budgets compiled in this country during the 20th century. Shows average dollar cost figures for the total and for the major components of each budget. Hawes, Mary H., “ Measuring Retired Couples’ Living Costs in Urban Areas,” Monthly Labor Review, November 1969, pp. 3-16. Includes estimates of the costs of budgets for three levels of living for a retired couple in spring 1967 and budget-based locality indexes. Describes uses of these budgets as tools in de termining eligibility for various programs and in helping older couples to evaluate their own spending habits. Lamale, Helen H., “ Changes in Concepts of Income Adequacy Over the Last Century,” Ameri can Economic Review, May 1958, pp. 291-299. An analysis of the relationship over time between actual levels of living in the United States and the goals or standards of living which have been accepted in different historical periods and for different purposes. Lamale, Helen H., “ Poverty: The Word and the Reality,” Monthly Labor Review, July 1965, pp. 822-827. Discusses the standard budget approach to the evaluation of income adequacy for different family types and in different geographical locations and estimation of the extent of poverty in the United States. Lamale, Helen H. and Stotz, Margaret S., “ The Interim City Worker’s Family Budget,” Monthly Labor Review, August 1960, pp. 785-808. Estimates the cost of a “ modest but adequate” standard of living for a husband, wife, and two children (living in rented housing), at autumn 1959 prices, in 20 large cities and their suburbs. Includes the detailed list of the goods and services used to define the living standard for the 1950’s, and describes the way this list was developed and priced. Orshansky, Mollie, “ Budget for an Elderly Couple: Interim Revision by the Bureau of Labor Statistics,” Social Security Bulletin, December 1960, pp. 26-36. A summary report on “ The BLS Interim Budget for a Retired Couple.” (See Reference No. 10.) Includes a discussion of various conceptual problems encountered in developing norm ative living costs estimates for a retired couple, and some limitations of this particular bud get. Ruiz, Elizabeth, “ Spring 1970 Cost Estimates for Urban Family Budgets,” Monthly Labor Review, January 1971, pp. 59-61. Summarizes annual costs and comparative cost indexes at three levels of living for fam ilies of four persons in Spring 1970. Equivalence scale values are applied to four-person family consumption costs to obtain comparable estimates for other family types. Stotz, Margaret S., “ The BLS Interim Budget for a Retired Couple,” Monthly Labor Review, November 1960, pp. 1141-1157. Estimates of the cost of a “ modest but adequate” standard of living for a man age 65 or over and his wife (living in rented housing), at autumn 1959 prices, in 20 large cities and their suburbs. Includes the detailed list of the goods and services used to define the living standards for 1950’s; and describes how this representative list was developed and priced. U.S. Department of Labor, Bureau of Labor Statistics, “ City Worker’s Family Budget for a Moderate Living Standard, autumn 1966” (Bulletin 1570-1, 1967). Describes changes in this budget over the last two decades, and gives autumn 1966 costs 73 74 BLS HANDBOOK OF METHODS Technical References—Continued 12. 13. 14. 15. 16. 17. 18. 19. for urban United States and costs and comparative indexes for 39 metropolitan areas, and 4 nonmetropolitan regions. U.S. Department of Labor, Bureau of Labor Statistics, “ Revised Equivalence Scale for Esti mating Equivalent Incomes or Budget Costs by Family Type” (Bulletin 1570-2, 1968). / Includes scale values for selected family type which can be used to approximate total costs of consumption for the three budget levels. Also includes a summary and discussion of the status of research on family equivalence scales. U.S. Department of Labor, Bureau of Labor Statistics, “ City Worker’s Family Budget Pric ing, Procedures, Specifications, and Average Prices, Autumn 1966” (Bulletin 1570-3, 1968). Reports on pricing methodology used in the intermediate budget and includes U.S. urban average prices and averages for five metropolitan areas for selected items priced for the budget. U.S. Department of Labor, Bureau of Labor Statistics, “ Retired Couple’s Budget for a Mod erate Living Standard, Autumn 1966” (Bulletin 1570-4, 1968). Describes changes in this budget over the last two decades, and gives autumn 1966 costs for urban United States and costs and comparative indexes for 39 metropolitan areas and four nonmetropolitan regions. U.S. Department of Labor, Bureau of Labor Statistics, “ Three Standards of Living for an Urban Family of Four Persons, Spring 1967” (Bulletin 1570-5, 1969). Describes budgets for a four-person family at three levels of living. Explains in detail the concepts, procedures, data sources, and estimating methods, and provides lists of goods and services priced. Includes spring 1967 costs and locality indexes. U.S. Department of Labor, Bureau of Labor Statistics, “ Three Budgets for a Retired Couple in Urban Areas of the United States, 1967-68” (Bulletin 1570-6, 1970). Describes budgets for a retired couple at three levels of living. Explains in detail the con cepts, procedures, data sources, and estimating methods, and provides lists of goods and serv ices paid. Includes spring 1967 costs and locality indexes. (A supplement to this Bulletin pro vides costs and indexes for 1969-70. U.S. Department of Labor, Bureau of Labor Statistics, “ Report of the Advisory Committee on Standard Budget Research,” June 1963. Recommendations formulated by a committee of experts with extensive experience in us ing standard budgets on the needs for various types of budgets, general concepts of the stan dards of living to be described by the budgets, and methodological and other problems asso ciated with estimating and publishing budget costs. Includes a selected bibliography on the major uses of standard budgets. U.S. Department of Labor, Bureau of Labor Statistics, “ Workers’ Budgets in the United States: City Families and Single Persons, 1946 and 1947,” (Bulletin 927, 1948). Concepts, definitions, and techniques used in developing the original City Worker’s Fam ily Budget for a four-person family, detailed list of goods and services priced, and 1946-47 cost estimates for 34 cities. Also an historical survey of family budgets, and summary data on State budgets for single women workers. “ A Budget for an Elderly Couple,” Social Security Bulletin, February 1948, pp. 4-12. Estimates of the cost of a “ modest but adequate” standard of living for a couple age 65 or older, at March 1946 and June 1947 prices, in eight large cities. (Concepts and techniques used to compile this budget were parallel to those employed in developing the original BLS City Worker’s Budget. See Reference No. 18.) — Je a n C. B r ack ett Chapter 10. Consumer Prices Background The Consumer Price Index was initiated during World War I when rapid changes in living costs, particularly in shipbuilding centers, made such an index essential in wage negotiations. To provide appropriate weighting patterns for the index, studies of family expenditures were conducted in 92 industrial centers in 1917-19. The Bureau of Labor Statistics began publication of indexes for 32 individual cities in 1919. Regular publication of U.S. city average indexes was not begun until 1921, but indexes were estimated back to 1913.1 Since that time the weighting factors, the list of items included in the market basket, and the cities in which price data were collected for calculating the index have been updated several times. Because people’s buying habits had changed substantially by the mid-1930’s, a new study was made covering expenditures in the years 1934-36 which provided the basis for a comprehensively revised index introduced in 1940 with retroactive calculations back to 1935. During World War II, when many commodities were scarce and goods were rationed, the index weights were adjusted temporarily to reflect these shortages. Again in 1950, the Bureau made in terim adjustments, based on surveys of consumer expenditures in seven cities between 1947 and 1949, to reflect the most important effects of immediate postwar changes in buying patterns.1 This ad 2 justment was followed by the first comprehensive postwar revision of the index, which was com pleted in January 1953.3 At that time, not only were the weighting factors, list of items, and sources of price data updated, but many improve ments in pricing and calculation methods were in troduced. Also, coverage of the index was extended to small cities so as to represent all urban wage-earner and clerical-worker families. The most recent revision was completed in 1964, with the introduction of new expenditure weights based on spending patterns in 1960-61, and updated samples of cities, goods and services, and retail stores and service establishments. The manner in which the index has been used and its acceptance by the public have changed from time to time. It has seen many appraisals, criticisms, and investigations. Perhaps the most far-reaching study was conducted during World War II by the President’s Committee on the Cost of Living.4 The House Committee on Education 5 and Labor conducted a detailed examination of the index in 1951.® The most recent study was made by the Price Statistics Review Committee, ap pointed by the National Bureau of Economic Research, at the request of the Office of Statistical Standards of the Bureau of the Budget, to review all government price statistics.6 As a result of these investigations and the Bureau’s continuing efforts to improve the index, changes in coverage, collection, and calculation procedures have been introduced at various times. Examples of these changes include the addition of medium and small cities to the city sample in 1953, the extension of coverage to include single workers in 1964, and institution of direct pricing of restaurant meals in 1953. 1 Collection of food prices back to 1890 had been initiated in 1903. During the course of the 1917— 19 expenditure survey, retail prices for other articles were collected in 19 cities for December of each year back to 1914 and In 13 other cities back to December 1917 only. Retail prices of food and wholesale prices of other items were used to estimate price change from 1914 back to 1913. 2 See Interim Adjustment of Consumers’ Price Index (BLS Bulletin 1039, 1951). 3 See Consumer Prices in the United States 195S-58 (BLS Bul letin 1256). 4 Report of the President’s Committee on the Cost of Living, Office of Economic Stabilization, Washington, 1945. 5 Consumers’ Price Index— Report of a Special Subcommittee of the Committee on Education and Labor. House of Representa tives, 8 2 /1 , Subcommittee Report No. 2, Washington, 1951. • Oovernment Price Statistics— Hearings before the Subcom mittee on Economic Statistics, Joint Economic Committee, Con gress of the United States, 8 7 /1 , Part 1, Washington, Janu ary 24, 1961. 75 76 BLS HANDBOOK OF METHODS Description of the Index Concept and Scope.7 The Consumer Price Index (CPI) is a statistical measure of changes in prices of goods and services bought by urban wage earners and clerical workers,8 including families and single persons. The index often is called the “ cost-of-living index,” but its official name is Con sumer Price Index for Urban Wage Earners and Clerical Workers. It measures changes in prices, which are the most important cause of changes in the cost of living, but it does not indicate how much families actually spend to defray their liv ing expenses. Prior to January 1964, the com plete name for the index was: Index of Change in Prices of Goods and Services Purchased by City Wage-Earner and Clerical-Worker Families to Maintain Their Level of Living. The Consumer Price Index is a weighted aggre gative index number with “fixed” or “ constant” annual weights, or it often is referred to as a “mar ket basket” index. Thus, in the Consumer Price Index the procedure is to measure price change by repricing at regular time intervals and com paring aggregate costs of the goods and services bought by consumers in a selected base period. The quantities of these goods and services are kept constant except at times of weight revisions. Since new weights .are introduced without affecting the index level, any change in aggregate costs is due to price change. The quantities represent not only annual consumption of the goods and services actually priced for the index but also consumption of related items for which prices are not obtained, so that the total cost of the market basket repre sents total consumer spending for goods and services. The index represents price change for every thing people buy for living—food, clothing, auto mobiles, homes, housefumishings, household sup plies, fuel, drugs, and recreational goods; fees to doctors, lawyers, beauty shops; rent, repair costs, transportation fares, public utility rates, etc., in cluding all taxes directly associated with the pur chase of an item and its continued ownership. It deals with prices actually charged to consumers, including sales and excise taxes, since these are an inherent part of the market price the consumer must pay for goods and services subject to such taxes. It also includes real estate taxes on owned homes which are part of the price of homeowner- ship. However, it does not include income or other personal taxes, since they are not associated with prices of specific goods and services although they may have an indirect impact.9 Since 1953, it has treated the purchase of a home in the same way as the purchase of su6h durable goods as automo biles, refrigerators, etc. In the 1964 revision, the index coverage was extended to include single consumer units1 * in 0 addition to families of two or more.1 The average 1 size of families represented in the index' is about 3.7 persons, and their average family income in 1960-61 was about $6,230 after taxes. The average income after taxes of single persons represented in the index was about $3,560. 'Weighting Structure. The annual consumption patterns represented in the index since January 1964 were determined in the Survey of Consumer Expenditures (C E S)1 in 66 Standard Metropol 2 itan Statistical Areas (SMSA’s) and smaller cities 7 See article by Sidney A. Jaflfe, “ The Statistical Structure of the Revised CPI” , Monthly Labor Review, August 1964, pp. 9 1 6 924. * 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 crafts men, foremen, and kindred workers, such as carpenters, book binders, e tc.; operatives and kindred workers, such as apprentices in the building trades, deliverymen, furnacemen, smelters, and pourers, e tc.; clerical and kindred workers; service workers, ex cept private household, such as waitresses, practical nurses, e tc .; sales workers; and laborers, except farm and mine. It excludes professional, technical, and kindred workers, such as engineers and teachers; farmers and farm managers; managers, officials and proprietors, except fa rm ; private household workers; and farm laborers and foremen. A consumer unit Included In the 1960-61 Survey of Consumer Expenditures was classified in the index group If more than half the combined income of all family members was obtained in a wage-earner and clericalworker occupation and at least one family member was a full time earner (i.e., worked 37 weeks or more during the survey year). •For a more detailed discussion, see “ Taxes and the Con sumers’ Price Index,” Monthly Labor Review, January 1953, pp. 53-57. (Reprint 2090.) 10 A person living alone or in a household with others from whom he was financially independent; i.e., his Income and ex penditures were not pooled. Terms such as "single persons, single workers, singles, etc.” used subsequently, refer to single consumer units and not to unmarried persons. u From January through November 1964, the Bureau computed a separate index applying only to families of two or more, for comparability with the previous series. Since the same items were priced for families and singles, the overall movement of the separate index was Identical with the index including single workers and it was discontinued. 7 The Survey of Consumer Expenditures is discussed in 3 Chapter 8 of this bulletin. The selection of the expenditure sur vey and CPI city samples is described in detail in an article by Marvin Wilkerson, “ The Revised City Sample for the Consumer Price Index,” Monthly Labor Review, October 1960, pp. 10781083. (Reprint 2352) CONSUMER PRICES T able 1. 77 C it i e s , P o p u l a t io n W e ig h t s , a n d P r ic in g S c h e d u l e f o b t h e R e v is e d C o n s u m e r P r ic e I n d e x Pricing schedule * Other items City and size stratum Population weight > Food* Schedule* Samples M 1 A. Standard metropolitan statistical areas of 1,400,000 or more in 1960: Baltimore, Md.................... ................................... ........................................ Boston, Mass......... ........... ............................................................................... Chicago-Northwestern Indianai.................................................................. 4 * Cleveland, Ohio................................................................... ........................... 1.402 1.930 5.552 1.325 1A, 2B 1A, 2B 1A, IB, 2A, 2B 1A, 2B X Detroit. Mich....... .......................................................................... ................. Los Angeles-Long Beach, Calif........................................ .......................... New York-Northeastern New Jersey4 .......................................................... Philadelphia, Pa.............................................................................................. 2.895 5.017 12.577 2.703 1A, 2B 1A, 2B 1A, 2B 1A, 2B 1.565 1.428 2.372 1.255 1A, 2B 1A, 2B 1A, IB, 2A, 2B 1A, IB, 2A, 2B B. Standard metropolitan statistical areas of 260,000 to 1,399,999 in 1960: * Atlanta, Qa............................................................................ .......................... Buffalo, N .Y ..................................................................................................... Cincinnati, Ohio-Ky... .............................................................. ................... Dallas, Tex....................................................................................................... Dayton, Ohio................................................................................................... 2.934 2.347 .740 2.934 1.096 1A, 2B 1A, 2B 1A,2B 1A, 2B 1A, 2B Denver, Colo..................................................................................................... Hartford, Conn.................... ............... ..................................... .................... Honolulu, Hawaii........................................................................................... Houston, Tex.......................................................................................... ...... Indianapolis, In d ................................... ............................................... ........ 1.838 2.348 .354 .999 1.095 1A, 2B 1 1A, 2B 1A, 2B 2 Kansas City, Mo.-Kans................................................................................ Milwaukee, Wis............................................................................................ Minneapolis-St. Paul, Minn____________ ______________________ ______ .710 .850 1.042 1A, 2B 1A, 2B 1A, 2B Nashville, Tenn................................................................................................ San Diego, Calif....................................................... ....................................... Seattle, Wash....... ........... ............ .................................................. .............. Wichita, Kans.................................................................................... ........... . . 2.933 .672 1.837 1.096 2 1A, 2B 1A, 2B 1A, 2B X C. Standard metropolitan statistical areas of 50,000 to 249,999 in 1960: Austin, Tex............................ .............................. .......................... ............... Bakersfield, Calif..................................................... ...................... ................ Baton Rouge, L a............................................................................................ Cedar Rapids, Iowa........................................................................................ Cbampaign-Urbana, 111............................................................................. . 1.250 1.323 1.250 1.284 1.284 1 2 2 1A, 2B 1A, 2B 1 2 2 1A, 2B 1A, 2B X Durham, N .C....................................................... ............ ............... ............. Green Bay, Wis...................................... ....................................................... Lancaster, Pa........... ..................... ........................................................... ...... Orlando, Fla................................................... ................................................ Portland, Maine............................................................................................... 1.250 1.284 1.803 1.250 1.803 1A,2B 1 1 1A, 2B 2 1A,2B 1 1 1A, 2B 2 D. Urban places of 2,500 to 49,999 in 1960: Anchorage, Alaska.................. ................................... ................................... Crookston, Minn_________________________________ _________________ Devils Lake, N.Dak.................................................................................. ...... Findlay, Ohio.............. .............................. .................................................... Florence, Ala....... ............................................................................... ............. .065 1.352 1.352 1.352 1.227 1.2 1 2 1 1 1.2 1 2 1 1 Kingston, N .Y ______________________________________________________ Klamath Falls, Oreg........................................................................... ............. Logansport, Ind....... ....................................................................... .............. Mangum, Okla................................................................................................ . Martinsville, Va.............................................................................. ............ 1.171 1.338 1.352 1.226 1.227 2 1 2 1 2 2 1 2 1 2 X McAllen, Tex.................................................................................................... Millville, N.J..................................................................................................... Niles, Mich........................................................................................................ Orem, U tah .......................................................... ..................................... . Southbridge, Mass.................................................. .................................. ...... 1.227 1.171 1.351 1.339 1.170 2 2 1 2 1 2 2 1 2 1 X X Union, S.C................................................................... .................................. Vicksburg, Miss................................................................................................ 1.227 1.226 1 2 1 2 3 X X X X Pittsburgh, Pa........................................................... ..................................... St. Louis, Mo..... ................................................ ..................................... ...... San Francisco-Oakland, Calif......................................................................... Washington, D.C_________________ _____ ____ ____ _______ ____________ 2 X i The 18 largest Standard Metropolitan Statistical Areas as defined for the 1960 Census of Population were selected on a certainty basis and repre sent 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. * 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. 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 * 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. l=January, April, July, and October. 2=February, May, August, and November. 3 = March, June, September, and December. 4 Standard Consolidated Areas. Population weights revised for this group beginning January 1966 78 BLS HANDBOOK OF METHODS covering the period 1960-61, except for Anchor age, Alaska, which was surveyed for 1959. Ex penditure records were obtained from the 4,343 urban families of two or more persons and from the 517 single workers included in the survey who were classified as wage earners or clerical workers. Sixteen cities in the smallest size class which were included in the 1960-61 survey are not in cluded in the CPI sample for pricing. However, in the weight derivation, expenditures by con sumer units in these small cities were included with those for the 16 small cities priced for the index, so that the resulting weights are based on the total sample of 32 small cities. In establishing index weights from the detailed expenditure data, about 400 items were selected objectively to compose the “market basket” for current pricing, beginning with the January 1964 “new series” indexes. Not all items are priced in every SMS A or city. In order to make possible estimates of sampling error, two subsamples of items have been established. These are priced in different areas and in different outlet samples, as indicated in table 1. The population weights shown in the table are used to combine price data for the 56 individual areas in the CPI sample into a U.S. city average. They were derived from the 1960 Census of Population but adjusted to repre sent the wage-earner clerical-worker coverage of the CPI. For the 18 largest SMSA’s, which are included in the sample with certainty, the weights are based on their respective populations only. For the remaining cities, which were selected by prob ability sampling methods to represent all other urban places, the weights represent not only the specific city population, but also the population of all cities in the same region and size class. Thus, every city in the same region and size class (other than the 18 largest) has identical population weights.1 * 1 3 The list of items priced, includes all the most important goods and services and a sample of the 1 Six additional B size Standard Metropolitan Statistical Areas 1 were added to the national index in January 1966. Since they were selected outside the probability framework, they were as signed only their own population weight which was substracted from that of other stratum B cities in the same region. 1 Minor weight revisions are introduced by linking. 4 1 For a more complete discussion, see The Consumer Price 5 Index: Pricing and Calculation Procedures, unnumbered paper by Doris P. Rothwell, Bureau of Labor Statistics, U.S. Depart ment of Labor, 1964. less important ones. In combination, these repre sent all items purchased. Table 4 contains a com plete list of the items priced for the index. The content of this market basket in terms of items, quantities, and qualities is kept essentially un changed 1 in the index calculation between major 4 revisions so that any movement of the index from one month to the next is due solely to changes in prices. A comparison of the total cost of the mar ket basket from period to period yields the measure of average price change. In the selection of the item sample for the re vised CPI, except for the choice of the particular quality or variety of the item to be priced (speci fication), probability sampling techniques were used, as described later. The more important items are included in the sample with certainty. The remaining items within a given expenditure class were selected with probability. The relative im portance of a certainty item represents consumer spending for that item only. “ Probability” items represent all other items within the expenditure class. The total weight of these items is divided equally among the probability items within an ex penditure class. Table 2 shows relative im portances in the national index as of December 1963. Individual relative importances are not shown in the table for probability items; rather their combined" importance is shown as “other priced items” in each expenditure class. Data Sources and Collection Methods Prices are obtained in the 56 area CPI sample by personal visit to a representative sample of nearly 18,000 stores and service establishments where wage and clerical workers buy goods and services, including chain stores, independent grocery stores, department and specialty stores, restaurants, pro fessional people, and repair and service shops.1 5 Rental rates are obtained from about 40,000 ten ants. Reporters are located both in the city proper and in suburbs of each urban area. Co operation is completely voluntary. Prices are collected in each urban location at intervals ranging from once every month to once every 3 months, as indicated in table 1, with a few items surveyed semiannually or annually. Be cause food prices change frequently, and because 79 CONSUMER PRICES T a b l e 2. C o n s u m e r P r ic e I n d e x ( N e w S e r ie s ) 1 R e l a t iv e I m p o r t a n c e o p M a jo r G r o u p s , S p e c ia l G r o u p s I n d iv id u a l I t e m s S e l e c t e d w it h C e r t a i n t y * D e c e m b e r 1963 Components Percent of all items December 1063 100.00 M ajor G roups 22.43 33.23 10.63 13.88 19.45 5.70 2.75 5.04 5.06 S pecial G roups All items less shelter..____ ____ ____ _____________________ Nondurables_______ ______ ___ _____________________ Services... _________ .. _________________________ Commodities less food_______ _________________________ _ Nondurables less food.___ ________ ____ __________ Housefumishings_______ _________ _________________ 79.85 77.57 65.97 47.19 18.78 34.03 43.54 24.76 9.16 7.65 15.60 5.69 4.72 28.53 13.47 4.86 4. 56 5.64 I ndividual I tems Food__________________ ________ _______________________ Food at home______________________________ ____ ___ Cereals and bakery products_____________________ Meats, poultry, and fish_________________________ Meats___II . . ___________________________ .. Pork.. _____________________________ Other priced items________ __________ Fish.......................... ............................................ Dairy products___________________ _________ Mflk, fresh (grocery).. _____________________ Milk! fresh (delivered)___ __________ _______ • Butter_____________ 1 ___________ _ ... ... Other priced items.. ______ _________ ______ _ Fruits and'vegetables____________________________ Potatoes_________________________ _______ Other priced items__________ ____ _______ Other priced itemsl______________________ Between meal snacks____ _______________________ See footnotes at end of table. 22.43 17.89 2.45 .80 1.65 .60 1.05 5.63 4.45 2.21 .57 .80 .84 1.30 .36 .30 .64 .94 .73 .51 .22 .45 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 4.54 3.75 .79 Components an d Percent of all items December 1963 Housing____________ __________________________________ !3helter_________ ___________________________________ Rent_____________ ____ _______________ _____ ____ Hotels and motels_______________________________ Homeownership___ _____________________________ Purchase and financing............. ................. ........ Home purchase.. .1________ ________ _____ Mortgage interest____________ _____ ______ Taxes and insurance_________________________ Real estate taxes____ _____ _______________ Property insurance.__________ ___________ Maintenance and repairs_____________________ Commodities...I_________________________ 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 housefumishings!_______________ _________ Furniture__________ _____ ______________________ Bedroom suite______________________________ Living room suite...________ ________________ Other”priced items.. _________ __________ . . . Floor coverings____________________ _____________ Rugs, soft surface____ ____ ____________ ____ _ Other priced items.. . . . __________________ Appliances______ ____ __________ _______________ _____________________ ’ 'Refrigerators_____ . Other priced items____________ _______ ______ Other housefumishings _____ _______________ . . Housekeeping supplies________ _____________ .. Housekeeping services_________ __________________ Domestic service.___________________________ Babysitter__ ________________ ____________ Postage__ ___________________________ _____ Other-priced items______________ _______ _____ 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 7.82 .61 1.44 .28 .28 .88 .48 .34 .14 1.36 .28 1.08 .83 1.55 1.55 .26 .29 .23 .77 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’coats_________ ________ ______ ________ 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__________ _______________ 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 1.49 .79 .44 .35 .68 Transportation_______ _________________________________ Private transportation______________________________ Autos and related goods_____ _____ ______________ 13.88 12.64 9.02 5.02 2.55 2.47 3.28 3.05 .23 .72 3.62 .98 2.64 1.42 .37 .04 .18 .63 1.24 .78 .14 .07 .20 .05 New cars_______ ____ ____ _______________ Used cars.................. ............ ....................... Gasoline and motor oil................................... ...... Gasoline________________________ ________ Motor oil________________ ____ ______ ____ Auto parts___________________ ____ _______ _ Automobile services___________________ ______ Auto repairs and maintenance....................... Other automobile expenses________ _______ Auto insurance.."_____________________ Registration fees.____________________ Drivers’ license______ ________________ Parking fees_________________________ Auto financing chargesJ .......................... Public transportation_______I____ I___ ____ _____ ____ Local transit____________________________________ Taxicabs_________________________________ ______ Train fares___________ ______________ ___________ Airplane fares............... ............... .............................. Intercity bus fares........................................................ BLS HANDBOOK OF METHODS 80 T a b l e 2. C o n s u m e b P r ic e I n d e x ( N e w S e r ie s ) 1 R e l a t iv e I m p o r t a n c e o f M a jo r G r o u p s , S p e c ia l G r o u p s I n d iv id u a l I t e m s S e l e c t e d w it h C e r t a i n t y * D e c e m b e r 1963— Continued Percent ol all Items December 1963 Components Health and recreation................. ............................................ Medical care........................................................................ Drugs and pharmaceuticals_______ ______ _________ Professional services............... Family doctor, house visit Family doctor\office visit Dentists’ tees Other priced Items .......... . . 7 . . . . . . . . . Nonhospital services_________________________ Toilet goods Services________________________________________ Rending and recreation__ _ ....... Recreation ................................. Recreational goods................................................ TV sets ............. Other priced Items.......................................... 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 5.94 4.36 2.78 .63 2.15 Percent o all Items December 1963 Components Health and recreation—Continued Reading and recreation—Continued Recreation—Continued Bowling fees_____________________________ Other priced items_______________________ College tuition ._ . . . . . _____ . Other priced items___________________________ Other priced items___________________________ Alcoholic beverages______________________________ Beer_______~_______________________________ Away from home M iscella n eou s 3 3 _ an d . ....... . . . . . . . . . . 1.58 .38 .36 .84 1.58 .50 .23 .85 5.06 1.89 1.74 .15 2.64 1.06 .78 .80 .53 .28 .12 .13 .38 » For a description of the new series, see The Consumer Price Index, (.Revised January 1981). A Short Description. > The list of items priced includes all the more Important goods and services and a sample of the less important ones. In combination, these represent all items Included in the CPI. Weights for Individual certainty items are shown separately. Some of them, however, are represented by more than one specification, but the weights for the individual specifications are not shown. The remaining weight of each expenditure class having both cerainty and probability items was shared equally by the probability items as of December 1963, except in a few cases where weights for duplicated items have double weights. * Not actually priced; imputed from priced items. * Cost of health insurance is imputed to price changes for representative services plus the cost of overhead. For a more complete discussion, see article by James C. Daugherty, “ Health Insurance in the Revised CPI," Monthly Labor Review, November 1964, pp. 1299-1300. * Personal financing charges other than mortgage interest and auto financ ing. foods are a significant part of total spending, food pricing is conducted every month in each urban location. Prices of most other goods and services are collected every month in the five largest urban areas and every 3 months in all other places. Pric ing of food is done each month on 3 consecutive days early in the month; rents and items for which prices are obtained by mail are reported as of the 15th of the month; pricing of other items extends over the entire calendar month. The Bureau uses mail questionnaires to obtain data on streetcar and bus fares, public utility rates, newspaper prices, and prices of certain other items which do not require personal visit by Bureau agents. For a number of items, e.g., home purchase, college tui tion, used cars, magazines, etc., data collected by other Government agencies or private organiza tions are used. To insure that the index reflects only changes in prices and not changes due to quantity or quality differences, the Bureau has prepared detailed spec ifications which describe the physical character istics of the items in the market basket. Specially trained Bureau representatives examine merchan dise in the stores to determine whether the goods and services for which they record prices conform to the specifications. Where the precisely specified item is not sold at a particular retail establishment, the Bureau’s representative quotes prices and obtains a detailed technical description of the item nearest to the physical characteristics of the spec ification, in order to insure that prices will be quoted on the same quality and quantity from time to time. At the first pricing in an outlet the agent selects the volume selling item meeting spec ification, making sure that it is regular merchan dise in good condition and available in a customary assortment of colors, patterns, etc. At the subsequent pricings the agent prices the identical item if it is still available in a reasonable assortment and selling in substantial volume. I f it is not, she must substitute another volume selling item, meeting specification, if possible. I f she cannot, she prices an item deviating from specifi cation. Prices of substitute items meeting specifi cation are compared directly. Prices of deviating items are introduced by linking or splicing in such a way that the difference in price between the specification and the deviating item is not re flected as a price change. I f it is possible to obtain an estimate of the value of the quality difference, prices of the previous item are adjusted by the CONSUMER PRICES quality estimate and compared with prices for the current month. When the sample of reporters is changed for any reason, prices from the new reporter also are intro duced by linking. Sampling A complicated index such as the CPI must be based on a whole complex of samples.1 A sample 6 o f cities or areas is required in which expenditure surveys and price collection will be conducted. Within each area there must be a sample of fam ilies or consumer units, from which consumer ex penditures will be obtained. It is convenient, but not essential, that the city sampling points be the same for price collection as for the Consumer Ex penditure Surveys. 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. Samples of outlets are needed at each sampling point in which price quotations are obtained for the selected items. Finally, pricing usually is done at a specific time of the month or quarter so there is, in effect, a sampling of time. In the 1964 revision, the Bureau of Labor Statistics used probability sampling to a greater extent than had been done previously, despite the well-known difficulties involved. At the same time, the Bureau attempted to include in the CPI sampling design some method for obtaining an estimate, even if only a crude one, of sampling error. Probability sampling is a necessity, of course, if this is to be done in a conventional man ner. However, even if probability sampling could be followed rigorously through all the complicated CPI structure, the mere computational load would be so extensive that it would be impractical to com pute measures of error except by some “ simple” approach. The objective, therefore, has been ap proached by the “ replication” method. The sample design includes an ex post facto pair ing of probability cities (or Standard Metro politan Statistical Areas), two replicated item samples, and replicated outlet samples. In addi tion to the minimum program, designed to produce an estimate of the total sampling error in the index 81 from all sources, the structure includes more ex tended replication in selected cities aimed at per mitting some evaluation of the components of the error, that is, variation in sampling results due to sampling of cities, items, and outlets. City Sample. A core sample of 50 SMSA’s (see footnote 13) or smaller cities for the index, sup plemented by 16 additional D size cities for the family expenditure surveys was the maximum size consistent with available budget. These addi tional D size cities were surveyed because expendi ture patterns are more variable among small cities than among large cities. The primary sampling units (PSU’s) are Standard Metropolitan Sta tistical Areas as they were 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. Because 1960 Census data were not then available, the measure of size used in sample selection was the estimated urban population as of January 1,1959. The population weights actually assigned are based on 1960 data. The PSU’s were stratified by broad region and by size into 12 regional-size strata. The 12 largest SMSA’s were selected with certainty, that is, they represent themselves in the sample design. Since Alaska and Hawaii have been added in the revised CPI, one sample selection has been allocated to each of these two States. The remaining 36 selec tions are allocated to the 12 regional-size strata on the basis of relative population and relative costs of pricing cities of different size. Four size strata are defined as follows: A. The 12 largest SMSA’s on the basis of urban population, in effect those with population over 1,400,000; B. Other large SMSA’s with urban population greater than 250,000; C. SMSA’s with urban population of 50,000250,000; and D. Nonmetropolitan urban places with popula tion less than 50,000. The method of selection used is known generally as “ controlled selection” which was described by Roe Goodman and Leslie Kish in the September w For a more detailed discussion, see Sampling Aspects of the Revised CPI, unnumbered paper by Marvin Wilkerson, Bureau of Labor Statistics, U.S. Department of Labor, 1964. 82 BLS HANDBOOK OF METHODS 1950 issue of the Journal of the American Statis tical Association (pp. 350-372). This method accomplishes a good geographic dispersion of sampling points across the country. After the initial 50-area sample was selected, the BLS received funds to prepare city indexes for six additional large SMSA’s—Cincinnati, Hous ton, Kansas City, Milwaukee, Minneapolis-St. Paul, and San Diego—as part of a plan to publish indexes for each SMS A with 1,000,000 total popu lation in 1960. These areas were added to the national index in January 1966. Samples of Consumer Units. The CES samples were chosen as subsamples of housing units enu merated in advance Comprehensive Housing Unit Surveys (CHUS) conducted in each area late in the year preceding the actual survey date.1 The 7 CHUS also serve as the source of the samples of rental dwellings for measuring price change in rents, and of owner occupied units for measure ment of changes in property taxes. They also provide data for weights for home purchase. The actual size of the CHUS sample in an area is determined primarily by the rental sample desired and by the proportion of renters in the given area, as estimated from Census data. The number of addresses enumerated in the CHUS is usually many times larger than it is in the samples for surveys of consumer expenditures. Sampling of Items. A classification system has been developed to provide a logical publication framework containing the traditional major ex penditure groups, subgroups, etc., but, in a broader sense, to divide the thousands of goods and services purchased by consumers into meaningful and manageable components of the universe. It pro vides the framework for the selection of the item sample and for the derivation of index weights. Two levels of the classification system are of critical importance. These are: {1) the item level, and (2) the level which defines the finest stratifica tion for the item sampling; that is, the strata to which allocations of items are made and within which probability samples of items are selected. The term “expenditure class” (EC) is given to this 1 The selection of the CES sample Is discussed In chapter 8 7 of this bulletin. level. The expenditure classes are primarily groupings of items which serve similar human needs. Items are grouped within an EC so that they are as homogeneous as possible with respect to their physical characteristics. It is not possible to confine groupings to items which are similar with respect to price movements. Within an expenditure class base period expen diture weights will be held constant; that is, the EC expenditures serve as a way of defining the level of living which is to be held constant until the next major revision of the CPI. The Bureau plans to resample items within an EC between major revisions whenever there is evidence of a major redistribution of relative expenditures or indications that the previous sample of priced items does not adequately represent the class. The connotation of “ item” in the sampling frame is necessarily fairly broad and the items are not of equal homogeneity in the different classes. Gen erally the listing is above that of the final “ speci fied-in-detail” items for which prices are collected. For the most part no attempt has been made to carry probability sampling to this ultimate stage. There were about 1,800 line items in the’expendi ture survey schedule. After extensive experi mentation, using expenditure data from a 1959 pilot survey in Cincinnati, a final sampling frame containing 52 EC’s and 812 items was developed. The list of EC’s and the number o f items in each are shown in table 3. The first step in the selection of the item sample for the revised index was to make a roughly opti mum allocation of the total number of items to be priced to each EC. Factors considered were the relative importances of the EC’s and a rough measure of variability of price movement. As in past revisions of the CPI, the samples were selected on a national basis. Selection of independent samples, city-by-city, is not practical since it would result in a huge list of items to be priced in at least one city and an impossible burden of writing and keeping up with changes in specifications. The two replicated samples of items of the re vised CPI have been selected with “ probability proportional to size,” size being defined as the relative importance of the expenditures for the item to total expenditures for all items. The gen CONSUMER PRICES eral procedure was to array items within a stratum and by using a random start to make regu lar selections along the array. Each of the two replicated samples thus contains “ certainty items;” that is, items which are certain of inclu sion because their relative importance is greater than the selecting interval. The replicated samples also contain some duplicates of items selected but not with certainty. Table 4 contains a listing of the items priced in each sample. For the final selection, relative importances (in the family expenditure pattern) for the condensed sampling frame (52 EC’s and 812 items) were obtained from expenditure data for nine of the cities surveyed for 1960. Ideally, of course, the data should have covered all 66 cities, but such data were not available in time for use in selection of items. Expenditure data for these nine places were weighted together to give preliminary esti mates of U.S. average expenditures. (Final index weights of course are based on complete data for all cities.) The selection of one or more specifications or “ specified-in-detail” items to represent the items selected from the sampling frame has been made in most cases by commodity specialists from expert knowledge of the item. Factors taken into con sideration are the importance and representative ness of particular qualities and the feasibility of describing a selected item clearly enough to permit repetitive price collection. In a few cases where sufficient data existed, it is possible to make a second stage probability selection of specifications. Outlet Sampling. The first big problem encount ered in attempting probability sampling of outlets was to obtain information about the universe of retail and service establishments in a given area. Ideally, names and addresses of such places, infor mation as to type of store or outlet, some indica tion of volume of sales, and preferably fairly specific information as to types of merchandise carried would have been desirable. Comprehensive establishment data were ob tained from a list of firms which report to the Bureau of Old-Age and Survivors Insurance (Social Security Administration, U.S. Depart ment of Health, Education, and Welfare). Using sampling ratios furnished by BLS, master sam ples of retail and service outlets were selected by 83 BOASI. These were supplemented with listings from other sources. In the larger SMSA’s, a two-stage sampling pro cedure has been followed. Samples of neighbor hood and suburban localities and shopping centers have been selected in which pricing outside the downtown area is conducted. These were selected with probability proportional to sales vol ume, using the best available sales data. The list ings of sample outlets were limited to those falling within the sampled areas. The number of food stores priced varies from less than 10 in the smallest cities to about 80 in New York. The number of quotations for non food items per city is quite small; the basic num ber in each outlet sample is four. This means that for the cities in which both item samples are priced eight is the maximum sample size even for items appearing in both item samples. In a few “A ” cities, the sample sizes are set at 5 per sample or a maximum of 10. At the U.S. level, however, the number of quotations is sizeable. In selecting the sample, allocations of quotations were made for each item by type of outlet, based on available sales data, “ where bought” surveys, etc. As a specific example, if eight quotations are required for a particular woman’s shoe specifi cation, the allocation might be three quotations to department stores, two to women’s specialty shops, two to women’s shoe stores, and one to family shoe stores. Specific allocations also are made by loca tion within the SMSA (central business district, neighborhood centers, and suburbs) and, in some cases, to multiunit and independent establishments. In addition to the pricing in regular retail and service outlets, there are a number of special items whose nature requires separate samples of specific types of “ outlets;” for example, samples of phy sicians and other medical specialists, restaurants, dairies, hotels, property owners, etc. Each of these offers its own particular problems. When the original samples, which were selected in Washington, were sent to the regional offices a great many practical problems were encountered, and many expedients and compromises with strict probability procedures were required to complete the initiation of pricing for the revised index. However, even though some deviations from prob ability sampling were inevitable and had been an 84 BLS HANDBOOK OF METHODS T a b l e 4. L is t o f I t e m s P r ic e d f o r t h e R e v is e d C o n s u m e r Sample A EC-1 EC-2 Bakery products_________ EC-3 . 3A Meats, poultry, and fish: Meats: Beef and veal........ 3B Pork. 3C Other meats. EC-4 Poultry___ EC-S Fish........... EC-fl Dairy products EC-7 Fruits and vegetables: Fresh fruits........... EC-8 EC-9 EC-10 EC-11 EC-12 EC-13 EC-14 EC-15 EC-16 EC-17 EC-18 EC-19 EC-20 See footnotes at end of table. I n d e x a s o f D e c e m b e r 19 63 Sample B Corn flakes________ _____ Rice, long and short grain. White bread____________ Whole wheat bread______ Layer cake, plain_______ Flour, white, all-purpose. Cracker meal. White bread. Cookies, cream filled. Cinnamon rolls, frosted. Hamburger, preground......................... Steaks, round, bone-in.______ _______ Steaks, porterhouse, bone-in................. Rump roasts, standing______________ Chuck roasts, bone-in_______________ Veal cutlets, bone-in________________ Pork chops, center cut____ __________ Bacon, sliced_______________________ Pork roasts, loin halves________ _____ Picnics, smoked____________________ Lamb chops, loin___________________ Salami sausage, sliced.......... ............... Frankfurters, skinless.................... ...... Frying chickens, ready-to-cook.......... Chicken breasts, fresh_______________ Fillets or steaks, fresh or frozen *_____ Tuna fish, chunk style______________ Milk, fresh, grocery_________________ Milk, fresh, delivered_______________ Milk, fresh, skim___________________ Ice cream, prepackaged_____________ Butter, salted______________________ Hamburger, preground. Steaks, round, bone-in. Steaks, sirloin, bone-in. Rump roasts, standing Rib roasts, bone-in. Beef liver, sliced. Pork chops, center cut. Bacon, sliced. Pork roasts, loin halves. Ham, whole. Bologna sausage, sliced. Llverwurst sausage, sliced or whole. Ham, canned, domestic or imported. Frying chickens, ready-to-cook. Turkey, fresh or frozen. Shrimp, raw, frozen. Sardines, Maine. Milk, fresh, grocery. Milk, fresh, delivered. Milk, evaporated, canned. Cheese, American process. Butter, salted. Apples, all purpose................................ Bananas, yellow variety....................... Oranges, except Temple or King_____ Grapes, Thompson seedless...... ........... Grapefruit, fresh, pink or white______ Orange juice, fresh____ _____________ Fresh vegetables. Head lettuce_______________________ Potatoes, white__________ ___________ Tomatoes__________________________ Asparagus, green___________________ Carrots, topped, prepackaged.............. Cucumbers------------------------------------Spinach, prepackaged_______________ Processed fruits and vegetables.. Pears, Bartlett, can or jar..................... Lemonade, concentrate, frozen............. Beets, sliced, can or Jar______ _______ Tomatoes, can or jar________________ Dried beans, Navy or Great Northern Other foods at home: Eggs.............................. Eggs, fresh, large, Grade A ---------------Fats and oils................. Margarine, colored......... ...................... Salad dressing, Italian........................... Sugar and sweets.......... Sugar, white, granulated...................... Chocolate bar, plain milk..................... Coflee, can or bag.................................. Nonalcoholic beverages. Carbonated drinks, fruit-flavored........ Tea bags, orange pekoe and pekoe tea. Prepared and partially pre Bean soup, canned, condensed........— pared foods. Spaghetti, in tomato sauce, canned— Mashed potatoes, instant..................... Potatoes, French fried, frozen.............. Food away from home. Restaurant meals: Lunch.............................................. Breakfast......................................... Between meal snacks: Coflee, cup...................................... Carbonated beverages, cup............ Frankfurter on roll.......................... Ice cream, dish............................... Housing: Shelter: Rent.................................................... Rent of house or apartment.......................... Hotel, motel room rates................................ Homeownership: Homo purchase and financing... Home purchase............................................ . Mortgage interest rates................................. Taxes and insurance................... Property taxes, residential........................ . Property Insurance rates: Fire and extended coverage................ . Comprehensive homeownership policy. Maintenance and repairs: Commodities............. Exterior house paint.................................... . Furnace air filters......................................... Packaged dry cement mix........................... . Services...................... r ic e Priced items Groups, subgroups, expenditure classes Food: Food at home: Cereals and bakery products: Cereals and grain products. P Residing houses............................................ Reshingling roofs......................................... Replacing sinks............................................ Apples, all purpose. Bananas, yellow variety. Oranges, except Temple or King. Grapes, Thompson seedless. Strawberries, fresh. Watermelons, whole or sliced. Head lettuce. Potatoes, white. Tomatoes. Cabbage, all varieties except red. Celery, Pascal, stalk. Onions, yellow. Peppers, sweet, green. 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, colored. Salad or cooking oil, vegetable. Grape jelly, pure. Chocolate flavored syrup. Coffee, can or bag. Coffee, instant. Cola drink, carbonated. Chicken soup, canned, condensed. Baby foods, strained. Sweet pickle relish. Pretzels, hard, salted. Restaurant meals: Lunch. Dinner. Between meal snacks: Coflee, cup. Carbonated beverages, cup. Pie, slice. Candy bar. Rent of house or apartment. Hotel, motel room rates. Home purchase. Mortgage interest rates. Property taxes, residential. Property insurance rates: Fire and extended coverage. Comprehensive homeownership policy. Interior house paint. Shelving, Ponderosa pine. Shrubbery, evergreen. Residing houses. Repainting living and dining rooms. Repairing furnaces. 85 CONSUMER PRICES T a b l e 4. L is t of I t e m s P r ic e d for th e R e v is e d C o n s u m e r P r ic e I n d e x Sample A EC-22 Housing—Continued Fuel and utilities. Household furnishings and operation: Textile housefurnishings.............. EC-23 Furniture. EC-24 Floor coverings. EC-25 Appliances____ EC-20 Other housefurnishings. EC-27 Housekeeping supplies. EC-28 Housekeeping services. EC-29 Apparel and upkeep: Men’s and boys’ apparel: Men’s apparel______ EC-30 Boys’ apparel................. EC-31 Women’s and girls’ apparel: Women’s apparel______ EC-32 EC— 33 Girls’ apparel. Footwear. 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........... EC-34 EC-35 Services......... See footnotes at end of table. ecember 1963— Continued Sample B 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................... 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....... .................................. Rugs, soft surface: Broadloom, wool................... ............................ Broadloom, nylon___________________ ______ Rugs, hard surface____ _________ ____ _____ _____ 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_______________________ Carpet sweepers, manually operated____________ Venetian blinds, white, steel or aluminum slats.— Electric drills, hand held_______________________ Detergent, liquid, laundry...................................... Laundry soap for fine fabrics....................... ........... Scouring pads, steel wool__________ ________ ____ Toilet tissue_______ ____ _______________ _______ Domestic service, general housework....................... Baby sitter service___________ _______ __________ Postal services_________________________________ Laundry flatwork, finished service_______ ______ _ Licensed day care service, preschool child............. Washing machine repairs________ ____ __________ 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. Dining room suites. Sofas, standard, upholstered. Box springs. Cribs. 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. Flatware, stainless steel. Table lamps, with shade. Lawn mowers, power, rotary type. Nalls, 8d (penny) common. Detergent, granules or powder. Air deodorizers, spray type. Paper napkins, embossed. Stationery, envelopes. Domestic service, general housework. Baby sitter service. Postal services. Laundry flatwork, finished service. Reuphoistering furniture. Moving expenses. Suits, year round weight, 2 qualities. Topcoats, wool___________________ Suits, tropical weight........................ Slacks, wool or wool blend_________ Shirts, work, cotton_______________ Shirts, sport, cotton, short sleeves... Shirts, sport, cotton, long sleeves___ T-shirt___________________________ Coats, all purpose, cotton or cotton blend.............. Dungarees, cotton or cotton blend_______________ Suits, year round weight, 2 qualities. Jackets, lightweight. Trousers, work, cotton. Slacks, cotton or manmade blend. Shirts, business, cotton. Socks, cotton. Handkerchiefs, cotton. Coats, heavyweight, wool or wool blend, 2 quali ties. Carcoats, heavyweight, cotton_______ _______ ___ Skirts, wool or wool blend........................................ Skirts, cotton or cotton blend___________________ Dresses, street, chiefly manmade fiber, 2 qualities. Dresses, street, wool or wool blend______________ Dresses, street, cotton__________________________ Housedresses. cotton___________________________ Slacks, lightweight, cotton and carded cotton........ Slips, nylon____ ____________________. __________ Brassieres, cotton____ _________________________ Hose, nylon, full fashioned and seamless, 2 styles.. Anklets, cotton__________________________ _____ Handbags, rayon faille or plastic________________ Raincoats, vinyl plastic or chiefly cotton____ ____ Skirts, wool or wool blend______________________ Slips, cotton blend_____________________________ Handbags, plastic_________________________ ____ Men’s: Shoes, street, oxford, 2 qualities........................ Women’s: Shoes, street, pump, 2 styles. Shoes, evening, pump........... Shoes, casual, pump_______ Houseslippers, scuff.............. Children’s: Sneakers, boys’, oxford type. Other apparel: Commodities. D Priced items Groups, subgroups, expenditure classes EC-21 as o f Diapers, cotton gauze............................................... Yard goods, cotton............ ....................................... Earrings, Pearl, simulated or imitation................... Dry cleaning, men’s suits and women’s dresses___ Shoe repairs, women’s heel lift ............................... Laundry, men’s shirts............................................... Sport coats, wool or wool blend. Undershorts, cotton. Coats, heavyweight, wool or wool blend, 2 qualities Coats, lightweight, topper. Sweaters, wool or acrylic. Dresses, cocktail, street length. Dresses, street, chiefly manmade fiber, 2 qualities. Dresses, street, wool or wool blend. Dresses, street, cotton. Blouses, cotton. Bathing suits, one piece. Girdles, manmade blend. Panties, acetate. Hose, nylon, full fashioned and seamless, 2 styles. Gloves, fabric, nylon or cotton. Coats, lightweight, topper. Slacks, cotton. Shorts, cotton. Dresses, cotton. Robes, duster style, quilted tricot, or percale. Shoes, street oxford, 2 qualities. Shoes, work, high. Shoes, street, pump, 2 styles. Shoes, evening, pump. Shoes, oxford. Dress shoes, girls’, strap. Wrist watches, men’s, imported movement. Wrist watches, women’s, imported movement. Zipper, skirt or neck placket. Dry cleaning, men’s suits and women's dresses. Automatic laundry service. Tailoring charges, hem adjustment. BLS HANDBOOK OF METHODS 86 T a b l e 4. L is t of I t e m s P r ic e d for t h e R e v is e d C o n s u m e r P r ic e I n d e x as of D ecember 1963— Continued Priced items Groups, subgroups, expenditure classes Sample A EC-36 Transportation: Private: Auto purchase- EO-37 Gasoline and motor oil. EC-38 EC-39 Auto parts........................................ Automobile services: Auto repairs and maintenance. EC-40 Other automobile expenses. EO-41 EC-42 Public transportation. Health and recreation: Medical care: Drugs and prescriptions. EC-43 Professional services. EC-44 Hospital services and health in surance: Hospital services.................. Health insurance ’ . New cars: Chevrolet, Impala, 2-door, hardtop.................. Chevrolet, Impala, 2-door hardtop. Chevrolet, Chevelle, 2-door hardtop................ Ford, Falcon, Future, 4-door sedan. Ford, Galaxie 600,2-door hardtop..................... Ford, Galaxie 600,2-door hardtop. Plymouth, Fury III, 4-door sedan.................... Pontiac, Catalina, 4-door sedan. Rambler, Classic 660,4-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. Do. 6 years old.........do............. —............................ . 5 years old, Gasoline, regular and premium.............................. . Gasoline, regular and premium. Motor oil, premium.................................................. Motor oil, premium. Tires, tubeless, retread............................................ . Tires, tubeless, new. Chassis lubrication, complete............................... . Motor tune-up.............. ......................................... . Automatic transmission repair............................... . Auto insurance rates, liability and physical damage. Auto financing charges *.......................................... . Auto registration and inspection fees..................... . Driver’s license fees.................................................. Parking fees, private and municipal...................... . Local transit fares.................................................... . Taxicab fares............................................................ . Railroad fares, coach................................................. Airplane fares, chiefly coach................................... . Bus fares, intercity.................................................. . EC-46 EC-46 Personal care services. See footnotes at end of table. Water pump replacement. Replacing muffler. Front end alignment. Auto insurance rates, liability and physical damage. Auto financing charges.9 Auto registration and inspection fees. Driver’s license feees. Parking fees, private and municipal Local transit fares. Taxicab fares. Railroad fares, coach. Airplane fares, chiefly coach. Bus fares, intercity. Over-the-counter items: Multiple vitamin concentrates........................ . Aspirin compounds. Liquid tonics................................................... . Cough syrups. Cold tablets or capsules..................................... Adhesive bandages, package. Prescriptions: Anti-infectives: Penicillin G buffered tablets..................... . Tetracycline capsules. Sulflsoxazole tablets. Sedatives and hynotics: Phenobarbital tablets.............................. . Secobarbital sodium capsules. Ataractics: Chlordlazepoxide-hydrocbloride capsules.. Meprobamate tablets. Antispasmodics: Propantheline Bromide tablets.................. Phenobarbital and belladonna extract. Cardiovasculare and antihypertensives: Reserpine tablets........................................ . Crystalline digitoxin tablets. Chlorothiazide tablets. Antiarthritics...................................................... Prednisone, tablets. Cough preparations: Terpin hydrate with codeine, elixir. Family doctor, office visits_____________________ Family doctor, office visits. Family doctor, house visits________ _____ _______ Family doctor, house visits. Pediatric care, office visits..... ..................... .......... . Obstetrical cases. Psychiatrists, office visits......................................... Chiropractors and podiatrists, office visits. Routine laboratory tests........................................... Herniorrhaphy, adult. Examination, prescriptions and dispensing of Examination, prescriptions and dispensing eyeglasses. eyeglasses. Fillings, adult, amalgam, one surface.................... . Fillings, adult, amalgam, one surface. Dentures, full upper................................................ . Extractions, adult. Daily service charges: Semiprivate room............................................. . Private room..................................................... . Hospital services: Daily service charges, semiprivate room.......... Daily service charges, private room............... . Operating room................................................. . Nonhospital services: Family doctor, office visit................................. Surgeon’s fees, tonsillectomy/adenoidectomy.. Prescriptions and drugs..................................... Retained earnings (overhead).................................. Personal care: Toilet goods. Sample B Daily service charges: Semiprivate room. Private room. Hospital services: Daily service charges, semiprivate room. Daily service charges, private room. X-ray diagnostic series, upper G.I. Nonhospital services: Family doctor, office visit. Surgeon’s fees, herniorrhaphy, adult. Obstetrical cases. Prescriptions and drugs. Retained earnings (overhead). Toothpaste, standard dentrifloe............................. . Toilet soap, hand milled. Hand lotions, liquid................................................ . Shaving cream, aerosol. Face powder, pressed.............................................. . Deodorants, cream or roll-on. Cleansing tissues.................................................... . ■Home permanent refills. Men’s haircuts......................................................... . Men’s haircuts. Shampoo and wave sets, plain............................... . Shampoo and wave sets, plain. Women’s haircuts_____________ ________________ Permanent waves, cold. 87 CONSUMER PRICES T a b l e 4. L is t of I t e m s P r ic e d fo r t h e R e v is e d C o n s u m e r P r ic e I n d e x a s of D ecember 1963— Continued Priced items Groups, subgroups, expenditure classes Sample A EC-47 EC-48 EC-49 EC-50 Health andrecreation—Continued Reading and recreation: Recreation: Recreational goods----- Recreational services. Reading and education. Other goods and services: Tobacco products___ EC-51 Alcoholic beverages. EC-52 Financial and miscellaneous personal expenses. Sample B TV sets, portable and console.................................. Radios, portable and table models, AM band only. TV replacement tubes............................................. Sports equipment: Golf balls, liquid center..................................... Basketballs, rubber or vinyl cover.................... Outboard motors....................................................... Tricycles.................................................................... Dolls...................................................................... . Stuffed animal........................ ...... .......................... Dog food, canned and boxed.............................. ...... Indoor movie admissions: Adult................................................................... Children’s........................................................... TV repairs, picture tube replacement..................... Bowling fees, evening............ ................................ Golf green fees................ ........................................... Newspapers, street sale and delivery....................... College tuition and fees, undergraduate.................. Magazines, single copy and subscription................ College textbooks, undergraduate........................... TV sets, portable and console. Radios, portable and table models, AM band only. Tape recorders, portable. Sports equipment: Fishing rods, fresh water spincast. Bowling balls. Phonograph records, stereophonic. Bicycles, boys’, 26". Movie cameras, 8 mm, fully automatic lens. Film, 35 mm, color. Cigarettes, nonfilter tip, regular size, pack............. Cigarettes, filter tip, king size, carton..................... Cigars, domestic, regular size................................. Beer, at home, local and national brands............... Whiskey, spirit blended and straight bourbon....... Wine, dessert and table.................................. ......... Beer, away from home............................................. Cigarettes, nonfilter tip, regular size, carton. Cigarettes, filter tip, king size, pack. Cigars, domestic, regular size. Beer, at home, local and national brands. Whiskey, spirit blended and straight bourbon. Wine, dessert and table. Beer, away from home. Indoor movie admissions: Adult. Children’s. Drive-in movie admissions, adult. Bowling fees, evening. Film developing, black and white. Newspapers, street sale and delivery. College tuition and fees, undergraduate. Paperback books, not school or technical. Piano lessons, beginner. Funeral services, adult_____________ ___ ________ Funeral services, adult. Bank service charges, checking account.................. Legal services, short form will. • Two of the largest volume sellers among the following types of fish are priced within each city, since within any given city, all varieties of fish are not available: Frozen ocean perch and haddock; fresh cod, catfish, king salmon, halibut, sole, and haddock. s 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. ticipated, the final samples adhere to the original basic structure to the maximum extent possible. As a result, the main benefits of probability sampl ing have been achieved: lack of bias, representa tion of different types of outlets, sections of each SMSA, etc. Formula. In the absence of major weight revi sions, and ignoring the problems of sampling, the index formula is most simply expressed as: ( 1) 2 \P oQ .o) or by its algebraic equivalent, the dollar weighted average of price relatives: Calculation Procedures (la) The index is a time series. As previously ex plained,TtTs~a weighted average of price changes for a sample of priced items, expressed as a rela tive of avearge prices in a reference base as 100. Weights, which are based on annual consumer expenditures, are kept constant from month to month. The index measures changes as they oc cur. It is not adjusted for seasonal variation.*8 1 The Bureau began publication of seasonally ad justed indexes in 1966, for selected components which show a significant seasonal pattern of price change.1 9 Iio=- Xioo This is the customary, oversimplified way of writing a price index formula to show that the q’s are held constant between major revisions. In actual practice, the basic data for weights are values which include allowances for unpriced 1 For a discussion of the problems involved in using varying 8 seasonal weights, 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. 1 Factors used to compute seasonally adjusted indexes are avail 8 able on request. BLS HANDBOOK OF METHODS 88 items, and the current index is computed by a chain computation procedure, as shown below: (2) 2 (p0q0 ) (p -sQ ) _a xioo where q is a derived composite of the annual quantities purchased in a weight base period for a bundle of goods and services to be represented by the specific item priced p and p' are the average prices of the specific commodities or services selected for pricing (the superscript indicates that the average prices are not necessarily derived from identical samples of outlets and specifications over long periods) i —s is the month preceding a weight revision (most recently, December 1963) i is the current month a is the period of the most recent Consumer Ex penditure Survey (1960-61) from which the revised weights are derived o is the reference base period of the index (1957-59). urban locations included in the index. The fol lowing hypothetical example for pork illustrates the index procedure: Sample item September price October price $0.75 .80 1.00 $0.7725 .82 1.02 Pork chops___ Ham________ Bacon......... (PiQ.a) = ( P i- lfi. In practice, then, the index formula is as follows: (Dec. 1963 (Change from Dec. Index) 1963 to month i— 1) (3) Iuo 2(p0 ) Az(^;_,ga ?0 ) (Change from month i— to month i) 1 2 ( £ x - ^ S(^-lga) > Xioo Illustrative Calculation. Average price changes from the previous pricing period to the current month are expressed as relatives (or ratios) for each item, and the price changes for the various goods and services are combined, using weighting factors based on the importance of the item in consumer spending and that of other items which it represents. This composite importance is called the cost weight of the market basket item. There is a set of separate cost weights for each of the 56 $15.00 8.00 10.00 $15.45 8.20 10.20 33.00 1.03 1.025 1.02 Total........ 33.85 Identical results could be obtained for pork by multiplying prices each period by the implied physical quantities included in the market basket, as the following illustrates. Sample item The (p 0_ or (p'i-sqa) base “ weights” for a g0) given priced item are the average annual expendi tures in a weight base period represented by that item and other similar non-priced items. Al though constant physical weights are implicit in the index, in reality the constant g’s are not calculated separately. In actual practice, the base expenditure for each item is projected forward for each pricing period by the price relative for the priced item! October Ratio September October-icost cost weight September weight (Sept. X ratio) Pork chops______ Ham____________ Bacon___________ Total............ Implied September September cost quantity price weight (pounds) October price $15.00 8.00 10.00 $0.7725 .82 1.02 20 10 10 $0.75 .80 1.00 33.00 October cost weight $15.45 8.20 10.20 33.85 The average change in pork prices is computed by comparing the sum of the cost weights in Octo ber with the comparable sum for September, as follows: October cost weight________ $33.85X100=102.6 September cost weight______ $33.00 This means that pork prices in October were 102.6 percent of (or 2.6 percent higher than) pork prices in September. Although the second method may appear simpler, in reality it is not. Deriving the implied quantity weights is an extra operation, and these implicit quantities change as revised samples are linked in. Furthermore, the second formulation greatly complicates the handling of the numerous substitutions of reporters and items which occur constantly in repetitive index work. Con sequently, the first method is the one actually used for the CPI. The second illustration, however, may assist the user to understand the meaning of the index mechanism. After the cost weights for each of the items have been calculated, they are added to area totals for commodity groups and all items. The U.S. totals are obtained by combining area totals, with each area total weighted according to the proportion of CONSUMER PRICES 89 BLS HANDBOOK OF METHODS the total wage-earner and clerical-worker popula tion which it represents in the index based on 1960 Census figures. Finally, the U.S. totals for the current and previous months are compared to com pute the average price change. Reference Base Period. Since 1962 the index has been calculated on the reference base of 195759=100.2 This means that current prices are 0 expressed as a percentage of prices for the average of the 3 years—1957, 1958, and 1959. An index of 110 means that prices have increased 10 percent since the base period; similarly, an index of 90 means a 10-percent decrease. The index can be converted to any desired base period for which the index is available. This is done by dividing each index number to be converted by the index for the desired base period. Tables of conversion factors are provided on request for most series, enabling users to convert indexes for prior periods on other bases to the current 1957-59 base. Since the 195759 base was adopted in 1962, some indexes have been continued on the bases of 1947-49=100 and 1939=100. These are calculated by the applica tion of appropriate conversion factors to indexes computed on the 1957-59 base. Imputation Procedures. Although prices are not obtained in all 56 cities every month (see table 1 for pricing cycle), it is necessary to represent all 56 cities in each monthly index computation. Be tween quarterly survey dates, for every item ex cept new automobiles, the weights are held at the level of their last pricing, which in effect means prices are estimated unchanged. For new auto mobiles, a price change is imputed to the unpriced cities on the basis of changes in cities surveyed every month For food and apparel items which are sold only at certain seasons of the~yr ar, the index calcula e tion is made in the off-season as if prices of these items changed proportionally with prices of items of a similar nature which are available all year. For example, prices for strawberries when not in season are carried forward on the basis of changes in prices of all other fresh fruits. When the item returns to the market the current price is, in effect, compared with the estimated price implicit in the procedure described. Average Prices. In the calculation of average food prices for publication, the prices used in the index are given special editing since they are not necessarily restricted to a single specified quality and size. Procedures have been devised to calcu late city and U.S. prices for publication which use index values and price relatives extensively. These procedures employ benchmark prices for defined specifications for each of the 56 cities, in which quotations not meeting the specified quality are excluded. Benchmark prices are computed in an independent operation, pooling prices for all outlets rather than as an average of average prices for the two subsamples. The benchmark prices are then adjusted month by month by the price changes reflected in the index. The first bench mark calculation was for April 1964, from which date prices were estimated back to December 1963 and forward to December 1964. New benchmark calculations are planned as of each January. City prices are combined to U.S. averages by the use of the 1960 index population weights.2 1 Average bills for specified quantities of gas and electricity and average prices o f fuel oil, which are published for the largest cities, are the same as those used in the index calculation. Since these are for identical quantities and qualities from month to month, no special editing is required. Item Indexes. Indexes for selected items and groups (commonly referred to as item indexes), were published semiannually during 1964 and 1965. Quarterly publication was resumed in 1966. Although the published indexes refer to March, June, September, and December, some prices for earlier months must be used in their calculation for cities not priced in these months. In June, for example, the indexes are based on (a) June prices in the 5 cities surveyed every month and the 17 cities surveyed in June on a quarterly cycle; and (b) April and May prices in the 17 cities surveyed in each of those months on a quarterly cycle, car ried forward to June as in the composite index calculation. “ The index base was 1 9 4 7 -4 9 = 1 0 0 from 1953 through 1961, 1935-39 = 100 from 1935 through 1952, and 1913 = 100 from 1913 through 1934. " F o r a more detailed discussion, see article by Doris P. Rothwell, “ Calculation of Average Retail Food Prices,” Monthly Labor Review, January 1965, pp. 61-66. 90 BLS HANDBOOK OF METHODS Analysis and Presentation The CPI is made available first at a press con ference, usually held near the end of the month following that to which the data relate. On the same day, the press release is mailed to a list of sub scribers who have immediate need for the data. This release contains a brief description of price changes during the month and several tables of major group and subgroup indexes and percentage changes from selected dates, for the U.S. city aver age and selected large metropolitan areas. It in cludes seasonally adjusted indexes for selected components. A report containing the same text with some additional tables is published about two weeks after the date of the press release. In addi tion each of the Bureau’s six regionaLoffices pre pares and mails a press reteaseforeach of the cities in its region for which CPI figures are published. These releases are timed to coincide with the na tional release. Other monthly reports contain average prices of selected foods and fuels in the largest metropolitan areas. A quarterly publi cation presents the U.S. city average indexes for individual goods and services. The CPI for the United States and for selected areas is published also in the Monthly Labor Re view in the issue dated two months later than the index. The annual Statistical Supplement to the Monthly Labor Review contains indexes for indi vidual goods and services (item indexes) as well as the relative importance of the items in the total index as of December. Average prices for foods and fuels are published in Estimated Retail Food Prices by Cities and Retail Prices and Indexes of Fuels and Electricity. Uses of the Index The most widespread use of the CPI is in wage adjustments and collective bargaining negotia tions. Although this was the primary reason for its beginning, use of the CPI for this purpose declined during the post-World War I and de pression periods. Its use in this way was revived during World War II, but escalation by the index did not receive widespread acceptance until the principle was written into a contract between the United Automobile, Aircraft, and Agricultural Implement Workers of America and the General Motors Corporation in 1948. The number of workers covered by such contracts in 1965 was about 2 million. However, movements of the index have an indirect effect on wages and salaries of many more workers.2 2 The CPI is used extensively to measure changes in purchasing power of the consumer dollar. It is the basis for most estimates of changes in real earnings of labor, and for comparison with pro ductivity measures. Changes in purchasing power are used for such diverse purposes as adjusting royalties, pensions of government and non-govern ment workers, welfare payments, rental contracts, and occasionally alimony payments. One of the most important uses of the index is as a guide to broad economic policy. It is one of the most widely used measures of inflationary pressures. During wartime periods the index and its components have served an important adminis trative function in connection with determination of policies concerning price control and subsidies. In peacetime the index and its underlying sta tistics have played an important part in the gov ernment’s effort to maintain stable wage-price relationships and to judge the advisability of mak ing monetary or tax adjustments. It is one of the chief statistical tools for conversion of the national accounts to constant dollars. Limitations of the Index The CPI is not an exact measure of price changes. It is subject to sampling errors which may cause it to deviate somewhat from the results which would be obtained if actual records of all retail purchases by wage earners and clerical workers could be used to compile the index. These estimating or sampling errors are limitations upon the precise accuracy of the index rather than mis takes in the index calculation. The accuracy could be increased by using much larger samples, but the cost is prohibitive. Furthermore, the index is believed to be sufficiently accurate for most of the practical uses made of it. With the changes in sampling techniques introduced in 1964, the Bu-2 3 2 See article by Francis S. Cunningham, “ The Use of Price 3 Indexes In Escalator Contracts,” Monthly Labor Review, August 1963, pp. 948-952. Reprint 2424. CONSUMER PRICES T a b l e 5. D efinition of the Sum m ary of C h a r a c t e r s rics of th e 91 CPI B e g in n in g 1964 I ndex Title....................................................... Formula (Simplified expression). Base period...... ....................................... Duration............................. ............ Definition of index expenditure weights. Consumer Price Index—U.S. City Average for Urban Wage Earners and Clerical Workers. (£ ) 2p.-j0. 1957-69=100. Series also published on 1947-49 and 1939 bases. January 1964 forward. Average expenditures for urban wage-earner and clerical-worker 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 1963 except for 6 cities added in 1966. P opulation C o verage of E xpenditure Su r vey Place of residence.................. Family size.................................. '* ....... Occupation............................... Length of employment......................... . Income______ ______________________ Urban places of 2500 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 halfof tota lfamiW income from wage-earner and clerical-worker occupations.) At least 1 family member or single consumer unit must have been employed for 37 weeks or more during the survey year in wage-earner or clerical-worker occupations. No criterion as to Income except the qualification above. C ity C overage Population weights................................ Sample of priced cities_______________ Published indexes___________________ 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. 50 metropolitan areas and cities selected originally to represent all urban places in the U.S. including Alaska and Hawaii with populations of 2500 or more in 1960. Six additional areas added in 1966. U.S. and 17 large metropolitan areas for families and single consumer units combined. Indexes for six more large metropolitan areas available in the latter part of 1965. I tem Sample Basis of sample selection........................ Basis for allocation to priced items____ Commodity coverage________________ Number of items priced_____________ Pricing cycle.......................................... . 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 probability selections within expenditure classes. Goods and services purchased for family living, including necessities and luxuries; excluding personal in surance, income and personal property taxes but including real estate taxes and sales and excise taxes. About 400 represented in U.S. index and published city indexes. Certainty items priced in all unpublished cities; other items in 1 of 2 subsamples of unpublished cities. Prices of foods, fuels and a few other items priced monthly in all cities, except for San Diego and Milwaukee where all items are priced quarterly. Prices of most other commodities and services priced monthly in the 5 largest cities, and quarterly I d re maining cities. R epo rt er C o v e ra ge Location............................................. Number of reporters..... ......... , ___ ” Number of quotations............... ...... Pricing technique............ ................ 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). Personal visit of BLS agent except for a few items collected by mail or from secondary sources. Specification pricing but agent is permitted to price deviations from specification under prescribed con ditions. reau is attempting to measure the sampling error in the index.2 3 Another kind of error occurs because people who give information do not always report accurately. The Bureau makes every effort to keep these errors to a minimum, obtaining prices wherever possible by personal observation, and corrects errors when ever they are discovered subsequently. Precau tions are taken to guard against errors in pricing, which would affect the index most seriously. The field representatives who collect the price data and the commodity specialists and clerks who process* ** Preliminary estimates of sampling error were computed and published in Measurement of Sampling E rror in the Con sumer Price In d ex: F irst R esults, by Marvin Wilkerson, paper presented at American Statistical Association meetings, Decem ber 29, 1964. Additional estimates will be made available as work continues on this project. them are well trained to watch for unusual devia tions in prices which might be due to errors in reporting. The CPI represents the average movement of prices for urban wage earners and clerical workers as a broad group, but not the change in prices paid by any one family or small group of families. The index is not directly applicable to any other occu pational group or to non-urban workers. Some families may find their outlays changing because of changes in factors other than prices, such as family composition. The index measures only the change in prices and none of the other factors which affect family living expenses. In many instances, changes in quoted prices are accompanied by changes in the quality of con sumer goods and services. Also new products are 92 BLS HANDBOOK OF METHODS CON SU M ER introduced frequently which bear little resem blance to products previously on the market; hence, direct price comparisons cannot be made. Quoted prices are adjusted for changes in quality, whenever necessary data are available. Technical specifications and highly trained personnel are re lied on to insure comparability of quality of items compared from period to period.2 Nevertheless, 4 some residual effects of quality changes on quoted prices undoubtedly do affect the movement of the CPI either downward or upward from time to time.2 5 Another important limitation of the index is that it measures only time-to-time price change in a given area. City indexes do not show intercity differences in either prices or living costs. They PRICES show only differences in rates of price change from one time to another. Other types of measures are required to show place-to-place differences in liv ing costs. The most recent such measure is “ The Interim City Worker’s Family Budget” which shows the estimated dollar costs of a “ modest but adequate” level of living in 20 large cities and their suburbs in the fall of 1959, which is described in Chapter 9.* M Hoover, Ethel D.f “ The CPI and Problems of Quality Change,” M on thly L abor R eview , November 1961, pp. 1175-1185. Reprint 2378, and Larsgaard, Olga A., and Louise J. Mack, “ Compact Cars In the Consumer Price Index,” M onthly Labor R eview , May 1961, pp. 519-523. Reprint 2368. * See testimony of Evran Clague In Hearings before th e Sub com m ittee on Econom ic S tatistics, Joint Economic Committee, Congress of the United States, 8 7 /1 , Part II, p. 588, Washing ton, May 1, 2, 3, 4, and 5, 1961. Technical References Number 1. Cunningham, Francis S. “ The Use of Price Indexes in Escalator Contracts,” Monthly Labor Review, August 1963, pp. 948-952. Reprint No. 2424. Discusses the techniques of escalation using the two major price indexes published by the Bureau of Labor Statistics— the Consumer Price Index (CPI) and the Wholesale Price Index (W PI). Examines the basic elements of an escalator clause and procedures for carrying out the agreement. 2. Daugherty, James C. “Health Insurance in the Revised CPI,” Monthly Labor Review, November 1964, pp. 1299-1300. Explains and justifies the major change in the treatment of the health insurance component of medical care as initiated in the recent revision of the Consumer Price Index. Compares the former method of pricing actual premium rates with the new method of pricing the benefits received for hospital and professional services combined with a measurement for retained earnings. 3. Hoover, Ethel D. “ The CPI and Problems of Quality Change,” Monthly Labor Review, November 1961, pp. 1175-1185. Explains and illustrates problems of quality measurement met in the index calculation procedures. Defines quality as used by the BLS, specification pricing, direct price compari sons, and linking procedures. Concludes that there is no evidence to support the argument that the index is not a true measure of price change because of not fully eliminating the effect of quality changes. 4. Humes, Helen and Schiro, Bruno. “ The Rent Component of the Consumers’ Price Index. Part I— Concept and Measurement,” Monthly Labor Review, December 1948, pp. 631-637. “ The Rent Component of the Consumers’ Price Index. Part II— Methodology of Measure ment,” Monthly Labor Review, January 1949, pp. 60-68. Combined in Reprint No. 1947. Part I discusses trie basic concepts underlying the rent index. Part II explains the methods of obtaining and calculating rental data, the “ new unit bias” which existed during World War II and the problem of comperfsating for depreciation of quality caused by aging. 5. Jaffe, Sidney A. “ The Statistical Structure of the Revised CPI,” Monthly Labor Review, August 1964, pp. 916-924. Describes the concept and formulation, population and expenditure coverage, statistical techniques and problems of the revised index. Examines some operational aspects, especially sample replication. Presents the index formula in general, simplified, and in operational form. 6. -------. The Consumer Price Index— Technical Questions and Practical Answers. Paper presented before the American Statistical Association Meeting, Washington, D .C ., December 30, 1959 (Mimeographed by the U.S. Department of Labor, Bureau of Labor Statistics). 27 pp. General explanation and definition of the index concepts, coverage and calculation. Dis cussion of problem areas including sampling, seasonality, and alleged quality bias. CONSUMER PRICES Technical References—Continued ffumber 7. Lamale, Helen Humes. “Housing Costs in the Consumer Price Index, Part I ,” Monthly Labor Review, February 1956, pp. 189-196. “ Housing Costs in the Consumer Price Index. Part II, Monthly Labor Review, April 1956, pp. 442-446. Combined in Reprint No. 2188. Part I defines the housing component of the index and describes the derivation of expendi ture weights used in the calculation of the shelter index. Part II describes the procedures used to measure changes in the prices of the various items of shelter cost. 8. Larsgaard, Olga A. and Mack, Louise J. “ Compact Cars in the Consumer Price Index,” Monthly Labor Review, May 1961, pp. 519-523. Reprint No. 2368. Summarizes and explains the methodology used to link compact cars into the Consumer Price Index in 1961. Discusses the historical treatment of quality changes in standard size cars. 9. National Bureau of Economic Research. The Price Statistics of the Federal Government: Revieiv, Appraisal, and Recommendations, (Washington, National Bureau of Economic Research, General Series, Number 73, 1961), 496 pp. Also appears in Government Price Statistics: Hearings, Subcommittee on Economic Statistics of the Joint Economic Committee, 87th Cong., 1st sess., Part 1, January 24, 1961, ([Washington,] U.S. Government Printing Office, 1961), 526 pp. Report of the detailed investigation by the Price Statistics Review Committee of the NBER in 1959 of the main price indexes compiled by the Federal Government: The Consumer Price Index; the Wholesale Price Index; and the Indexes of Prices Received and Paid by Farmers. Reviews and analyzes the various aspects of the indexes and presents general and specific recom mendations for improvements. Twelve staff reports appended. 10. Rothwell, Doris P. “ Calculation of Average Retail Food Prices,” Monthly Labor Review, January 1965, pp. 61-66. Explains the BLS methods of collecting prices, and computing indexes and avreage prices for food items in the index. Emphasizes the unsuitable nature of index data for comparison of prices between cities. Presents estimated retail prices of food from December 1963 through November 1964, the-'cities covered, and the pricing diagram for food in the index. 11. -------. The Consumer Price Index: Pricing and Calculation Procedures, (Unnumbered paper, U.S. Department of Labor, Bureau of Labor Statistics, March 1964), 22 pp. Discusses sampling, pricing by specification; price collection, processing, and editing. Describes index formula, calculation of price relatives, expenditure weights, indexes, aggregation, and correction policy. 12. -------. “ Use of Varying Seasonal Weights in Price Index Construction,” Journal of the American Statistical Association, March 1958, pp. 66-77. Describes a formula based on varying seasonal weights for month-to-month measurements of price change which does not exhibit the “ biases” of chain indexes and which satisfies classical index theory with respect to year-to-year comparisons. Results of experimentation with alternative formulas are presented. 13. U.S. Congress, House of Representatives, Committee on Education and Labor. Consumers’ Price Index, Special subcommittee of the Committee on Education and Labor, 82nd Cong., 1st sess., Report No. 2 (1951), 39 pp. Nontechnical summary of results of hearings on the reliability of the Consumer Price Index. Presents details of history, uses, and method of construction of the index. Recommends continued support of the index by the Congress. 14. U.S. Congress, Joint Economic Committee. Government Price Statistics: Hearings, Subcommittee on Economic Statistics, 87th Cong., 1st sess., Part I, January 24, 1961 (Washington, U.S. Gov ernment Printing Office, 1961), 526 pp. Part 2, May 1, 2, 3, 4, and 5, 1961. (1961), 265 pp. Part 1 presents findings of an investigation by the Price Statistics Review Committee of the National Bureau of Economic Research in 1959-60 of all government price statistics. Also includes 12 staff papers on specific subjects. The detailed technical report includes recommenda tions for improvement of all indexes and, specifically for the Consumer Price Index, suggests extended coverage to include single consumers, probability sampling techniques, establishment of a research division, and regularly scheduled weight revisions. Part 2 presents testimony before the subcommittee of members of the Price Statistics Review Committee, government officials, and other interested parties concerning the committee report and recommendations. 93 BLS HANDBOOK OF METHODS 94 Technical References—Continued Number 15. U.S. Congress, Joint Committee on the Economic Report. The Consumers' Price Index— Report of the Joint Committee on the Economic Report on the Consumers' Price Index of the United States Bureau of Labor Statistics, 80th Cong., 2nd sess. (1949), 20 pp. Brief statement of the results of examinations of methodology, compilation, composition, and presentation of the Consumer Price Index as of 1949. Extensive bibliography. 16. U.S. Department of Labor, Bureau of Labor Statistics. The Consumer Price Index (Revised January 1 9 6 4 ): A Short Description, (U.S. Department of Labor, Bureau of Labor Statistics, September 1964). A nontechnical description of the index, its scope and computation. Explains the market basket, formula, uses and limitations of the index. Tables show cities included, population weights, pricing schedules, groups of goods and services priced, their relative importance, and the number of items priced as of December 1963. 17. ------- . Seasonal Factors, Consumer Price Index: Selected Series, June 1958-M ay 1961, Bulletin 1366 (1963), 47 pp. Provides basic data with which Consumer Price Index old series indexes can be adjusted for seasonal variation. Users are cautioned that the 1964 revision may have a very different effect on the series. Includes a description of the BLS method of computing seasonal factors, a discussion of its application to consumer price series, comments on specific series and tables providing indexes and seasonal factors for 66 selected series through May 1961. 18. ------- . “ Taxes and the Consumers’ Price Index,” Monthly Labor Review, January 1953, pp. 53-57. Reprint No. 2090. Discusses the present treatment of taxes in the index and the specific taxes included. Jus tifies the BLS policy of continuing to exclude income taxes from the index and including sales and excise taxes. 19. -------. Interim Adjustment of Consumers' Price Index: Correction of Neiv Unit Bias in Rent Component of Consumers’ Price Index and Relative Importance of Items, Bulletin 1039 (1952), 49 pp. Military developments in Korea in 1950 emphasized and made urgent the need for reweight ing of certain segments of the index before the already initiated revision could be completed in 1952. The failure to reflect the difference between rents for new dwellings when they first enter the market and comparable dwellings already on the market during and after the Second World War is discussed and the method of adjustment presented. Tabulation of adjusted indexes from 1940-50, relative importances and weights generated by the interim adjustment also are presented. 20. ------- . Consumer Prices in the United States, 1953-58: Price Trends and Indexes, Bulletin 1256, (1959), 126 pp. Analyzes and explains retail price trends and their effect on the economy from 1953 to 1958. Brief history* of the index and comparison of features of the old index based on the 1934-36 expenditures survey, with the adjusted index based on the 1947-49 expenditures survey in 7 cities and the revised index based on the 1950 expenditures survey. Explains the 1952 revision in detail. Presents historical indexes for various segments of the index. 21. U.S. Office of Economic Stabilization. Report of the President’s Committee on the Cost of Living, (1945), 423 pp. Summarizes the findings of the investigation in 1943-44 of the suitability of the Consumer Price Index for measurement of the change in the cost of living during wartime. Includes detailed discussions of the definition, scope, and statistical methodology of the index. 22. Wilkerson, Marvin. Measurement of Sampling Error in the Consumer Price Index: First Results. Paper presented at the American Statistical Association meeting, Washington, D.C., December 1964 (mimeographed by the U.S. Department of Labor, Bureau of Labor Statistics), 16 pp. With the computation of the Revised Consumer Price Index, completed in 1964, a first attempt was made to produce estimates of error for a comprehensive national price index. Pre sents a brief description of the replication design, estimates of error through October 1964. Dis cusses limitations of the error estimates and provides an interpretation of the results. 95 CONSUMER PRICES Technical References—Continued Number 23. Wilkerson, Marvin. Sampling Aspects of the Revised C P I (Unnumbered paper, U.S. Depart ment of Labor, Bureau of Labor Statistics, October 1964), 33 pp. Explains two significant improvements in methodology made in the Consumer Price Index at the time of the most recent revision: (1) application of probability sampling, and (2) estima tion of sampling error through a system of replicated samples. Tables present the sampling frame for selection of the index item sample and pairing of index cities for replication computa tions. 24. -------. “ The Revised City Sample for the Consumer Price Index,” Monthly Labor Review, October 1960, pp. 1078-1083. Reprint No. 2352. Describes the selection procedures used to derive the core sample of 50 cities used in the revised Consumer Price Index computation. Tables show probability patterns for the selected areas and basic and alternate city samples for the Consumer Price Index and the Consumer Expenditures Survey. — D o r is P. R othw ell and C arlyle P. S t a l l in g s Office of Prices and Living Conditions Chapter 11. Wholesale Prices Background The Wholesale Price Index (W PI) is the oldest continuous statistical series published by the Bureau of Labor Statistics (BLS) and one of the oldest in the Federal Government. It was first published in 1902, and covered the years 1890-1901. The origins of the index are as sociated with a resolution of the U.S. Senate in 1891, which authorized the Senate Com mittee on Finance to investigate the effects of the tariff laws “ upon the imports and exports, the growth, development, production, and prices of agricultural and manufactured articles at home and abroad.” 1 The index published in 1902 on the base 1890-99 was an unweighted average of price relatives and included from 250 to 261 com modities. Since that time, many changes have been made in the sample of commodities, the base period, and in the method of calculating the index. The first major change was com pleted at the end of 1914, when a system of weighting was introduced and the index was recalculated back through 1890.1 By 1940, the 2 number of commodities had increased to ap proximately 900, based on about 2,000 individual price quotations. Then, in 1952, the most ex tensive revision in the history of the index was completed.3 The number of commodities and quotations was doubled, weights were based on 1947 Censuses and changes were made in the calculation method. Some changes in classifica 1 W h olesa le P rices, W a ges, and T ransportation, Senate Re port No. 1394, "T he Aldrich Report,” Senate Committee on Finance, Congress of the United States, March 3, 1893, Part I, (52d Cong., 2d session), Government Printing Office (1893); and C ou rse o f W h olesa le P rices, 1890-1901, Bulletin of the Department of Labor, No. 39, March 1902, pp. 205-209. 2 See also, Allan D. Searle, “Weight Revisions in the Whole sale Price Index, 1890-1960,” Monthly Labor Review, February 1962, pp. 175-182. 3 A large number of the newly introduced commodity prices were carried back to 1947. The presently published index contains the new commodities for the period 1947-51 and displaces the older, less comprehensive index on the 1926 base published for the same period. tion were made also, including expansion to the present 15 major groups. A major reclassifica tion was implemented in January 1967, when the 8-digit classification structure was initiated. Also at that time, new weights from the 1963 industrial censuses were introduced. By January 1971, the number of commodities had increased to more than 2,500, the number of price quotations had increased to over 8,000, and the index had become increasingly repre sentative of general primary market price changes. Description o f Survey Concepts Throughout its history, the WPI has* been a measure of price changes for goods sold in primary markets in the United States. “ Whole sale” as used in the title of the index refers to sales in large quantities, not prices received by wholesalers, jobbers, or distributors. From its inception, the index has been con sidered a general purpose index designed to measure the general price level in other than retail markets. From the beginning of the in dex, however, attention was directed to some specific needs of users, and indexes for indi vidual commodities and for major commodity groups were published. As early as 1903, two special group indexes by stage of processing— Raw Commodities and Manufactured Com modities— were published “ to meet the wishes of students of price statistics.” In recent years, emphasis has been placed on the development of more subdivisions within major groups and special combinations of indexes such as by stage of processing and by durability of product. Most of the quotations reported to the Bu reau are the selling prices of representative manufacturers or producers, but some prices 97 98 BLS HANDBOOK OF METHODS are those quoted on organized exchanges (spot prices) or at central markets. Prices for im ported commodities are those received byimporters— the first commercial transaction involving the commodity in the United States. Since the index is intended to measure “ pure” price change, that is, not influenced by changes in quality, quantity, shipping terms, product mix, etc., commodities included in the index are defined by precise specifications which in corporate their principal price-determining characteristics.* So far as possible, prices are 4 3 2 f.o.b. production point, and refer to sales for immediate delivery. Prices applicable to longrun contracts and “ futures” are usually not included. Universe The WPI universe consists of all commodi ties sold in commercial transactions in primary markets of the United States, including Alaska and Hawaii. Commodities produced in the United States are included, as well as those imported for sale. The universe covers manu factured and processed goods and the output of industries classified as manufacturing, agricul ture, forestry, fishing, mining, gas and elec tricity, public utilities, and goods competitive with those made in the producing sector, such as waste and scrap materials. All systematic production is represented, but individually priced items, such as works of art, are excluded. Also excluded are goods transferred between establishments owned by the same company (interplant or intra-company transfers). Goods sold at retail by producer-owned retail estab lishments also are excluded because they conceptually belong to a retail (consumers') universe, rather than to primary market transactions. Civilian goods normally purchased by the Government are in the universe, but military goods are not. Government sales of some com 4 An example of a commodity specification for steel strip is: "Strip, cold-rolled, carbon steel, coils, No. 4 temper, No. 2 finish, No. 3 edge, base chemistry, 6 " x .050", in quantities of 10,000 to 19,999 lb., mill to user, f.o.b. mill, per 100 lb.” 3 The prices used in the index through 1951 were the simple arithmetic averages of prices for all Tuesdays in the month. From January 1952 through December 1966, Tuesday of the week containing the 15th was the pricing date. modifies (e.g., electric power) are included if they can be considered competitive with free market sales. Prices To the extent possible, the prices used in constructing the index are those that apply to the first significant commercial transaction in the United States. Transactions for the same item at later stages of distribution are not included. However, as raw materials are trans formed into semifinished and finished goods, the resulting products are represented. With some exceptions, the prices refer to one particular day of each month. In most cases, the pricing date is Tuesday of the week con taining the 13th day; but for some commodities (farm products, particularly) a day other than Tuesday is used because it is considered more representative.5 The Bureau attempts to base the WPI on actual transaction prices. Companies are re quested to report prices less all discounts, al lowances, rebates, free deals, etc., so that the resulting net price is the actual selling price of the commodity for the specified basis of quo tation. The Bureau periodically emphasizes to reporters the need to take into account all dis counts and allowances. However, list or book prices are used if transaction prices are un obtainable. Prices are generally f.o.b. production or cen tral marketing point to avoid reflection of changes in transportation costs. Delivered prices are included only when the customary practice of the industry is to quote on this basis and the Bureau cannot obtain a price at the production point. Subsidies to the producer and excise taxes are excluded since they are not considered part of the price, but import duties are included as part of the selling price of im ported goods. Although the same commodity is priced gen erally month after month, it is necessary to provide a means for bridging over changes in detailed specifications (or descriptions of items priced) so that only real price change will be measured. An adjustment is particularly im portant when new commodities are introduced, WHOLESALE PRICES but even when specifications of existing com modities are changed, care is exercised to help insure that only price changes influence the index. A new price series resulting from a physical change in an article or a change in its selling terms is substituted for the earlier series by direct comparison or by linking. The objec tive of the linking procedure is to insure that the index will reflect only those changes due to actual price differences.6 Each time a change in the item priced occurs, the Bureau appraises the significance of the specification change to ascertain whether an actual price change oc curred. If the specification change is minor and does not involve price-making factors, the sub stitution is effected by direct comparison, and any reported price change between the old and the new specification is reflected in the index. If changes in specification are major, and if either no real price change occurred or no in formation can be obtained concerning the value of the difference in specification (perhaps in dicative of a change in quality), the substitu tion is made by linking and no change is reflected in the index. In this case, any reported 6 The following example illustrates the linking procedure: The September price for a certain machine used in the cal culation of the index was $2,347.50. In October, a new model of the machine was introduced, priced at $2,562.60. The new model was considered essentially comparable with the old, except that it had a more powerful motor and larger tires. These were valued at $186.20 more than the value of those used on the former model. For linking, the September price of the new model was estimated at $2,533.70 ($2,347.50 Sep tember price of former model plus $186.20 increase in value of motor and tires). The price comparison between Septem ber and October was based on the estimated September price of $2,533.70 and the reported October price of $2,562.60. Thus a 1.1-percent increase was reflected in the October index, but the price change due to quality improvement (more powerful motor and larger tires) was not reflected. 99 difference in price level is not permitted to affect the index level. When differences are major, an attempt is made to obtain data from the reporters on the value of the additional (or deleted) features and to adjust the price index accordingly. This is particularly important in the case of some durable goods, such as automobiles, which have periodic model changes. Also, price increases which result from the addition of features that formerly sold at extra cost are not reflected in the index. Conversely, price changes attrib utable to deletion of equipment which was formerly standard are not treated as decreases. In the event production of a specified com modity is discontinued by a reporter, or its im portance is reduced, the Bureau collects price data for a similar or a replacement item. Prices are obtained for the new and the discontinued series for a 1-month overlap period. The index is extended by linking, and the difference, if any, between the new item price and the orig inal price is taken as a measure of the quality difference between the two items. Linking is also used for the addition to or deletion of commodities or groups of commodi ties from the index; the addition to or deletion of a company report from the sample of com panies priced, or, on occasion, a change in the source of price. Whenever a new commodity is added to an existing commodity group, link ing of the new item to any one of the existing items is not pertinent. Instead, the weights of the entire group are redistributed to include the new item and the link is made at the group level instead of at the commodity level. A similar procedure is used to handle items that drop out of the index. BLS HANDBOOK OF METHODS 100 W h o l e s a l e P r ice I n d e x R e l a t iv e I m p o r t a n c e , N u m b e r o f I t e m s a n d P r ic e Q u o t a t i o n s for M a jo r G r o u p s a n d S u b gr o ups Grouping Relative importance in total 1963 weights December 1970 All commodities ________________________________ 01 Farm products _____________________________________ 01-1 Fresh and dried fruits and vegetables_________ 01-2 Grains ____________________________________________ 01-3 Livestock _________________________________________ 01-4 Live poultry _____________________________________ 01-5 Plant and animal fibers ________________________ 01-6 Fluid milk _______________________________________ 01-7 Eggs ---------------------------------------------------------------------01-8 Hay, hayseeds, and o ilseed s_____________________ 019 Other farm products _________________________ 02 Processed foods and f e e d s ________________________ 021 Cereal and bakery products ________________ 02-2 Meats, poultry, and fish ________________________ 02-3 Dairy products ___________________________________ 02-4 Processed fruits and vegetables _______________ 02-5 Sugar and confectionery________________________ 02-6 Beverages and beverage m aterials_____________ 02-7 Fats and oils 1 ____________________________________ 02-8 Miscellaneous processed foods _________________ 029 Manufactured animal f e e d s __________________ 100.000 10.038 1.176 1.198 2.851 .255 .437 2.081 .510 .694 .835 16.298 2.017 4.153 2.300 .866 1.268 2.118 .627 1.185 1.765 03 Textile products and a p p a r e l_____________________ 031 Cotton products _______________________________ 03-2 Wool products ___ ______________________________ 03-3 Manmade fiber textile p rodu cts________________ 03-4 Silk' products1 ___________________________________ 03-5 Apparel _________ _____________________ .________ 03-6 Textile housefumishings ________________________ 037 Miscellaneous textile products ______________ 6.875 1.086 .346 1.320 04 Hides, skins, leather, and related p rodu cts______ 04- 1 Hides and skins _______________________________ 04-2 Leather ___________________________________________ 04-3 Footwear _________________________________________ 044 Other leather and related produ cts_________ 1.239 .077 .179 .694 .289 05 Fuels and related products, and p o w e r__________ 051 Coal ___________________________________________ 05-2 Coke _ ___________________________________________ 05-3 Gas fuels _________________________________________ 05-4 Electric power ___________________________________ 05-6 Crude petroleum and related products1 _______ 057 Petroleum products, refined _________________ 7.322 .700 .092 .677 1.767 .628 3.459 06 Chemicals and allied products ____________________ 061 Industrial chemicals ____ _____________________ 06-2 Paint and paint materials 1 ______________________ 06-3 Drugs and pharmaceuticals _____________________ 06-4 Fats and oils, inedible __________________________ 06-5 Agricultural chemicals and chemical products 06-6 Plastic resins and materials ____________________ 067 Other chemicals and allied products _______ 07 Rubber and plastic products _____________________ 071 Rubber and rubber products _______________ 072 Plastic products ______________________________ 08 Lumber and wood p rodu cts_______________________ 08- 1 Lumber ________________________________________ 08-2 Mi 1 work _________________________________________ 1 08-3 Plywood __________________________________________ 084 Other wood products ________________________ 09 Pulp, paper, and allied p rodu cts_________________ 09- 1 Pulp, paper, and products, excluding building paper and board ______________________________ 092 Building paper and b o a r d ____________________ 5.937 1.814 .698 .807 .189 .545 .366 1.517 3.631 .359 .132 December 1966 Number of items and price quotations, January 1971 Price quotations Items 7 9 193 22 40 15 35 11 18 17 20 15 8,017 143 48 8 12 4 29 8 4 13 17 664 79 99 42 172 23 104 24 67 54 183 38 17 45 519 93 39 108 67 11 5 235 32 12 49 12 11 17 9 100 12 12 55 21 67 7 6 2 18 6 28 275 39 6 2 173 22 33 335 97 30 107 7 43 10 41 550 101 71 167 7 43 38 123 335 163 172 367 224 68 41 34 266 100.000 10.637 1.171 1.357 3.086 .332 .553 2.001 .576 .780 .781 16.533 2.038 4.404 2.275 .856 1.192 2.047 .603 1.183 1.935 7.149 1.152 .403 1.488 .021 3.562 .384 .139 1.264 .097 .196 .667 .304 2,503 101 30 8 12 4 19 7.130 .439 .070 .691 1.808 .614 3.508 6.378 1.968 .706 .888 .163 .675 .456 1.522 8 4 ! i | 2.346 1.475 .872 2.483 1.259 .686 .399 .140 4.796 2.339 2.339 2.418 1.215 .658 .416 .129 4.877 88 45 43 77 50 15 7 5 74 4.655 .141 4.719 .158 66 8 222 44 10 Metals and metal p rodu cts________________________ 101 Iron and steel _________________________________ 10-2 Nonferrous metals ______________________________ 10-3 Metal containers ________________________________ 10-4 Hardware ________________________________________ 10-5 Plumbing fixtures and brass fittin gs___________ 10-6 Heating equipment _____________________________ 10-7 Fabricated structural metal p ro d u cts__________ 108 Miscellaneous metal products _______________ 13.456 4.797 3.376 .488 .575 .181 .256 1.771 2.012 12.799 4.547 3.205 .462 .548 .177 .254 1.716 1.890 375 108 92 8 44 13 21 41 48 1,295 404 275 28 144 65 90 133 156 11 Machinery and equipment ________________________ 111 Agricultural machinery and equipment ____ 11-2 Construction machinery and equ ip m en t_______ 11-3 Metalworking machinery and equipm ent______ 11-4 General purpose machinery and equipment ___ 11-6 Special industry machinery and equ ip m en t_ _ 11-7 Electrical machinery and equ ip m en t__________ 11- 9 Miscellaneous machinery ____________________ 12.508 .705 .885 1.545 1.770 1.641 4.348 1.614 627 55 49 105 78 76 182 82 2,098 202 169 319 287 226 633 262 12 Furniture and household d u rables_______________ 12- 1 Household furniture _________________^_______ 12-2 Commercial furniture ___________________________ 3.529 .925 .449 12.110 .665 .807 1.469 1.665 1.506 4.462 1.536 3.584 .904 .416 93 21 7 498 151 34 See footnotes at end of table. WHOLESALE PRICES 101 W h o l e s a l e P r ic e I n d e x — C o n t i n u e d R e l a t iv e I m p o r t a n c e , N u m b e r o f I t e m s a n d P r ic e Q u o t a t i o n s for M a jo r G r o u p s a n d S ubgro ups Relative importance in total 1963 weights Grouping December 1970 December 1966 Number of items and price quotations, January 1971 Price Quotations Items .336 .862 .383 .574 .385 .891 .459 .529 9 28 10 18 54 134 52 73 13 Nonmetallic mineral Droducts 13-1 Glass 1 13-2 Concrete ingredients 13-3 Concrete products 13-4 Structural clay prods., excluding refractories 13-5 Refractories 13-6 Asphalt roofing 13-7 Gypsum products ............ ... 13-8 Glass containers 13-9 Other nonmetallic minerals 3.175 .388 .630 .922 .170 .159 .120 .090 .305 .392 3.040 .364 .612 .882 .168 .139 .127 .106 .273 .369 45 4 4 4 6 5 5 3 7 7 405 14 14 TransDortation eauiDment 14-1 Motor vehicles and equipment 14-2 Aircraft 7 14-4 Railroad eauiDment 7.441 6.642 .490 .310 2.556 .515 .863 .101 .371 .705 7.244 6.932 .312 77 62 10 5 152 132 10 10 2.498 .513 .802 .102 .386 .695 119 30 9 5 40 35 12-3 12-4 12-5 12-6 Floor coverings Household appliances Home electronic equipment Other household durable goods 15 Miscellaneous Droducts 15-1 Toys, sporting goods, small arms, etc. 15-2 Tobacco products 15-3 Notions 15-4 Photographic equipment and supplies 15-9 Other miscellaneous products "’ __ 1 Subgroup index not published. NOTE: Relative importance represents the basic value weight of an item or items multiplied by the relative of price change between the weight date and a later date, and Prices for individual commodities reported by the individual companies are averaged (usually by means of an unweighted average). Month-to-month price change should be com puted from matched-company data. In order that a change in the company-reporter sample itself not affect the measure of percent change, the change is calculated for any 2 months from identical-company data. Thus, a new report affects the index no earlier than the second month. Glassification The classification system of the WPI follows commodity lines. Products are grouped by sim ilarity of end-use or material composition, rather than by industry of origin. The WPI classification does not match the Standard In dustrial Classification (SIC), the Standard Commodity Classification, the United Nations Standard International Trade Classification (SITC), or any other standard classification. Historical continuity and the needs of index users have been important in developing the classification. No single classification plan can 169 68 49 5 18 9 26 47 350 103 48 10 69 120 the result is expressed as a percentage of the total for all commodities. The differences between the relative impor tances as of December 1966, the date of last major weight change, and that of December 1970 are the result of pnce changes only. meet all of the requirements for wholesale price statistics, but the plan adopted should be flex ible enough to facilitate regrouping of price series to make special grouping indexes. In January 1971, the index was made up of 15 major groups, 87 subgroups, 291 product classes, 554 subproduct classes, and 2,503 items.7 To meet the needs of index users, a number of special group indexes are calculated and published each month. Among these are indexes by stage of processing,8 indexes by durability of product, and indexes of construction ma terials, in addition to about 22 other special group indexes. Except for the stage of processing indexes, these special groupings consist of rearrange ments of the WPI data into different combina tions of price series, so that the appropriate prices and weights are those of the WPI. The 7 See table for the major groups and subgroups included in the WPI. 8 The broad stages of processing are: Crude materials for further processing; Intermediate materials, supplies, and components; and Finished goods. Each of these is subdivided further. 102 BLS HANDBOOK OF METHODS stage of processing indexes, however, regroup each item priced in the WPI according to the amount of processing, manufacturing, or as sembling it undergoes before entering the market. A commodity may appear in several different categories in this scheme. Thus, 29 percent of the fresh vegetables (by valueweight) was assigned to crude foodstuffs and feedstuffs for further processing and 71 per cent to consumer foods (as “ finished” goods). The value weights are the same as those of the WPI and the allocations among the stages of processing are from an inter-industry trans action study made for the year 1958 by the Office of Business Economics. Data Sources and Collection Methods Prices Price data are collected by mail question naire, and reporting is voluntary and confi dential. Most prices are collected each month. For a few commodities, for which price changes are infrequent, the shuttle schedule is mailed quarterly, but monthly prices are re quested. Generally, the price data used in the index are obtained directly from the producing company, but some trade publications are used when the publication generally is accepted as reliable by the Bureau and the industry. For fish and most agricultural products, the prices used are those collected and published by other Government agencies. Price reporting is initiated, wherever possi ble, by a personal visit by a Bureau representa tive to the prospective respondent. Pricing of additional products from established reporters often is started by mail. In any event, a de tailed report describing all of the price-making characteristics of the commodity is prepared for each new price series. This commodity price information sheet (BLS 1810) is shown on pages 107 and 108. The form becomes a part of the permanent record for the series. After the initial collection of prices, monthly infor mation is collected by mail on a shuttle sched ule. (BLS 473, shown on pages 109 and 110.) Weights The price data are combined using weights based on value of shipments. The major sources of the value data are: Bureau of Census --Census of Manufactures Census of Mineral Industries Bureau of M in e s___ Various publications, e.g., Minerals Yearbook Department of Agriculture______ Various publications, e.g., Agricultural Statistics Bureau of F ish eries_________Various publications, e.g., Fisheries of the United States In addition, many other sources of data, such as trade associations, are used. Import data are obtained from a report of the U.S. Depart ment of Commerce, United States Imports for Consumption. Sampling The monthy index is based on a judgment sample of commodities, a sample of specifica tions (descriptions), and a sample of reporters. The sample of commodities is chosen after a review of the data of the industrial censuses and other statistics of value of transactions. Generally, the commodities chosen are those of the largest shipment values. Starting with January 1967, expansion of Industry Sector Price Index sample coverage has been a major influence in selecting new products for the WPI. New items are not added until they have become established in the market.8 They are 9 added, normally, in December of any year, and have their first effect on the index in January. Samples of specifications and of reporters are selected after consultation with trade as sociations or other industry representatives and with staff of other government agencies. Individual commodity specifications are se lected also on the basis of net dollar sales. That is, the “ volume seller” of the industry (not of 8 If new items are added before they become fully estab lished, the sharp price decline experienced by most products, as they move from development to mass production, imparts a downward bias to the index. Also, many new products turn out to be of only transitory significance. 103 WHOLESALE PRICES the company) is preferred. The specification describes not only the popular physical charac teristics but also the most common quality, grade, level of distribution, and market. How ever, terms of sales (discounts, etc.) are based on the company’s own most common practice. For some commodities, prices are quoted by producers and sellers in terms of a single spe cification taken as standard; all other prices are quoted as differentials from the standard. The latter is true for some farm products such as wheat and cotton. When no standard com modity basis exists, the specification to be priced is selected with the help of industry experts. The number of reporters is determined, to some extent, by the variation of price move ments among them and the degree of price leadership. Whenever possible, a minimum of three companies is obtained, so that data for specified commodities can be published without disclosure of information supplied by individ ual companies. For commodities with more than one major production area and a definite regional pattern, a larger sample is selected. Among these commodities are waste materials and building materials such as brick, cement, and stone. A comparatively small list of properly se lected commodities would produce a reliable index, if only an All Commodities index were desired. However, historically interest has been great in indexes for groups of commodities and for individual commodities. To meet these needs, the Bureau has increased the sample in order to provide more detailed indexes as well as many special-purpose indexes. An alternative formulation more closely ap proximates the actual computation procedure: 2. I, = [2 (Q.P„) (P ,/P „) / 2 Qap „] x 10°- In this form, the index is a weighted average of price relatives for each item (P i/P G The ). expression (QaP0) represents the weights in value form and the “ P” and “ Q” elements (both of which originally relate to period “ a” but are adjusted for price change to period “ o” ) are not derived separately. Each value weight in cludes not only the value of items priced but also the values of unpriced items whose price movements are assumed to behave similarly. When new weights are introduced, the index with new weights is linked to the index con structed with the earlier weights. The weight adjustment itself, therefore, affects only the later calculations of average price change. When specifications or samples change, the item relatives must be computed by linking (multiplying) the relatives for the separate periods for which the data are precisely com parable. (For a somewhat more detailed treat ment, see chapter 10, Consumer Prices.) Base Period The Wholesale Price Index has been com puted on the government-wide standard refer ence base 1967=100 since January 1971.1 It 0 had been based 1957-59=100 from January 1962 through December 1970. Earlier bases were 1947-49, 1926, and 1913. New items (or new index groupings consisting primarily of new items) introduced into the index after 1967 cannot be calculated on the 1967 base. Such indexes are published with separate bases related to the date of introduction. Estimating Procedures Weights Formula and Calculation In concept, the Wholesale Price Index is cal culated according to a modified Laspeyres formula: 1. I, = [2 QaP ,/2 Q ,P 0] x 100> where P0 is the price of a commodity in the comparison period and P, is its price currently. Q, represents the quantity shipped during the weight-base period. The WPI weights represent the total net sell ing value of commodities produced, processed, 1 Conversion of indexes from the 1967=100 base to the 0 1957-59 base may be accomplished by multiplying the 1967 = 100 based index by that item’s 1967 annual (12-month) average (1957-59=100) and dividing the result by 100. Aver ages for 1967, calculated from 12 months’ data, and appear ing in the December 1968 issue of “ Wholesale Prices and Price Indexes” may be used for base conversion. 104 BLS HANDBOOK OF METHODS or imported in this country, including Alaska and Hawaii, and flowing into primary markets. The values are f.o.b. production point and are exclusive of excise taxes. The value of inter plant transfers, military products, and goods sold at retail directly from producing establish ments also are excluded. Thus the definition of the weights conforms to tho universe definition. Each commodity price series is considered representative of a class of prices and is as signed its own weight (the shipment value of the commodity) plus the weights of other re lated commodities not directly priced but whose prices are known or assumed to move sim ilarly.1 The assignment of price movements 1 for priced commodities to those for which quotations are not obtained is referred to as imputation. For some commodities— such as ships and some kinds of custom-made ma chinery— it is not possible to obtain direct measures of price movement. The weights for such items are assigned to other commodities or groups of commodities for which prices are available. Usually, this assignment is made to priced commodities that have a similar manu facturing process, on the assumption of similar price movements. Price movements for attach ments and parts for certain machinery often are imputed to the machine itself. The Bureau’s policy is to revise the WPI weights periodically when data from the in dustrial censuses become available.1 The 1 2 weights beginning in 1967 are based on the 1963 industrial censuses. The next revision which normally would follow the 1967 indus trial censuses has been postponed because of lack of resources. Indexes for 1947 through 1954 are based primarily on the 1947 censuses. In the January 1955 index, adjustments were made to align the major group weight totals with 1952-53 average shipment values as re ported in the Annual Surveys of Manufactures. Weights based on the 1954 census shipment values were introduced in January 1958. From 1961 through 1966, weights were based on 1958 census values. In January 1967, new weights from the 1963 industrial censuses were in corporated with the comprehensive reclassifica tion mentioned previously. Subsequent minor redistributions of the weights have been made each January to account for additions and deletions of commodities. The Bureau publishes the relative impor tance of each item in the WPI rather than the actual values used as weights. The relative importance of an item represents its basic value weight used in the index, including impu tations, multiplied by the relative of price change from the weight date to a later date; the result is expressed as a percent of the total for all commodities or for some index grouping.1 3 Imputing Missing Prices Whenever price data are not available for a particular month, it is necessary to estimate the missing price for use in the calculation of the index. For commodities in the farm prod ucts and processed foods groups, out of the market seasonally, the price in off-season is imputed from the combined movement of the related commodities for which prices are avail able for the two periods being compared. For other commodities, delinquent prices are held unchanged from the preceding month. Prices for some custom-made items are re ported to BLS as estimates. For example, prices for fabricated structural steel for build ings and bridges are obtained from producers who reprice, each month, steel of the same specifications as used in structures on which they had been engaged at the time pricing for the WPI was initiated. Elevators, normally sold including installation, are reported f.o.b. plant— i.e., excluding transportation and in stallation cost— for use in the WPI. Analysis and Presentation The monthly WPI is published first in a press release, usually issued in the first week of the 1 Before the 1952 revision (calculated back to 1947), priced 1 items in the index were weighted only by their own market values. 1 In general, the censuses are collected at 5-year intervals. 2 1 The use of relative importance data to construct indexes 3 for groups of products is discussed in the January 1970 issue of “ Wholesale Prices and Price Indexes.” Relative importances as of December for all WPI items are published in each January issue. WHOLESALE PRICES month following the reference month. Indexes are shown for all groups and subgroups as well as for All Commodities, Farm Products and Processed Foods and Feeds combined, and In dustrial Commodities. Analytic tables also are included which show monthly percent changes for the preceding 12 months for major group ings, and selected seasonally adjusted and un adjusted changes for some stage of processing classifications. A brief description and analysis of the causes of price movements are included. The monthly detailed report, issued some time after the press release, carries all data for which wholesale price indexes are published, including item indexes and all special group indexes. Prices for many individual commodi ties also are included. Each quarter this report includes a more comprehensive analysis than that given in the press release. Annual sum maries appear in the monthly report as they become available. In addition, numerous his torical tabulations at various levels of detail are available on request. The monthly indexes are published as final. Beginning with data for January 1967, only major corrections are made and published im mediately at the time the error is discovered. Each year, after calculation of the December index, all corrections reported during the year are made, and the indexes for all 12 months and the annual average are republished as revised. Selected seasonally adjusted indexes or per cent changes are published in the press release and monthly detailed report. About 50 indexes which historically show significant and con sistent seasonal movement are presented each month seasonally adjusted and unadjusted. The applicable season adjustment factors are avail able on request from the Bureau. These factors are recalculated annually to include more recent data, and the most recent set of factors may differ somewhat from those previously in use. Uses and Limitations The WPI is used by government and private research agencies for many purposes, including 105 market analysis, escalation of long-term pur chase and sales contracts, and formulation of monetary policies. It is used, as well, as an indicator of economic trends. A 1961 survey of users of the WPI revealed that more than one-half use the All Commodi ties index as a general economic indicator. About 40 percent use that index or its com ponents to compare with their selling or buying prices. The survey revealed that over 10 billion dollars (in terms of unexpired value) in long term contracts for purchase of material or lease of industrial property are escalated ac cording to changes in the total index or its components. Government agencies and private research groups also use the component series in deflating value data in preparation of the gross national product estimates and in studies of economic growth. The index also is used by buyers and sellers of commodities— purchasing agents and sales managers. In most of these cases, it is not the All Commodities index, but rather the group indexes and the individual price series that are employed. Buyers of commodities are able to check both the amounts which they pay for goods and the general movement of their pur chase prices against the index. The use of the index for checking absolute price levels is limited substantially, however. The Bureau’s main goal has been to measure the direction and amount of change, and only incidentally to measure actual selling prices. The index, as a measure of general and speci fic price trends, also is used widely in budget making and review, both in government and in industry; in planning the cost of plant expan sion programs; in appraising inventories; in establishing replacement costs; etc. Compo nents of the index also are used in LIFO (Last-In, First-Out) inventory accounting by some organizations. Although the WPI often is used to measure change in purchasing power of the dollar, it should not be used to measure changes in gen eral purchasing power, prices at retail, securi ties prices, etc. Comparisons between the level of the WPI, the Consumer Price Index, and the indexes of prices of farm products show relative change from a base period, but com- 106 BLS HANDBOOK OF METHODS parisons of the index levels should not be used as a measure of the actual margins between farm prices and manufacturing or between manufacturing and retail. Its commodity clas sification structure should be borne in mind when using it to measure price changes for industries, many of which make divers prod ucts not classified as their “ primary” products.1 4 Again, as in other measures, the WPI has some limitations even in the field for which it is conceptually designed. Segments of the in dex are used as deflators of gross national product data, but gaps in WPI coverage leave considerable areas for which deflators have not been provided. The WPI is based on a purposive, judgment sample. The All Commodities Index can be as sumed to be more reliable than a component group index, in general. Also, it can be assumed that the reliability of the index has increased over time as the sample has expanded.1 As 5 the economy has produced an increasing pro 1 See Chapter 12 on Industry-Sector Indexes. 4 1 The sample of priced items doubled in 1952 to about 1,850 5 items and has increased to about 2,500 since then. portion of fabricated finished goods (whose price changes are relatively infrequent), over the years, movement of the WPI has become somewhat smoother. Currently, new products are added each year. In earlier decades, there were also major additions of large numbers of new items at one time, in commodity areas previously underrepresented. These sudden ex pansions could have made it appear that prices had stabilized suddenly. To the extent that quality improves (or de teriorates) over the years, the index errs when no adjustment is made. However, the Bureau makes suitable adjustments whenever possible. Assuming quality improvement, the index would have an upward bias if direct compari son were made between unimproved and im proved articles. If, on the other hand, such changes were consistently made by linking, a downward bias would result. Since the Bureau has not adopted either method exclusively, and in many instances tries to evaluate the changes brought to its attention, the bias that may exist is considered to be small. However, no measure of its magnitude is available. 107 WHOLESALE PRICES BLS 1810 THIS FORM WILL BE HELD IN CONFIDENCE B udget Bureau No. 44-R 0602 U.S. DEPARTMENT OF LABOR E xpiration date: 1 2 /3 1 /7 3 B U R E A U OF L A B O R S T A T I S T I C S Washington, D.C. 20212 Code No. Commodity COMMODITY PRICE INFORMATION SHEET Mfr. | | Plant or division Other (S p e e lly ) Address (Street) (C ity and State) Information authorized bv (Z ip co d e) Title Information furnished by Title Mail schedule to Title (R eporter) Address (C ity and State) (Street) 1. 2. COMMODITY DESCRIPTION (include style n o., model no., lot n o., grade, brand, etc) PRICE HISTORY FOR COMMODITY DESCRIBED ABOVE Price Date 3. (Z ip code) Remarks Date Price Remarks CHECK OR FILL-IN PERTINENT INFORMATION ABOUT PRICES REPO RTED ABOVE A. C la ss of seller B. Prices are: ________ _______ ________________________to C lass of customer __________________________________ (m lr., Importer, e tc .) Actual transaction prices’ Q L ist prices less discounts Q ; Other Q ; (w h o lesa ler, u se r, e tc .) L ist prices subject to discounts Q ; (s p e c ify )_____ ______________________________________________________ C. Unit quoted_____________________________________________ D. Size of order____________________________________________ E. Shipping terms (f. o. b .; frt. allowed; ect.) F. Type of package used; Crate Q ; Carton Q G. Is refund allowed for returnable container? ; Y es H ] Bag. Q No Q ; O th er_______________________ ._______________ If “ Y e s ” , e x p la in __________________________ A ctual s e llin g p r ic e s to c la s s o f custom er, for s iz e o f order, shipping term s, and d isc o u n ts reported. BLS HANDBOOK OF METHODS 108 4. ENTER DISCOUNTS AND ALLOWANCES A P P L IC A B L E TO R EPO RTED PRICES WHEN COMMODITY DESCRIBED IS SOLD TO THE CLASS OF CUSTOMER SPECIFIED IN 3A . A. Trade discount ________________________ % B. Quantity discount (based on size of order specified in 3D) __________________ % FOR THE FOLLOW ING, INDICATE DISCOUNT TERMS AND ESTIMATE THE P E R C E N T OF SALES AFFECTED Estimated % of sale s Terms C. Cash d is c o u n t ____________________________________________________________________________________ --------- ------------------------------------------------------------------------------------------------------- --------------------------% D. Seasonal discounts % E. Cumulative volume discount % F. Rebates (monthly) % G. Other discounts, allow ances, free d eals, etc. (explain fully) CIRCLE A L L DISCOUNTS OR ALLOWANCES ABOVE WHICH HAVE BEEN DEDU CTED IN ARRIVING A T PRICES R EPO RTED 5. LIST DUTIES OR EXCISE T A XE S A P P L IC A B L E TO R EPO RTED PRICES. A. B. 6. These are included in prices quoted Not included Q ] If tax is included, give example of how to calculate price excluding tax. ENTER APPROXIM ATE P ER CENTAGE OF SALES TO EACH CLASS OF CUSTOMER. O th e r m fr. ( O . E .M . or D ist r ib u t o r Jo b b e r W h o le s a le r S a la s (a p p ro x . % ) U ser O th e r T o ta l 100% Remarks: BLS Representative R e t a ile r ( s p e c if y ) a s s e m b le r ) Date WHOLESALE PRICES U S. DEPARTMENT OF LABOR B ureau of La b o r WAS HIN GT ON. O C S tatistics 202 12 INFORMATION FOR THE WHOLESALE PRICE INDEX ALL REPORTS WILL BE HELD IN CONFIDENCE Dear Sir: The price data which you provide is used in computing the Wholesale Price Index which is the officially accepted indicator of primary market price movements. The index is widely used by industry and government. These voluntary reports, submitted by you and other businessmen, are the major source of information used in preparing this index. The information you provide is strictly confidential and open to inspection only to sworn employees of the Bureau of Labor Statistics. Please use the enclosed envelope, which requires no postage, for returning this schedule. Your continued cooperation is greatly appreciated. COMMISSION EH OF LABOR STATISTICS IMPORTANT INSTRUCTIONS In the boxes provided on the other side, pleose be sure to indicote oil chonges in COMMODITY D E S C R IP T IO N , BASIS OF QUOTATION, DISCOUNTS, ALLO W AN CES. AND T A X E S that m ay hove occurred since your last report. Your cooperation in keep ing oil inform ation current is a g reat ard in com puting a r e lia b le , a c cu ra te W h o le sa le Price In d e x. (Remarks) 109 110 BLS HANDBOOK OF METHODS INFORMATION FOR THE WHOLESALE PRICE INDEX C o da No COM MODITY DESCRIPTION (Pleoia md.cote a l l changes ]• CHANGES C iv e dote, notwre. ond estim ated vo I im of chonge 2 RASIS OF Q U O TA TIO N (P h o t * *nd*cote a l l chonges > Dot* tn d noKirt of (K on g * U ni* C ln i of Mlhr and c w iH w S>ie af order Shipping termy Omar (Jpa«V ) 3. D IS C O U N T S , A L L O W A N C E S , AN D T A X E S In d ica ta a ll d is c o u n ts , a llo w o n c a s , and tan as a p p lic a b le to obova b a s is of q u o tatio n . T h is inform ation is naadad to a rriv e at the A C T U A L S E L L I N G P R I C E . (P le a s e i n d i c a t e a l l chonge S.)nmffffwn^^wwm^“ m ,“ , *mw"*^"mn,“'w“ w«““™ * “ ,™ r™ ,“ “ n,*» ™ "^ m *"r™ “ w“ ™^mmmnmmi^m. “ m DoN and notwre of (Kongo HO OventPy d'KOun* TreRe diKaunl f Ce«*> dnceuni \ dixauntt boon deducted Saesanef d>tceuni l Hero #ny indicated tram IK rapartad prue* # O W . t 'K a v r l 1 Oma* (Karpai Hare any at mate | bean included* (■I'M (*RR« 4 PRICE IN FO R M A TIO N For the com m odify described in item I , pleo te enter below the current price for the dote in d icate d , on the b o u t quoted in •tern 2 . Price a* of Mar O M IC IH e O tT I Apr 13 1971 Moy 11 1971 June 15 1971 e m c t O t T I o r C H I N K 9 17 91 O ffice M IC IN I (H any) OAT| Oct 12 9 o a r s O ffice o p C H I N K ft# a n y ) 1971 Nov * m c « 1971 Dec 14 1971 Ju ly 13 1971 Jon 11 1972 Aug 10 1971 Feb 15 1972 Mar 14 1972 Sept 14 1971 r n P E R M A N EN T O F F IC E R EC O R D KIN O LY RETURN L THIS FORM PROM PTLY WHOLESALE PRICES Technical References Number 1. Cunningham, Francis S., “ The Use of Price Indexes in Escalator Contracts,” Monthly La bor Review, August 1963, pp. 948-952 Reprint No. 2424. A statement of the use of the Wholesale and Consumer Price Indexes in escalating pur chase and sales contracts and wages, with some specific suggestions and pitfalls noted. 2. Evans, W . Duane and Hoffenberg, Marvin, “ Input-Output Relations and Appraisal,” in Studies in Income and Wealth, New York, National Bureau of Economic Research, (1955) Vol. 18. A statement of the conceptual framework, data, pricing problems and significance in economic analysis of the U.S. Government’s interindustry statistical study of 1947. 3. Evans, W . Duane and Hoffenberg, Marvin, “ The Inter-Industry Relations Study for 1947,” The Review of Economics and Statistics, May 1952, pp. 97 -142. A description of scope, uses, and method of the U.S. Government’s interindustry sta tistical study of 1947. Includes discussions of computational problems, areas of use, data requirements, etc. 4. National Bureau of Economic Research. The Price Statistics of the Federal Government: Review, Appraisal, and Recommendations, Washington, D.C., NBER General Series, Num ber 73 (1961). An appraisal of price statistics of the Federal Government by the Price Statistics Review Committee of NBER, covering uses, concepts, collection, and publication, sampling, and other aspects of the Consumer Price Index, Wholesale Price Index, Index of Prices Paid by Farmers, and other price measures. 5. Searle, Allan D., “ Weight Revisions in the Wholesale Price Index, 1890-1960,” Monthly Labor Review, February 1962, pp. 175-182. History of weight changes and weighting concepts, from inception of the Wholesale Price Index. 6. U.S. Congress, Joint Economic Committee. Government Price Statistics: Hearings: Subcom mittee on Economic Statistics, 87th Congress, 1st sess., Part 1, Jan. 24, 1961, 526 pp.; Part 2, May 1-5, 1961, 265 pp. Part I presents the report, Price Statistics of the Federal Government, prepared by the NBER (q.v.) ; Part II contains statementer of Labor Statistics. the response of Ewan Clague, Commissions of private and government economists including 7. U.S. Department of Labor, Wholesale Prices, 1890 to 1899 (Bulletin 27, 1900). Describes an inquiry into the course of wholesale prices for the purpose of continuing the study contained in the Report on Wholesale Prices, Wages, and Transportation made by the Senate Committee on Finance, March 3, 1893 (pp. 237-313). 8. U.S. Department of Labor, Course of Wholesale Prices, 1890 to 1901 (Bulletin 39, March 1902). Describes United States Senate Finance Committee index (pp. 205-211), and Depart ment of Labor index (pp. 212-243). 9. U.S. Department of Labor, Bureau of Labor Statistics, Wholesale Prices and Price Indexes, 195U-56 (Bulletin 1214, 1957). Method of Calculating Special Group Indexes, (pp. 1 2 -1 3 ), Calculating Relative Im portance Data (p. 14), Description of Indexes by Stage of Processing (Economic Sector Indexes) (pp. 15-22) ; A Possible Effect on Weight Revisions (p. 7). 10. U.S. Department of Labor, Seasonal Adjustment Factors; Wholesale Price Index: Selected Series 1948-1961 (BLS Bulletin 1379, 1963). Seasonal adjustment factors for 183 commodities and commodity groups, and descrip tion of BLS seasonal adjustment method. 11. U.S. Department of Labor, Wholesale Prices and Price Indexes, 1957, BLS Bulletin 1235 (1958). Indexes by Durability of Product (Economic Sectors by Durability of Product), pp. 11-14. 12. U.S. Department of Labor, Bureau of Labor Statistics, Wholesale Prices and Price Indexes, 1958, BLS Bulletin 1257 (1959). Describes Supplementary Inquiry on Wholesale Price Reports (discount study), pp. 1012, and January 1958 Revision of the Weighting Structure, pp. 14-16. 13. U.S. Department of Labor, Wholesale Prices and Price Indexes, 1961, (BLS Bulletin 1382, 1964). 111 112 BLS HANDBOOK OF METHODS Technical References— Continued Number 14. 15. 16. 17. 18. 19. January 1961 Revision of the Weighting Structure, pp. 14-16. U.S. Department of Labor, Bureau of Labor Statistics, Wholesale Prices and Price Indexes, January 1967 (final) and February 1967 ( final). Describes introduction of new 1963 weight values and major reclassification effected in January 1967. U.S. Department of Labor, Wholesale Prices and Price Indexes, January 1970. Describes derivation and use of relative importances (weights) and lists all W P I weights for December 1969. U.S. Department of Labor, Wholesale Prices and Price Indexes, January 1971. Introduces the new standard reference base, 1967=100, and describes conversion from the former base. U.S. Senate, Wholesale Prices, Wages, and Transportation, Senate Report No. 1394, Part I, “ The Aldrich Report” (1893). Contains a summary of the complete Senate report on wholesale prices, on wages, and on transportation made in response to a Senate resolution of March 3, 1891. Searle, Allan D., “ Toward Comprehensive Measurement of Prices,” Monthly Labor Review, March 1971, pp. 9-22. Describes how a general price index could be constructed, what it should accomplish, and virtues and limitations of various approaches. Clorety, Joseph A ., Jr., “ Measuring Changes in Industrial Prices,” Monthly Labor Review, November 1970, pp. 30-36. Relates Stigler-Kindahl study (N B E R , 1970) to BLS program. — T h o m a s R . T ib b e t t s Chapter 12. Industry-Sector Indexes Background During recent years a growing need for com prehensive measures of industrial prices, in addition to the market oriented prices of the Wholesale Price Index, has become increasingly apparent. As a result, the Bureau initiated a program of industry-sector price indexes based upon data collected for the WPI. An industry or sector price index is es sentially a composite index made up of price series that match the economic activity of a defined industry or economic sector. The Wholesale Price Index, on the other hand, is compiled according to commodity rather than industry groupings. The number of sectors or industries for which these series are compiled depends largely upon resources available for additional pricing. A set of industry-sector price indexes cover ing the years 1947 through 1953 was prepared in the early 1950’s as part of the Bureau’s project on interindustry economics. These in dexes, generally a regrouping of the Wholesale Price Indexes into the interindustry (inputoutput) classification structure, were designed for revaluing bills of goods and industry out puts. In 1959, another set of such indexes was compiled for the Bureau of the Census in con nection with that agency’s construction of the 1958 production index benchmark.1 This sec ond group of price indexes was used in deflat 1 None of these earlier indexes was published, but they were made available to other government research agencies. See Chapter 31, section on Economic Growth Studies, for background on interindustry studies. 2 The classification of establishments into industries, in this program, follows the guidelines established by the Office of Management and Budget in its Standard Industrial Classifica tion (SIC) system, as revised in 1957. Under this classification system, related products or services are grouped together and given an industry code number (consisting of 4 digits). Every establishment is assigned to the industry in which its most important products or services, in terms of values, are classi fied. Many industries contain establishments which produce significant quantities of goods and services that are classified in other industries. These goods usually are referred to as "secondary products.” See appendix B. 3 Government Price Statistics, Hearings before the Subcom mittee on Economic Statistics, of the Joint Economic Com mittee, Congress of the United States, Part I, January 24, 1961, page 64. Also see report of United Nations Economic and Social Council; Problems and Methods in the Gathering of Representative and Comparable Wholesale Price Series, E/CN. 3/264, 15 March 1960, Chapter II. ing values of shipments in those census product classes where physical production data were lacking or unsatisfactory. Again, these were essentially indexes of commodity prices, classi fied as primary to a given industry.1 2 The need for the Bureau to develop IndustrySector Price Indexes became increasingly ap parent in 1960 and 1961, when the Price Statistics Review Committee of the National Bureau of Economic Research recommended to the Bureau of the Budget that the basic ob jectives of an industrial price program should be comprehensiveness, maximum detail in re porting, and groupings most useful in economic analysis.3 The committee stated “ . . . It seems desirable that the subclassification should aim at fitting into the Standard Industrial Classi fication.” In 1962, the Bureau of Labor Statistics intiated the development of industry and sector price indexes. Because of its scope, the pro gram was viewed as a long-run program to be accomplished in several stages. The first stage was devoted to the study of conceptual and data problems with only a gradual expansion of commodity pricing. The first indexes to be developed were out put price indexes utilizing gross shipments weights. Priority was given to indexes for the manufacturing and mining divisions of the Standard Industrial Classification. The next stage will be the development of output price indexes for the trade and transportation and other nonmanufacturing sectors. Input price indexes, i.e., indexes representing the price of industrial purchases, will come later. Eventu ally the work will expand sufficiently to permit the development of a general price index of the entire economy. Description o f the Survey Concepts An industry or sector price index is a com posite index derived from several series of prices that closely match the economic activity of a specified industry or industry sector. These 118 114 BLS HANDBOOK OF METHODS indexes may be either output or input price indexes based upon either the products and services sold or the products and services pur chased by an industry. An output price index for a given industry represents price indexes for a sample of the products produced by that industry, averaged together according to the relative importance of production of each sam ple product to the industry. An input price index for an industry consists of an aggre gation of price indexes for a sample of the commodities and services purchased by the industry, averaged together according to the relative magnitude of the purchases. The Bureau’s work has been directed first toward two sets of output indexes. One set is weighted by gross shipments of products “ made in the industry” to be used for deflating industry shipments. The second set will be output price indexes of shipments classified on the industrial basis but weighted by shipments of the product produced anywhere in the eco nomy. A principal use of the second set is for input-output analysis.4 Universe Ultimately, the scope of the universe will be defined in terms of the Office of Management and Budget’s Standard Industrial Classification (SIC) system as revised in 1967, which covers all domestic economic activity. This system groups together related products or services and assigns them industry and sector codes. Currently, however, the scope of pricing is effectively restricted to the commodities’ cover age of the WPI because of the use of WPI price data. If price indexes are to parallel industry out put data, the indexes should cover the total output of each industry including the value of * 8 * See Chapter 31, section on Economic Growth Studies, for a discussion of the nature and uses of input-output data. See also W . Duane Evans and Marvin Hoffenberg, “ The Inter industry Relations Study for 1947,” The Review of Economics and Statistics, May 1952, pp. 97-142; and “Input-Output Rela tions and Appraisal,” Studies in Income and Wealth, vol. 18, National Bureau of Economic Research, New York, 1955. BThe SIC provides no product codes. 8 The Input-Output model referred to is that compiled by the Office of Business Economics, U.S. Department of Com merce. See National Economics Division, “ Input-Output Struc ture of the U.S. Economy: 1963,” Survey of Current Business, November 1969, pp. 16-47. interplant transfers, the value of sales to all classes of customers, and the value of indus trial services. They should include the value of sales for export but exclude excise taxes and costs of transporting finished goods to pur chasers. This is consistent with the “ total activity” coverage of statistical series on em ployment and production. Input price indexes of materials consumed in production should cover total materials in puts of the industry. This would include im ports for consumption, and also transportation and delivery costs. Prices and Base Period The prices used in the current IndustrySector Indexes are in general those used in the Wholesale Price Index. In the Wholesale Price Index, primary market prices, f.o.b. production point, are used. For the Industry-Sector In dexes, pricing eventually should be extended to all classes of customers, including retail, for use in the output indexes. Buyers’ prices, in cluding shipping costs, should be used for input indexes and should represent the particular mix of products purchased by the buying in dustry. As of January 1971, the reference base period for Federal Government indexes is 1967=100. Classification The Standard Industrial Classification (SIC) of the Office of Management and Budget pro vides the framework for the Industry-Sector Index classification scheme. Within this frame work, individual products are given a 7-digit code by the Bureau of the Census.5 The product indexes are aggregated to 5-digit product classes and 4-digit industries. Industry indexes can be aggregated to 3- and 2-digit levels as well. Four-digit industry indexes also can be aggregated to fit the sectoring plan of the latest Input-Output model.6 Sampling and Estimating Procedures Sampling Currently the Industry-Sector program largely depends on price data already available INDUSTRY-SECTOR INDEXES in the Wholesale Price Index. However, ex tension of industrial pricing is being geared to the new program. A sampling plan has been prepared which outlines industrial sectors that should have priority as new pricing is under taken. As the result of this analysis, pricing may be cut back in some sectors to permit ex tension of pricing into inadequately covered sectors. Price data used in computing an industryoutput price index should be representative of the output of the industry, that is, of the values of products made in plants classified in the industry, but should exclude prices of products primary to the industry but made in plants classified in other industries. Even though the products are the same, the prices received by the primary and secondary industries may be different, sometimes because these industries sell to different types of users. As a rough guide to the adequacy of sampling, the im mediate objective is to represent at least 50 percent by value of the commodities included in each 5-digit Census product class.7 This per 8 centage will be slightly less at higher levels of aggregation. This criterion will be adjusted to levels indicated by differences in variability of price change among product classes as ex perience makes possible the use of more sophis ticated sampling approaches. Weights Since January 1967, weights for the output indexes are 1963 value of shipments obtained from the Census of Manufactures, the Census of Mineral Industries, and data of the U.S. Department of Agriculture. Indexes for 1957 through 1966 are weighted by 1958 values. Values include interplant transfer values, values for goods produced and consumed in the same establishment, and the value of goods sold for export. Values of imported commodi ties are not included. The difference in the scope of the weights, as compared with the WPI, stems from the objective in this system to match price data with the scope of domestic industry production. Each priced product actually represents a class of commodities and is assigned its own 115 weight plus the weights of other products not directly priced in the index but whose prices are known or assumed to move similarly. Values for unpriced products which cannot be assigned to a specific priced commodity are imputed to the average movement of the prod uct classes in which they fall.7 For use in deflating industry shipments, the 4-digit (SIC) Industry Indexes are derived from 5-digit product class indexes weighted together by their shipments value for the par ticular industry, i.e., the “ made-in-the-industry” value. Formula and Calculation A modification of the Laspeyres fixed-weight formula is used. The underlying formula is: h = 2 ( P t/ P „ Q a)/zPoQa where P«/P„ is an indi vidual product price index. P „Q a, the value weight (base year price times weight year quantity), currently is the 1963 value of ship ments adjusted for price change from the year 1963 to 1967. In succeeding years, new weights will be introduced whenever weights are re vised for the comprehensive Wholesale Price Index.8 In actual practice the calculation may be somewhat more involved than indicated by the simple formula above. For example, indexes used for deflating industry shipments values should employ product weights based upon values of commodities made within the same industry. Since the Census data for such values are available only at the 5-digit (productclass) level, it is necesary first to construct product-class indexes based upon total output (wherever made) weights. Then, using 5-digit made-in-the-industry weights, the productclass indexes are combined to the 4-digit indus try level. 7 This procedure is the same as that employed in the WPI. However, as the product classes are defined differently, an unpriced commodity may have a different price movement imputed to it in the Industry Index program than it has in the WPI. 8 Much of the value used in weighting the WPI is derived from the periodic Census of Manufactures and Mineral In dustries which are currently collected on 5-year cycles. See Chapter 11, Wholesale Prices. 116 BLS HANDBOOK OF METHODS Analysis and Presentation The published indexes for selected 5-digit product classes and 4-digit industries are an nual averages for the period 1957 through 1964, and cover 44 manufacturing and eight mineral industries. Monthly indexes are avail able beginning January 1965 for the same limited number of industries and products. Additional indexes are published as they be come available. By January 1971, indexes were published for 102 four-digit industries and about 350 five-digit product classes. Uses and Limitations Price statistics organized along industrial lines have particular relevance to studies of economic growth, productivity, and other types of industrial and economic analysis where the emphasis is on industrial structure as distinct from market or commodity-use classifications. Whether an index meets a given specific need depends largely upon its commodity coverage and its weighting structure. An important use of an output index weighted by gross shipment values is to deflate value series in order to arrive at measures of output in constant dollars. Most measures of output and productivity rely pri marily upon physical quantity data for the various products of an industry, but in cases where quantity data are not available, deflated values can be used if suitable price indexes are available for use as deflators. Deflated value data also may serve as a check on production indexes prepared from quantity data and unitvalue weights. There are many sectors of the economy for which the analysis of industrial output is severely limited because appropriate price indexes are not available. Essentially the process of deflation provides a means of obtaining an estimate of quantity change from available data on total dollar value and a price index. If the dollar values themseves are divided by the price index, the result ing dollar values express the sales value in terms of purchasing power of the dollar as of the base period of the index. Or an index of dollar volume can be divided by the price index to obtain a production index.9 The output indexes also may be used for comparing movements of prices with other in dustry-based statistical measures such as em ployment, earnings, productivity, etc. Price indexes consistent with total shipments weights will be useful for deflating industry inputs. For example, the appropriate index for deflating the value of aluminum purchased by an indus try would be the index whose components rep resent shipments of aluminum to buyers in this industry rather than the aggregate output of the primary aluminum industry. Input price indexes will be especially useful to research departments in private industry as well as to public agencies in making cost stud ies. They should be consistent in coverage with BLS series on average hourly earnings, another important element of cost. For contract escala tion, they will'give index users a wider choice of indexes. Input price indexes, however, are not available yet. There are a number of uses which combine output and input indexes. For example, gross output price indexes and materials input price indexes can be used to yield a measure of value added in constant dollars. The prices used in contructing the currently published indexes are those regularly collected on a monthly basis and used in the calculation of the comprehensive Wholesale Price Index. These prices generally are at the primary market level but a few are at other levels. It must be assumed that these price movements are similar to the market level of sales repre sented by the Census data. To include inter plant transfer values and values of goods produced and consumed in the same industry, it is necessary to assume that price movements of goods in commercial markets represent the price changes of goods not sold in commercial markets. Until additional pricing can be done, these new indexes will be limited by the coverage— commodity and class of customer— of the com prehensive Wholesale Price Index. 0 It can be shown that division of the value index by the Laspeyres (base-year-weighted) price index yields a produc tion index of the Paasche (current-year-weight) form. Di vision by the Paasche price index, conversely, yields a quantity index of the Laspeyres type. See Chapter 2S, Output per Manhour Measures: Industries. 117 INDUSTRY-SECTOR INDEXES Technical References Number 1. Evans, W . Duane and Hoffenberg, Marvin, “ The Inter-Industry Relations Study for 1947,” The Review of Economics and Statistics, May 1952, pp. 97-142. A description of scope, uses, and methodology of the U.S. Government’s interindustry statistical study of 1947. 2. Goldman, Morris R .; Marimont, Martin L .; and Vaccara, Beatrice N., “ The Inter-Industry Structure of the United States,” Survey o f Current Business, November 1964, pp. 10-29. A report showing preliminary results of the 1958 interindustry relations study and con taining tables of the percent distribution of 1958 gross output in 86-industry detail. 3. Moss, Bennett R., “ Industry and Sector Price Indexes,” Monthly Labor Review, August 1965, pp. 974-982. Contains price indexes for about 50 4-digit (Standard Industrial Classification) indus tries, together with a technical note on concepts, methodology, and uses. 4. National Bureau of Economic Research, “ Input-Output Relations and Appraisal,” Studies in Income and Wealth, vol. 18, New York, 1955. A description of scope, uses, and significance in economic analysis of the U.S. Govern ment’s interindustry statistical study of 1947. 5 . --------- The Price Statistics of the Federal Government, New York, 1960. An appraisal of price statistics of the Federal Government by the Price Statistics Review Committee of NBER, covering uses, concepts, collection, and publication, sampling, and other aspects of the Consumer Price Index, Wholesale Price Index, Index of Prices Paid by Farmers, and other Price measures. 6. U.S. Congress, Joint Economic Committee. Government Price Statistics, Hearings before the Subcommittee on Economic Statistics, Part I and Part II, Washington, D.C., 1961. Part I represents the report, The Price Statistics of the Federal Government, prepared by the NBER (q .v .); Part II contains statements of private and government economists in cluding the response by Ewan Clague, Commissioner of Labor Statistics. Mary E. Lawrence Chapter 13. Spot Market Prices Background As early as January 1934, at the request of the U.S. Department of the Treasury, the Bu reau of Labor Statistics began the computation of a daily commodity price index, using quota tions for sensitive comodities. It was released first to the general public in January 1940. In 1952, in connection with the revision of all its major price index series, the Bureau issued a new Daily Index of Spot Market Prices. The new index was not a continuation of the old series, but was based on a new sample of 22 commodities and was calculated on a 1947-49 base; in contrast, the old index was based on 28 commodities and was calculated with August 1939 as base. In January 1962, the 22-commodity index was recalculated on a 1957-59=100 base to cor respond to the base period adopted for other Federal Government general purpose indexes. In January 1971, the index was rebased again in accordance with government-wide practice, this time to a 1967=100 base. In 1969, com putation of the index on a daily basis was dis continued. Since then the index has been pre pared for Tuesday of each week. Description o f Survey The Spot Market Price Index is a measure of price movements of 22 sensitive basic com modities whose markets are presumed to be among the first to be influenced by changes in economic conditions. As such, it serves as one early indicator of impending changes in busi ness activity. The commodities used are in most cases either raw materials or products close to the initial production stage which, as a result of daily trading in fairly large volume of stan dardized qualities, are particularly sensitive to factors affecting current and future economic forces and conditions. Highly fabricated com modities are not included for two reasons: (1) they embody relatively large fixed costs which fact causes them to react less quickly to changes in market conditions; and (2) they are less important as price determinants than the more basic commodities which are used throughout the producing economy. A spot price is a price at which a commodity is selling for immediate delivery. In the absence of a spot price, a bid or an asked price may be used. Some of the prices used are nominal prices in that they are not actual transaction prices. Often they are exchange prices— a price for a completely standard commodity which eliminates the effect of minor quality changes on actual transaction prices.1 Trade publica tions may use this type of price for commodities such as cocoa beans, coffee, and wool tops. The price for print cloth is an average of spot price and price for most distant forward contract because it was determined that a large part of the sales of print cloth are made on a contract basis. The 22 commodities are combined into an “ All Commodities” grouping, with two major subdivisions: Raw Industrials, and Foodstuffs. Raw Industrials include burlap, copper scrap, cotton, hides, lead scrap, print cloth, rosin, rubber, steel scrap, tallow, tin, wool tops, and zinc. Foodstuffs include butter, cocoa beans, corn, cottonseed oil, hogs, lard, steers, sugar, and wheat. The items upon which the index is based are classified further into four smaller groups: Metals, Textiles and Fibers, Livestock and Products, and Fats and Oils. However, some of the 22 commodities do not fall into one of these four groupings. For example, sugar is not in cluded in any special group. Furthermore, the groupings are not mutually exclusive. Lard, for instance, is in both the Livestock and Products Index and in the Fats and Oils Index. Data Sources and Collection Methods The prices used in the index are obtained from trade publications or from other Govern 1 Exchanges which issue spot prices have committees to make a determination of the spot price for the standard commodity. 119 120 BLS HANDBOOK OF METHODS ment agencies. Prices for cocoa beans, steers, sugar, wheat, burlap, copper scrap, cotton, lead scrap, print cloth (spot), rosin, rubber, steel scrap, wool tops, and zinc, are of the same specification and source as those used in the comprehensive monthly Wholesale Price In dex.2 Prices for butter, corn, hides, hogs, lard, tallow, and tin are either differently specified spot prices or from different markets. Selection o f Products The criteria for the selection of commodities were (1) wide use for further processing (basic), (2) freely traded in an open market, (3) sensitive to changing conditions significant in those markets, and (4) sufficiently homo geneous or standardized so that uniform and representative price quotations can be obtained over a period of time. Subject to these restrictions, efforts were made to include representative sensitive commodities from as large a segment of the economy as pos sible. Also, the influence of international mark ets upon the economy was taken into account by the inclusion of some key commodities (such as crude rubber and tin) which are important in international trade. Both in the sample and in the index structure, an attempt was made to prevent price movements of agricultural prod ucts from dominating the movement of the index. each of the commodities is unweighted in the index means that a price change for rosin, a comparatively unimportant commodity, has as much effect as an equal percentage movement in the price of a very important commodity such as wheat, cotton, or steel scrap. The computation procedure involves obtain ing for each commodity the ratio of its price in any given period to its price in the base period and taking the 22nd root of the product of these ratios. This product is then multiplied by 100 to obtain the index number for each period. The calculation is made by means of logarithms. The formula reduces to Log Log Pk—2 log P0+44 22 where 7fc = Index for a given day P k = Price for a given day P0 = Average (geometric) price in base period 44 = Logarithmic constant which when di vided by 22 equals log of 100. Monthly average indexes are obtained ac cording to the previous procedure, except that P7=the geometric average of the Tuesday c prices (daily prices prior to 1969) over the month. In maintaining the index over time, it may be necessary to change commodity specifi cations or substitute entirely new products. These changes are handled by a statistical link ing procedure so that only actual price move ments are reflected in the index. Estimating Procedures The Spot Market Index is an unweighted geometric mean 3 of the individual commodity 4 * price relatives, i.e., of the ratios of the current prices to the base period prices. The use of the geometric mean has the advantage that the index is not dominated by extreme price move ments of individual commodities. Since ex tremely large movements may be atypical, it was deemed better to minimize their effects, even at the expense of losing the effect of large representative changes. However, the fact that 2 See Chapter 11, Wholesale Prices. The geometric mean of n figures is the nth root of their product. Thus, the geometric mean of the numbers 1.5, 2.0, and 9.0 is 3.0 (1 .5 x 2 x 9 = 2 7 . 3V27—3). The arithmetic mean, is 4.2. Analysis and Presentation Tuesday spot market indexes and prices are published each week, on the Friday following the day of reference. A summary of weekly indexes and the average for each month are published with the first weekly release of the following month. Beginning with 1950, his torical indexes are shown for Tuesday of each week together with monthly averages; from July 1946 through 1949 indexes are listed for Tuesday of each week only. In addition, in dexes are published for selected earlier dates: August 15, 1939, December 6, 1941, August 17, 1945, and June 28, 1946. 121 SPOT MARKET PRICES Uses and Limitations A survey of users in 1964 showed that the Index is frequently used as a general economic indicator, for gaging the direction of basic prices, for forecasting general price move ments, and for current prices of specific com modities. Other uses, frequently mentioned, are for market research and for comparing price trends with the user’s selling or buying prices. The Tuesday Index of Spot Market Prices dif fers from the Wholesale Price Index in method of construction and weighting, as well as in the sample of items for which prices are included. While it is independent of the monthly compre hensive index, changes in the Tuesday Index or its components may foreshadow turns in Wholesale Price Indexes. However, the Tues day Index is not a good indicator of current price trends for the whole economy. For this purpose, the comprehensive Wholesale Price Index should be used. The Tuesday Spot Market Index is, by design, very sensitive to price changes in basic commodities but, because of its unweighted structure, the magnitude of changes in any of the index groups cannot be used as a reliable measure of the general price change of all commodities within the groups. For many of the 22 items, the commodity ex change prices are based upon transactions which cover as little as 25 percent of the total sold in all markets. In some cases, the price is set by a committee of experts from the com modity exchange for a standardized com modity. Also, when there are not enough trans actions from which to obtain an actual market price, a “ nominal” spot price is set. From this, it is apparent that the exchange prices may not always be representative of the large vol ume of private transactions occurring outside the organized market. However, it is believed that the reported exchange prices generally are used as the basis for private negotiations. Composition o f Grouping Indexes Metals: Copper scrap, lead scrap, steel scrap, tin, and zinc. Textiles and Fibers: Burlap, cotton, print cloth, and wool tops. Livestock and Products: Hides, hogs, lard, steers, and tallow. Fats and Oils: Butter, cottonseed oil, lard, and tallow. Specifications for Commodities Included in the Index as of March 1971 Market Specifications 10 oz„ 40", ex-dock or ex New York. warehouse, duty paid, per yd. Chicago. Grade A, 92 score, per lb. — Butter New York. _ Cocoa beans__ Accra, per lb. No. 1 heavy copper and wire, New York. Copper scrap refiners’ buying price, car load lots, delivered buyers’ works, per lb. Chicago. Corn _______ No. 3 yellow, per bu. - .. 12 markets. Middling, ll£ f " , per lb. Cotton Memphis. Crude, Valley, per lb. ___ Cottonseed oil. Cow, light native, packer 30/53 Chicago. Hides lbs., fleshed, packer to tan ner, dealer, or exporter per lb., f.o.b. shipping point. U.S. No. 2’s and 3’s, 200-220 Omaha. Hogs lbs., per 100 lb. Prime Steam, in tanks, per Chicago. Lard lb. Battery plates, smelters’ buy New York. Lead scrap ing price, East, carload lots; delivered buyers’ works, per lb. New York. Print cloth __ 48", 78x78 count, 4 yds./lb. spot and nearby, per yd. Print cloth __ 48", 78x78 count, 4 yds./lb., New York. most distant contract, per yd. Gum, W G grade, carlots, per New York. Rosin 100 lb. Crude, natural, No. 1 Ribbed New York. Rubber Smoked Sheets, per lb. Steel scrap __ No. 1 heavy melting, (dealer), Chicago. consumers’ buying price, in cluding brokerage, de livered, per gross ton. Steers ________ Choice, 900-1100 lbs., per 100 Omaha. lb. Raw, 96°, duty paid, per 100 New York. Sugar ______ lb. Tallow _______ Fancy, bleachable, inedible, Chicago. per lb. _ Tin __________ Grade A, spot delivery, per New York. lb. No. 1 Dark Northern Spring, Minneapolis. Wheat ___ per bu. Wheat _______ No. 1 Hard Winter Ord., per Kansas City. bu. Certificated spot price, nom New York. Wool tops inal, per lb. Prime Western, for New York. Zinc _________ Slab, prompt delivery, delivered, (f.o.b. New York equivaCommodity Burlap - L loyd E. W ig r e n Wag es and Industrial Relations Chapter 14. Occupational Pay and Supplementary Benefits Background The Bureau of Labor Statistics, for many decades, has conducted studies of wages by oc cupation and industry, based upon employer records. The Bureau’s first such study, growing out of a study by the U.S. Senate in 1891, re sulted in a wage rate record extending back continuously to 1860. Systematic collection of wage data by occupation and industry has con tinued since the turn of the century; changes in coverage has been dictated mainly by gov ernment requirements. A large survey program undertaken for the War Industries Board in 1919 produced occupational pay rates by indus try and State, and (for some industries) by city. Between 1934 and 1940, the selection of industries studied was determined largely by administrative needs under the National Re covery Act, Public Contracts Act, and the Fair Labor Standards Act, with emphasis on nation wide data for relatively low-wage industries. Survey activity shifted in the 1940-41 de fense period to heavy industries essential to war production. Implementation of wage sta bilization policy during the war required a large-scale program of occupational wage stud ies by industry and locality. The emphasis on data by locality has continued since 1945 within the framework of industry studies generally designed to also yield national and regional estimates. In addition, the Bureau developed two new types of surveys. Area wage surveys, initiated in the late 1940’s, were designed to meet the growing de mand for pay data related to office clerical and manual jobs that are common to a wide variety of manufacturing and nonmanufacturing in dustries within metropolitan areas. This survey program was firmly established and tem porarily expanded for use in the wage stabili zation effort during the Korean emergency. The need for nationwide estimates of white-collar pay in private industry for use in appraising the Federal white-collar salary structure re sulted in a survey design that would produce national averages, based on an area sample. Data for individual areas studied also serve the wage administration needs for other govern ment agencies. Prior to 1960, studies in a very few profes sions provided salary data. Beginning in that year, salary surveys have been made on a nationwide basis covering professional, admin istrative, and technical jobs in a broad spec trum of industries. Averages for these jobs, together with national averages for clerical and drafting jobs included in the area wage sur veys, are utilised by the administrative agen cies directly concerned with Federal pay matters. Description o f Surveys Although differing in industrial, geographic, and occupational coverage, the three types of surveys described form an integrated program of occupational wage surveys based upon a common set of administrative forms, manual of procedures, and common concepts and defini tions. Employer cooperation in surveys is on a voluntary basis. Confidential individual estab lishment data compiled by the Bureau’s field economists are grouped in published reports in a manner that will avoid possible disclosure of an establishment’s rates. Establishments in cluded in all surveys are classified by industry as defined in the 1967 edition of the Standard Industrial Classification Manual prepared by the U.S. Office of Management and Budget.1 Survey reports identify the minimum size of establishment (measured by total employment) studied. Definitions for Standard Metropolitan Statistical Areas are employed in all pro grams.2 1 See appendix B. a See appendix C. 123 124 BLS HANDBOOK OF METHODS Industry wage surveys provide data for occu pations selected to provide representativeness of the range of rates, methods of wage payment, and of men’s and women’s work ac tivities. Consideration also is given, in their selection, to the prevalence in the industry, definiteness and clarity of duties, and impor tance as reference points in collective bargain ing. In addition to collecting straight-time firstshift rates (or hours and earnings for incentive workers) for individual workers in the selected occupations, surveys in most industries also establish the wage frequency distribution for broad employment groups, i.e., production and related workers or nonsupervisory workers. Weekly work schedules; shift operations and differentials; paid holiday and vacation prac tices; and health, insurance, and pension bene fits are included in the information collected, along with the provisions made for other items, applicable to certain industries. The studies also provide estimates of labor-management agreement coverage, proportions employed un der incentive pay plans, and the extent to which establishments provide a single rate or range of rates for individual job categories. Fifty manufacturing and 20 nonmanufactur ing industries, accounting for about 22.5 mil lion employees, are surveyed on a regularly recurring basis. A majority are studied on a 5-year cycle, but a number of comparatively low-wage industries are on a 3-year cycle. In addition, special wage surveys also are under taken at the request of other government agencies. Nearly all of the manufacturing, utilities, and mining industries are studied on a nation wide basis and estimates are provided also for regions and major areas of concentration. Sur veys in trade, finance, and service industries usually are limited to a score or more of metro politan areas. Nationwide surveys generally develop separate estimates by size of establish ment, size of community, labor-management agreement coverage, and type of product or plant group. Area wage surveys provide data for occupa tions common to a wide variety of industries in the communities surveyed. The 76 occupational categories studied include 31 office clerical; 15 electronic data processing, drafting, and indus trial nurses; and 30 maintenance, toolroom, powerplant, and custodial and material move ment jobs. Thus, they provide representation of the range of duties and responsibilities as sociated with white-collar, skilled maintenance trades, and other “ indirect” manual jobs. Weekly salaries reported for individuals in white-collar jobs relate to regular straight-time salaries that are paid for standard workweeks. Average hourly earnings for maintenance and other manual jobs relate to first-shift hourly rates. Industry divisions included are (1) manu facturing; (2) transportation, communication, and other public utilities; (3) wholesale trade; (4) retail trade; (5) finance, insurance, and real estate; and (6) selected service industries. Establishments employing fewer than 50 work ers are excluded— with a minimum of 100 applying to manufacturing; transportation, communication, and other public utilities; and to retail trade in the dozen largest communities. In addition to the all-industry averages and distributions of workers by earnings classes, separate data are provided for manufacturing and nonmanufacturing in each area and, wherever possible, for individual industry di visions in the nonmanufacturing sector. Among the 89 Standard Metropolitan Statistical Areas in this annual survey program as of 1971, sep arate data are provided for transportation, communication, and other public utilities in 88 areas; for retail trade in 25 areas; for whole sale trade and finance, insurance, and real estate in 17 areas; and for the selected service industries in 8 large areas. In 22 of the larger areas, wage data are presented separately for establishments that have 500 workers or more. Data on weekly work schedules; paid holiday and vacation practices; and health, insurance, and pension benefits are recorded separately for nonsupervisory officeworkers and plant workers (nonoffice). Shift operations and dif ferentials are collected for plantworkers in manufacturing. Data on minimum entrance rates for inexperienced officeworkers are col lected in all industries. These items are studied biennially in all areas. This survey program OCCUPATIONAL PAY AND SUPPLEMENTARY BENEFITS also has developed information on profit-shar ing plans, characteristics of sick leave plans, wage payment systems, and other items related to employee compensation. Special area wage surveys have been con ducted annually since 1967 at the request of the Employment Standards Administration for use in administering the Service Contract Act of 1965. The surveys also meet the needs of the general public and provide information on hourly earnings for 14 office occupations; 22 maintenance, toolroom, powerplant, and cus todial and material movement jobs; 7 laundry jobs; and 6 food service jobs. The industrial scope includes manufactur ing; transportation, communication, and other public utilities; and wholesale trade industry divisions; and general merchandise stores; eating and drinking places; real estate, hotels and other lodging places; engineering and architectural services; personal services; and miscellaneous business services. Data on inci dence of paid holidays and vacation practices, and health insurance and pension benefits are provided biennially. The National Survey of Professional, Administative, Technical, and Clerical Pay provides a fund of broadly based information on salary levels and distributions in private employment. The 80 occupation-work levels studied were selected from the following fields: Accounting, legal services, personnel management, engi neering and chemistry, buying, clerical super visory, drafting, and clerical. Definitions for these occupations provide for classification of employees according to appropriate work levels (or classes). Although reflecting duties and responsibilities in industry, the definitions were designed to be translatable to specific pay grades in the General Schedule applying to Federal Classification Act employees. This sur vey, thus, provides information in a form suit able for use in comparing the compensation of salaried employees in the Federal civil service with pay in private industry. Average salaries, monthly and annual for all occupations and also on a weekly basis for clerical and drafting, relate to the standard salaries that were paid for standard work schedules, i.e., to the straight-time salary cor 125 responding to the employee’s normal work schedule, excluding overtime hours. Nationwide salary distributions and averages are presented for men and women combined. Averages also are presented for establishments in metropoli tan areas combined and for establishments employing 2,500 workers or more. Industry divisions included are: (1) manu facturing, (2) transportation, communication, electric, gas and sanitary services, (3) whole sale trade, (4) retail trade, (5) finance, insur ance, and real estate, and (6) engineering and architectural services, and commercially operated research, development, and testing laboratories. Limited to the Nation’s metropolitan areas for the years 1960 through 1964, the annual survey was expanded in 1965 to include non metropolitan counties. The minimum establish ment size included in the survey was raised from 100 to 250 in 1961. In 1966, the minimum establishment size was lowered to 100 in trans portation, communication, and the other public utilities; wholesale trade; and the service in dustries studied. The minimum was also low ered to include establishments with 50 workers in the finance, insurance, and real estate in dustries. Since the survey scope is subject to change, users are directed to the Scope and Method of Survey appendix in the reports for a description of current practice.3 Concepts. The Bureau’s occupational wage surveys summarize a highly specific wage measure— the rate of pay, excluding premium pay for overtime and for work on weekends, holidays, and late shifts, for individual work ers. In the case of workers paid under piece work or other types of production incentive pay plans, an earned rate is computed by di viding straight-time earnings for a time period by corresponding hours worked. Production bonuses, commissions, and cost-of-living bo nuses are counted as earnings. In general, bonuses that depend on factors other than the output of the individual worker or group of workers are excluded; examples of such non production payments are safety, attendance, 3 The terms “in scope" or “within scope” are used through out this chapter to refer to the coverage of the particular survey being described. 126 BLS HANDBOOK OF METHODS year-end or Christmas bonuses, and cash dis tributions under profitsharing plans. Unless stated otherwise, rates do not include tips or allowances for the value of meals, room, uniform, etc. The earnings figures, thus, rep resent cash wages (prior to deductions for social security, taxes, savings bonds, premium payments for group insurance, meals, room or uniforms) after the exclusion of premium pay for overtime, weekend, holiday, or late shift work. Hours shown for salaried occupations relate to standard weekly hours for which the em ployee receives his regular straight-time salary. Occupational classifications are defined in advance of the survey. Because of the emphasis on interestablishment and interarea compar ability of occupational content, the Bureau’s job descriptions may differ significantly from those in use in individual establishments or those prepared for other purposes. The job descriptions used for wage survey purposes are typically brief and usually more generalized than those used for other purposes. The pri mary objective of the descriptions is to identify the essential elements of skill, difficulty, and 4 An example of a job description: Machinist, Maintenance Produces replacement parts and new parts in making re pairs of metal parts of mechanical equipment operated in an establishment. Work involves most of the following: In terpreting written instructions and specifications; planning and laying out of work; using a variety of machinist’s handtools and precision measuring instruments; setting up and operating standard machine tools; shaping of metal parts to close tolerances; making standard shop computations relating to dimensions of work, tooling, feeds, and speeds of machin ing; knowledge of the working properties of the common metals; selecting standard materials, parts, and equipment required for his work; and fitting and assembling parts into mechanical equipment. In general, the machinist’s work normally requires a rounded training in machine-shop prac tice usually acquired through a formal apprenticeship or equivalent training and experience. 5 In general, workers are included in a classification if the duties as described are performed a major part of the time and the remainder is spent on related duties requiring similar or lesser skill and responsibility. However, in some jobs, particularly office and skilled production-worker categories, workers may regularly perform a combination of duties in volving more than one occupation. Unless indicated other wise in the description, in these situations consideration for classification purposes is given to those elements of the job which are most important in determining its level for pay purposes. Thus, a worker meets the basic concept of the stenographer classification if taking of dictation is a regular requirement of the job even though a majority of time is spent on routine typing. responsibility that establish the basic concept of the job.4 Although work arrangements in any one es tablishment may not correspond precisely to those described, those workers meeting the basic requirements established for the job are included.5 In applying these job descriptions, the Bu reau’s field representatives exclude working supervisors, apprentices, learners, beginners, trainees, handicapped workers, part-time or temporary workers, and probationary workers unless provision for their inclusion is spe cifically stated in the job description. Paid holidays, paid vacations, and health, insurance, and pension plans are treated sta tistically on the basis that these are applicable to all nonsupervisory plant or office workers if a majority of such workers are eligible or can expect eventually to qualify for the practices listed. Data for health, insurance, and pension plans are limited to those plans for which at least a part of the cost is borne by the em ployer. Informal provisions are excluded. Survey Methods Planning. Consultations are held with appro priate management, labor, and Government representatives to obtain views and recom mendations related to scope, timing, selection, and definitions of survey items, and types of tabulations. Particularly in planning surveys in specific industries, these discussions impor tantly supplement comments and suggestions received from the regional offices at the con clusion of the previous study. Reflecting its use in evaluation of Federal white-collar pay, the design of the National Survey of Professional, Administrative, Technical, and Clerical Pay was developed in conjunction with the Office of Management and Budget and the Civil Service Commission. Changes in the survey scope, item coverage, and job definitions are initiated by these agencies. The industrial scope of each survey is identi fied in terms of the classification system pro vided in the Standard Industrial Classification Manual. The scope may range from part of a OCCUPATIONAL PAY AND SUPPLEMENTARY BENEFITS 4-digit code for an industry study to a uniform combination of broad industry divisions and specific industries for the area wage surveys or the salary survey of professional, administra tive, technical, and clerical jobs. The needs of major users are a major consideration in de signing the multi-purpose occupational studies. The minimum size of establishment included in a survey is set at a point where the possible contribution of the excluded establishments is regarded as negligible for most of the occupa tions surveyed. Another practical reason for the adoption of size limitations is the difficulty encountered in classifying workers in small es tablishments where they do not perform the specialized duties indicated in the job defini tions. Considerations in timing of industry surveys include date of expiration of major labormanagement agreements, deferred wage ad justments, seasonality of production (e.g., garments), and interests of users. Wherever possible, area wage surveys are timed to follow major wage settlements as well as to meet the needs of government agencies engaged in wage administration as required by law. The types of occupations studied and criteria used in their selection were identified in the de scription of the various types of surveys. The job list for each survey is selected to represent a reasonably complete range of rates in the wage structure for the employment categories involved, i.e., production and related workers in a specific manufacturing industry or nonsupervisory office, maintenance, material han dling, and custodial workers in a metropolitan area. The established hierarchy of job rates to be found within establishments and industries permits the use of pay data for such key or benchmark jobs for interpolating rates for other jobs. Technological developments or user interests may dictate changes in the job lists and definitions. New definitions for jobs usually are pretested in a variety of establishments prior to their use in a full-scale survey. Questionnaires. Two basic schedules are used in obtaining data in all surveys. The first (BLS 2751A) includes items relating to products or services, employment, shift operations and dif ferentials, work schedule, overtime premiums, 127 paid holidays and vacations, insurance and pension plans, union contract coverage, and other items applicable to the establishment. The second (BLS 2753G) is used in recording oc cupation, sex, method of wage payment, hours (where needed), and pay rate or earnings for each worker studied. Supplementary forms are used to meet particular needs. Collection. Bureau field economists collect data by personal visit to each of the sample estab lishments. Job functions and factors in the establishment are carefully compared with those included in the Bureau job definitions. The job matching may involve review of rec ords such as pay structure plans and organiza tional charts, company position descriptions, interviews with appropriate officials, and, on occasion, observation of jobs within plants. A satisfactory completion of job matching per mits acceptance of company-prepared reports where this procedure is preferred by the re spondent. Generally, however, the field eco nomist secures wage or salary rates (or hours and earnings, when needed) from payroll or other records and data on the selected employer practices and supplementary benefits - from company officials, company booklets, and labormanagement agreements. Area wage surveys in all areas involve per sonal visits every second year with partial collection by mail in the intervening years. Establishments participating in the mail collec tion receive a transcript of the job matching and wage data obtained a year earlier by the field economist, together with the job defini tions. The up-dated returns are scrutinized and questionable entries are checked with the re spondent. Personal visits are made to establish ments not responding to the mail request and to those reporting unusual changes from yearearlier data. The work of all field economists is checked for quality of reporting, with particular atten tion directed to accuracy in job matching. The revisits are made by supervisory and senior economists. Systematic technical audits of the validity of survey definitions, made by staff with specialized training, also are maintained for the technically complex nationwide whitecollar salary survey. BLS HANDBOOK OF METHODS 128 B L S 2751A Budget Bureau No. 44-R0338 Approval expires November 30, 1972 U.S. D E P A R T M E N T OF LA BO R Bureau of Labor Statistics (R e v . January 1970) WAGE S U R V E Y G E N E R A L E S T A B L IS H M E N T IN FO RM A TIO N 1. Your report will be held in confidence E S T A B LISH M E N T ID E N T IF IC A T IO N A . Survey P a y ro ll P e rio d P a y ro ll P e rio d 19___ e s t a b l i s h m e n t STREET 19. n a m e ADDRESS COUNTY, STATE AND ZIP CODE NAME AND T I T L E AREA OF AUTHORIZING OFFICIAL NAME AND T I T L E CODE - TELEP H O N E OF OFFICIAL SUPPLYING DATA 1 9 ___ ADDRESS OF O FFICE FROM WHICH DA TA WAS O B T A I N E D . IF D I F F E R E N T FROM ABOVE T E L E P HONE B . C entral O ffic e NAME OF S TR E E T ( C o m p le te | i f clea r a n c e | a n d / o r d a ta | | o b t a i n e d fr o m COMPANY th is sou rce) NAME ADDRESS CITY, S TA TE AND OF AUTHORIZIN G O FFICIA L ZIP C O D E T IT L E 2. C U R R E N T PR O D U C T S OR S E R V IC E S AND P R O C E S S E S A. P R O D U C T O R S E R VI C E [APPROXIM ATE % ANNUAL VALUE APPROXIMATE % ANNUAL VALUE B. S C O P E O F O P E R A T I O N S 19____ 19____ 3. P R O D U C T OR S E R V I C E O F F IC E USE O N LY S C H E D U L E NO. 1 -5 1 9 _____ 1 9 _____ 1D E N T . 6 -8 AREA 9-11 REGION STATE CITY SIZE SIC C O D E EST. SI Z E UNION 12 13 -1 4 15 1 6 -1 9 20 21 • WEIGHT 22 -2 4 SPECIAL CHAR ACTERISTICS (1 ) 2 5 -27 (2) 2 8 -3 0 129 OCCUPATIONAL PAY AND SUPPLEMENTARY BENEFITS U .S . D E P A R T M E N T O F L A B O R B L S 2751 A —Continued (Rev. January 1970) Bureau of Labor Statistics SURVEY P A YR O LL PERIOD SCHEDULE NO. ESTABLI SHMENT 5. UNION C O N T R A C T C O V E R A G E No A . Are a m ajority o f your p rod u ction w orkers c o v e r e d by union agreem ents? B . Are a m ajority o f your o f f i c e w orkers c o v e r e d by union a g r e e m e n t s ? _____ a a 31 32 nam e and a ffilia tio n □ □ D. What o c c u p a t io n a l group s are co v e r e d by the co n tra ct? C . With what u n ion s d o e s th is esta b lis h m e n t h a ve co n tr a c ts ? ( G iv e Y es b e lo w .) (L is t grou p s b elo w o p p o site th e a p p r o p ria te u n io n .) P ro d u ctio n W orkers: O ffic e W orkers: 6 . E S T A B LISH M E N T E M P L O Y M E N T (A P P R O X IM A T E ) A . What is the approxim ate tota l em ploym ent* in th is e sta b lish m e n t? B . H ow many are n on su p e rv iso ry p rod u ction (p la n t) w orkers? M e n ___________________________________________________________________ ____ W omen____________________________________________________________________ C . N o n s u p e rv is o ry o ffi c e w o r k e r s ? _____________________________ M en _______________________________________________________________________ Women________ ____________________________________________________________ D . O ther e m p lo y e e s (e x e c u t iv e , p r o fe s s io n a l, s u p e rv is o ry , e t c . ) ? ____ E. __________________________________________________________________________ F . __ __________________________________________________________________ In clu d es s a la rie d o ffi c e r s o f c o r p o r a tio n s but d o e s not in clu d e p r o p rie to rs , members o f u n in co rp o ra te d firm s, p e n s io n e r s , members of the arm ed fo r c e s ca rried on the p a y r o ll, or unpaid fa m ily w o r k e rs . G . Rem arks 1 130 BLS HANDBOOK OF METHODS OCCUPATIONAL RATES * s 275 3 0 Rev. 65 Payroii period ' Schedule No. r Est. N am e O C C U P A T IO N A N D G RAD E P age Occupational code (1) Sex (2) M ethod of Num ber of pay workers (3) (4) Hours (5) Salary, rate, or earnings (6) Line No. (7) 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 (8) OCCUPATIONAL PAY AND SUPPLEMENTARY BENEFITS Sampling The sampling design employed is almost al ways highly stratified. Before the sample is selected, information on all known establish ments that might possibly fall within the scope of the survey is compiled from lists provided by regulatory governmental agencies (primarily State unemployment insurance agencies), sup plemented by data from trade directories, trade associations, labor unions, and other sources. Establishments then are stratified as pre cisely as available information permits. Each geographic-industry unit for which a separate analysis is to be presented is sampled inde pendently. Within these broad groupings, a finer stratification by product (or other perti nent attributes) and size of establishment is made. Stratification may be carried still further in certain industries: textile mills, for instance, are classified on the basis of integration, i.e., whether they spin only, weave only, or do both. Such stratification is highly important if the occupational structure of the various industry segments differs widely. The sample for each industry-area group is a probability sample, each establishment having a predetermined chance of selection. In order to secure maximum accuracy at a fixed level of cost (or a fixed level of accuracy at minimum cost), the sampling fraction used in the various strata ranges downward from all large estab lishments through progressively declining pro portions of the establishments in each smaller size group, in accordance with the principles of optimum allocation. Thus, each sampled stratum will be represented in the sample by a number of establishments roughly propor tionate to its share of the total employment. Though this procedure may appear at first to yield a sample biased by the over-representa tion of large firms, the method of estimation employed avoids the possibility of bias by the assignment of proper weights to the sample establishments. In the event a sample establishment within scope is unable to supply usable data, a sub stitute is assigned in the same industry-loca tion-size class. (Since no close relation exists between failure to participate in these surveys 131 and the items being studied, little bias is intro duced by this procedure.) The overall non response rate in published surveys averages about 5 percent, and in exceptional cases reaches 10 percent. The size of the sample in a particular survey depends on the size of the universe, the di versity of occupations, and their distribution, the relative dispersion of earnings among es tablishments, the distribution of the establish ments by size, and the degree of accuracy required. Estimates of variance based on data from previous surveys are used in determining the size of the sample needed. As indicated earlier, area wage surveys are limited to selected metropolitan areas. These areas, however, form a sample of all such areas, and, when properly combined (weighted), yield estimates of the national and regional levels. The sample of areas is based on the selection of one area from a stratum of similar areas. The criteria of stratification are region, type of industrial activity as measured by percentage of manufacturing employment, major indus tries, and level of earnings in manufacturing. Each area was selected with probability pro portionate to its nonagricultural employment. The largest metropolitan areas are selfrepresenting, i.e., each one forms a stratum by itself and is certain of inclusion in the area sample. The area sample contained about 80 percent of all nonagricultural employment of the metropolitan area complex of the entire country in 1970. In some strata it was impossible to give some areas their proper chance of selection, because of difficulties in making surveys in these areas primarily due to predominance of single em ployers. Then, too, some areas were chosen to represent some strata because of interest in the area of itself. These departures from a strict probability design, covering only 14 percent of the total, are believed to be negligible in their effect. Estimating Procedures Estimated average earnings (hourly, weekly, monthly, or annual) for an industry or an oc cupation are computed as the arithmetic mean 132 BLS HANDBOOK OF METHODS of the individual employees’ earnings. They are not estimated by dividing total payrolls by the total time worked, since such information al most never is available on an occupational basis. All estimates are derived from the sample data. The averages for occupations, as well as for industries, are weighted averages of indi vidual earnings and not computed on an es tablishment basis. The proportion of employees affected by any fringe provision likewise is estimated from the sample; all workers in each establishment are considered to be covered by the predominant benefit policy in effect, and the entire employment of the establishment is classified accordingly. As mentioned previously, the use of a vari able sampling ratio in different strata of the population would result in biased estimates if straight addition of the data for the various establishments were made. Therefore, each es tablishment is assigned a weight that is the inverse of the sampling rate for the stratum from which it was selected— e.g., if a third of the establishments in one stratum are selected, each of the sampled establishments is given a weight of 3. To illustrate the use of weights, suppose the universe were 7 establishments, from which a sample of 3 was selected. Assume that estab lishment A was drawn from a cell, or stratum, in which half of the plants were used in the sample. It therefore, is, given a weight of 2. Establishment B, on the other hand, was taken with certainty (or a probability of 1) and is thus given a weight of 1. Establishment C was taken from a group where a fourth of all plants were used in the sample, and hence is given a weight of 4. The following calculations are made in estimating average earnings for a given occupation. Workers in occupation in sample establish ments at specified rate Average Estimates of total hourly in stratum Total earn- ______________________ Establishment Weight number ings Workers earnings A B ______ ___ 2 1 C_____ ___ 4 Estimated universe 40 30 20 10 $1.60 1.70 1.95 1.65 2X40 1X30 1X20 4X10 ..... 170 2 X 40 X $1.60 1X 30X 1.70 1X 20X 1.95 4X 10X 1.65 $284.00 The estimated average hourly earning is thus $284.00 or $1.67. 170 A similar method applies to any character istic estimated from the sample. To estimate the proportion of employees in establishments granting paid vacations of 2 weeks after 2 years of service, for instance, the establish ments are classified according to the length of vacation granted after 2 years’ service, estab lishment weights are applied to employment, as in the previous example, and the proportion of the estimated employment in the 2-week category of the estimated total employment then is computed. Using the same three estab lishments as in the previous example, this can be illustrated as follows: Actual total Weighted establishment em ployment Establishment Weight employment 200 100 A _____________ 2 500 500 1 B _________ 300 75 C_____________ 4 1,000 Estimated universe Vacation provisions after 2 years 1 week. 2 weeks. 1 week. Thus, the estimated percentage of workers in establishments granting 2 weeks’ vacation 500 or 50 percent. after 2 years of service is 1,000 When a large establishment within survey scope, for which no substitute exists, is unable to supply data, the deficiency is alleviated by increasing the weight of the most nearly sim ilar units. Should any segment be affected by a substantial amount of such noncooperation, the publication of materials will be diminished by omitting separate presentation of sectors seriously affected. Where a sample of selected metropolitan areas is used to represent the totality of such areas, a second stage of weighting is used to expand the individual area totals to region and/or national estimates. Since, as indicated in the description of the sampling method, each area represents a stratum of similar areas, the total from each area are weighted to the esti mated stratum totals by multiplying by the inverse of the chance of selection. This pro cedure provides the ratio of nonagricultural employment in the stratum to that in the OCCUPATIONAL PAY AND SUPPLEMENTARY BENEFITS sample area (one in the case of the large selfrepresenting areas). Summing all such esti mated stratum totals yields the earnings and employment totals for the region and the country as a whole. Analysis and Presentation Where an industry survey is designed to yield estimates for selected States or areas, these are published separately as information becomes available from all sample firms in the State or area unit. Industry surveys limited to selected areas do not provide a basis for the examinations of pay levels by size of com munity, size of establishment, product, or labor-management agreement coverage that generally are included in bulletin reports on nationwide surveys. Regardless of geographic scope, industry survey reports record the inci dence of incentive pay plans and, to the extent possible, average pay levels separately for time and incentive workers. Individual bulletin reports on individual area wage surveys are supplemented by two sum mary bulletins. The first compiles the results of individual area surveys made during a fiscal year. The second contains information on oc cupational earnings, employer practices, and suplementary wage benefits for all metropoli tan areas combined and by industry division within the four broad census regions. Wage-rate indexes are constructed for broad occupational categories, e.g., office clerical workers, skilled maintenance workers, and un skilled plant workers. These indexes are pub lished annually, separately for all industries and manufacturing, for each metropolitan area studied, and for all metropolitan areas com bined and by region. Area pay relatives for the three occupational categories are published annually, permitting ready comparisons of average pay levels among areas. Estimates of labor-management agreement coverage are also presented annually. Occupational pay relation ships within individual establishments are summarized periodically. Bulletin reports on the National Survey of Professional, Administrative, Technical, and 133 Clerical Pay present occupational averages and distributions on an all-industry basis, nation wide and separately for all metropolitan areas combined, and for establishments employing 2,500 workers or more. Average pay levels for industry divisions are shown as percentages of the all-industry averages. Year-to-year per centage changes for occupation-work levels and trend estimates for occupations are reported. Industry and area wage survey reports are issued throughout the year as the surveys are completed. The bulletin on the National Survey of Professional, Admistrative, Technical, and Clerical Pay is made available in March. Summaries of the data in the bulletins and special analyses appear also in the Monthly Labor Review. Uses and Limitations Occupational wage data developed in these surveys have a variety of uses. They are used by Federal, State, and local agencies in wage and salary administration and in the formula tion of public policy on wages, as in minimum wage legislation. They are of value to Federal and State mediation and conciliation services and to State unemployment compensation agen cies in judging the suitability of job offers. Knowledge of levels and trends of pay rates by occupation, industry, locality, and region is re quired in the analysis of current economic de velopments and in studies relating to wage dispersion and differentials. Bureau data are used in connection with private wage or salary determinations by em ployers or through the collective bargaining process. To the extent that wages are a factor, survey data also are considered by employers in the selection of location for new facilities and in cost estimating related to contract work. Occupational wage survey programs are not designed to supply mechanical answers to questions of pay policy. As suggested earlier, limitations are imposed in the selection and definition of industries, of geographic units for which estimates are developed, of occupations and associated items studied, and in determina tion of periodicity and timing of particular 134 BLS HANDBOOK OF METHODS surveys. Depending upon his needs, the user may find it necessary to interpolate for occupa tions or areas missing from the survey on the basis of knowledge of pay relationships. Because of interestablishment variation in the proportion of workers in the jobs studied and in the general level of pay, the survey averages do not necessarily reflect either the absolute or relative relationships found in the majority of establishments. To illustrate, em ployment in the specialized maintenance crafts tends to be concentrated in the larger establish ments, whereas employment in custodial and material-movement jobs is distributed more widely within an industry or area. Thus, to the extent that pay rates in the larger establish ments vary from the average level, the skill differential measure based on the survey aver ages will differ to some degree from that obtainable within each of the larger estab lishments. The incidence of incentive methods of pay ment may vary greatly among the occupations and establishments studied. Since hourly aver ages for incentive workers generally exceed those for hourly-rated workers in the same job, averages for some incentive-paid jobs may equal or exceed averages for jobs positioned higher on a job evaluation basis but normally paid on a time basis. Wherever possible, data are shown separately for time workers and incentive workers in the industry surveys. In centive plans (generally plant-wide in applica tion) apply to only a very small proportion of the workers in the indirect plant jobs studied in the area wage program. Although year-to-year changes in averages for a job or job group primarily reflect general wage and salary changes or merit increases received by individuals, these averages also may be affected by changes in the labor force resulting from labor turnover, labor force ex pansions and reductions for other reasons, as well as changes in the proportion of workers employed in establishments with different pay levels. A labor force expansion might increase the proportion of lower paid workers and thereby lower the average, or the closing of a relatively high-paying establishment could cause average earnings in the area to drop. Reliability of surveys. Results of the surveys generally will be subject to sampling error. This error will not be uniform, since, for most occupations, the dispersion of earnings among establishments and frequency of occurrence of the occupation differ. In general, the sample is designed so that the chances are 9 out of 10 that the published average does not differ by more than 5 percent from the average that would be obtained by enumeration of all estab lishments in the universe. That error applies to the smallest breakdown published. Hence, the error present in broader groupings will be somewhat less. The sampling error of the percentage of workers receiving any given supplementary benefit differs with the size of the percentage. However, the error is such that rankings of predominant practices almost always will ap pear in their true position. Small percentages may be subject to considerable error, but will always remain in the same scale of magnitude. For instance, the proportion of employees in establishments providing more that 4 weeks’ paid vacation to long-service employees may be given as 2 percent, when the true percentage for all establishments might be only 1 percent. Such a sampling error, while considerable, does not affect the essential inference that the prac tice is a rare one. Estimates of the number of workers in a given occupation are subject to considerable sampling error, due to the wide variation among establishments in the proportion of workers found in individual occupations. (It is not unusual to find these estimates subject to sampling error of as much as 20 percent.) Hence, the estimated number of workers can be interpreted only as a rough measure of the relative importance of various occupations. The greatest degree of accuracy in these em ployment counts is for those occupations found principally in large establishments. This sam pling error, however, does not materially affect the accuracy of the average earnings shown for the occupations. The estimate of average earn ings is technically known as a “ ratio estimate,” i.e., it is the ratio of total earnings (not pay rolls) to total employment in the occupation. Since these two variables are highly correlated OCCUPATIONAL PAY AND SUPPLEMENTARY BENEFITS (i.e., the errors tend to be in the same direc tion), the sampling error of the estimate (aver age hourly earnings) is considerably smaller than the sampling error of either total earnings or total employment. Since completely current and accurate infor mation regarding establishment products and the creation of new establishments is not avail able, the universe from which the sample is drawn may be incomplete. Sample firms in correctly classified are accounted for in the actual field work, and the universe estimates are revised accordingly. Those firms which should have been included but were classified erroneously in other industries cannot be ac counted for. Since some measure of subjective judgment enters into the classification of occupations and other characteristics, there is some reporting variability in the results. A repetition of the survey in any establishment with different in terviewers and respondents would undoubtedly produce slightly different results. However, when spread over a large number of establish ments the differences, being random, would tend to balance out. Hence, analyses based on a small number of respondents must be used with care, even when all eligible establishments are included. No evidence of any consistent error has been uncovered. Technical References Number 1. Cohen, Samuel E., “ Studies of Occupational Wages and Supplementary Benefits.” Monthly Labor Review, March 1954 (pp. 292-297). An eariler description of the methods of wage surveys, similar to the present article. 2. Douty, H. M., “ Survey Methods and Wage Comparisons.” Labor Law Journal, April 1964 (pp. 222-230). A discussion of the uses of wage survey results, and the pitfalls to be avoided. A short discussion of the factors affecting survey methods is also included. 3. Kanninen, Toivo P., “ New Dimensions in BLS Wage Survey Work.” Monthly Labor Review,October 1959 (pp. 1081-1084). An outline of the occupational wage survey programs, as expanded in fiscal 1960. Lists the type of survey and cycle for each of 70 industries studied separately, and identifies the area sample as originally determined for the labor market survey program. 135 — G eo r g e L . S t e l l u t o Chapter 15. Employee Earnings and Hours Frequency Distributions Background An extensive program of studies of the fre quency distribution of employee earnings has been maintained by the Bureau of Labor Sta tistics since 1954. Although the need for such data had long been recognized and some work had been done in the area, the program did not receive its full impetus until 1955 when Congress, in an amendment to section 4(d ) of the Fair Labor Standards Act, directed the Secretary of Labor to include in his annual report an evaluation and appraisal of minimum wages established by the act and his legislative recommendations.1 To meet these requirements, the program of employee earnings distributions studies was established. Since 1964, frequency distributions of weekly hours of work have been a part of this program. This program was initiated to provide knowl edge about the internal structure of earnings and hours which an average does not reveal. For example, distribution studies show whether earnings or hours are dispersed evenly or tend to be clustered around one or more points; they show the gap between the highest and lowest paid workers; they reveal the proportions of employees working short or long hours, and similar information on individual earnings and hours. They serve to locate the employed “ poor,” i.e., those not sharing in the general rise in living standards available from work. Description Studies of employee earnings and hours dis tributions generally include within their scope all nonsupervisory employees. No attempt is made to classify employees by occupation, al though at times information is collected sep arately for some types of employees. The data collected relate to straight-time hourly eam1 The Bureau’s studies conducted in this area before 1954 frequently related to selected occupations, or were limited to relatively narrowly defined industries. 2 See technical references for a listing of publications con taining full descriptions of both types or studies. 1 See appendix B. * See appendix C. ings, excluding premium pay for overtime work and for work on weekends, holidays, and late shifts. Incentive pay, production bonuses, and cost-of-living payments are included in earnings, but nonproduction bonuses (e.g., Christmas bonuses) are not. Hours informa tion is collected for hours worked during the week and for hours spent on vacations, holi days, or sick leave for which pay is received. Earnings and hours information is obtained for each employee, making it possible to calculate average hourly earnings and weekly hours of work for the entire group as well as to show the distribution of employees by earnings, hours, or both. Two basic types of distributions studies are conducted— industry and area.1 Industry stud 2 ies may have broad coverage, such as manu facturing or wholesale trade, or they may be limited to specific industries, such as motor carriers. While these studies usually are con ducted on a nationwide basis, they sometimes (especially in the case of specific industry stud ies) are limited to one or several areas in which the industry is concentrated. Industries are defined on the basis of descriptions in the cur rent Standard Industrial Classification Manual, published by the Office of Management and Budget.3 Area studies are limited in geographic cover age, perhaps to a Standard Metropolitan Sta tistical Area,4 to a county or a group of contiguous counties, or to a region or part of a region. This type of study usually is con ducted on a cross-industry basis— that is, earn ings and hours data are collected for most industry divisions, including mining; manu facturing; transportation (except railroads), communications and other public utilities; wholesale and retail trade; finance, insurance, and real estate; and services. Minor exceptions may be made. Data Sources and Collection Methods Data generally are obtained from employer payroll records, using one of two collection methods. Establishments may be solicited for 187 138 BLS HANDBOOK OF METHODS information by mail or by personal visit. Those establishments contacted by mail receive a questionnaire form and a note requesting co operation in the survey and explaining the nature and purpose of the study. Typically, the questionnaire requests information on the prod uct or service of the establishment and number of employees, as well as other establishment information which may be pertinent to the survey. The respondent is requested to report each employee’s earnings and weekly hours worked— the primary focus of the survey. Pro vision is made for the reporting of earnings on an hourly basis or on a salary or incentive basis. In the last two instances, the respondent reports the number of hours paid for during the salary or incentive period, thus permitting the computation of average hourly earnings for employees paid on other than an hourly basis. Certain large establishments, or companies from which data for several establishments are requested, are visited personally by the Bu reau’s field economists who may prepare the data for the survey from company records, or arrange with the company for the completion of the questionnaire form. In addition, a sam ple of establishments which do not reply to the mail questionnaire is visited by Bureau field economists to obtain the desired information. To limit errors caused by incorrect reporting by the respondent, questionnaires received by mail are reviewed for reasonableness and con sistency. Where data are questionable, a letter is written to the respondent asking him to re view the item to assure its accuracy. Sampling Earnings and hours distribution studies are conducted on the basis of a sample of all es tablishments within the scope of the survey. The sample generally is derived from State Unemployment Insurance (UI) listings which show reporting units with four employees or more by location, number of employees, and industry classification. In industries where establishments with fewer than 4 employees are of numerical importance (e.g., retail trade) the UI lists may be augmented by sources such as other government agencies, or trade directories. The size of the sample depends on several factors, among which are the size of the uni verse, the distribution of establishments by number of employees, the relative dispersion of earnings among establishments, the degree of accuracy required, and the cost of obtaining the data. Estimates of variance based on data from previous surveys also may be used in de termining the appropriate size of the sample. The sample usually is selected using a highly stratified probability sampling design. Estab lishments are first grouped, or stratified, ac cording to industry, geographic location, and employment size. Establishments in specific industries or areas for which earnings and hours data are to be presented separately are grouped independently of establishments in other strata, and sampled separately. In order to obtain maximum accuracy per unit cost, the number of establishments in the sample is distributed among the various strata in the most efficient manner, in accordance with the principles of optimum allocation. A fraction of establishments in each stratum is included in the sample, with the sampling fraction diminishing as the employment size of the strata decreases, so that the probability of in clusion in the sample is greater for the large than for the small establishment. Frequently the entire stratum containing the largest estab lishments is included in the sample. The following example shows a hypothetical universe of 47 establishments in three employ ment-size groups. By applying the appropriate sampling ratio to each size group (stratum), the number of sample establishments is de termined. Employment size group Under 25 _____ 25-49 __________ 50 and over Number of establishments in universe 32 12 3 Sampling ratio 1/4 1/3 1/1 Number of sample establishments 8 4 3 Weight of each sample establishment 4 3 1 No assumption is made that the earnings and hours structures of establishments not respond ing to the mail questionnaire are similar to those of establishments which do respond. 139 EMPLOYEE EARNINGS AND HOURS Therefore, a sample is taken of the non respondents following a procedure similar to the one just described. Establishments in this subsample are visited by Bureau field repre sentatives in order to obtain the required data. Data are not always obtained for every es tablishment in the original sample. Generally, approximately 60 to 70 percent of the sample establishments supply usable data to the survey. Estimating Procedure Although a greater proportion of large than of small establishments is included in the sam ple, any possible bias which might result from this difference is avoided by means of the esti mating procedure. Each establishment in the sample is assigned a weight which is the re ciprocal of the sampling ratio in the stratum from which it was selected. That is, an estab lishment selected from a stratum in which a sampling ratio of 1 out of 4 is used is assigned a weight of 4, so that it represents itself and three other establishments in the stratum (see previous example). Data for each establish ment are multiplied by the weight assigned to the establishment. Thus, all establishments, regardless of their size, are represented ap propriately in the final estimates. An establishment in the subsample of non respondents is weighted to represent all nonrespondents in the stratum. It is assigned a new weight— the product of the original weight and the inverse of the subsampling frac tion. Thus, if a third were subsampled of a group originally sampled at the rate of 1 out of 2, the weight of 6 would be assigned. In the case of an establishment included in the sam ple with certainty, another establishment which is similar to the nonrespondent would be weighted to represent it. In industry surveys, estimated employment totals derived from the weighting process are further adjusted to the employment levels for the payroll period studied, as reported in the Bureau of Labor Statistics monthly establish ment employment series.5 This adjustment is 5 See chapter 2, “Employment, Hours, and Earnings” . necessary to reduce the hazards of sampling, and because the State UI listings, which con stitute the universe or a large part thereof, are prepared prior to the time of the survey and thus do not account for establishments opened or closed between the compilation of the lists and the date of the survey. Estimated average hourly earnings or weekly hours of work are the arithmetic mean of weighted individual employees’ earnings or hours. Generally, they are derived by totaling weighted individual hourly earnings or weekly hours and dividing the sum by the weighted number of employees in the group. However, in industries such as retail trade, in which hourly earnings frequently vary by weekly hours of work and in which the length of the workweek covers a broad range among employees, the most representative group average hourly earnings figure is considered to be the quotient of total individual weekly earnings divided by total individual weekly hours worked. The following example illustrates the more common method of estimating group average hourly earnings. Referring to the sampling scheme described in the preceding example, as sume that the eight sample establishments in the under 25 size group had 30 employees, each earning $1.50 an hour, 20 earning $2 and 15 earning $3, and that the 12 establishments in the 25-49 size group and the 3 in the 50- and over-size group had employees with earnings as shown. Making the calculations that follow, estimated group average hourly earnings are obtained. Weight X 4 3 1 employm ent= 30 20 15 50 40 30 150 100 50 Weighted employm entx 120 80 60 150 120 90 150 100 50 920 Hourly earnings= Weighted group hourly earnings $1.50 2.00 3.00 1.50 2.50 3.00 2.00 2.50 3.00 $180.00 160.00 180.00 225.00 300.00 270.00 300.00 250.00 150.00 2,015.00 Estimated group —Total weighted group hourly earnings = $2,0 15 .0 0 average Total weighted group employm ent 920 hourly earnings ^ 140 BLS HANDBOOK OF METHODS If the average is to be computed following the second procedure (such as would be done in retail trade studies), employment would be re placed by weekly hours in the example and the remainder of the calculations carried through. In this case, the ratio now shown in the ex ample would be replaced by the ratio of total weekly earnings to total weekly hours. a description of the area or industry studied and provides a summary and analysis of the survey results. Important relationships and dif ferences among areas and industries are high lighted and, where data are available from earlier surveys, period-to-period changes in earnings and hours are discussed. In addition, a detailed description of the scope and method of survey is included in the report. Analysis and Presentation The Bureau generally issues a bulletin on the results of each survey of employee earnings and hours distributions. The report contains a description and analysis of the survey results as well as tabulations of data collected. In industry studies, tabulations are pre sented for the entire industry, and frequently they are also presented for important segments of the industry. For example, in a survey of manufacturing, in addition to data for the en tire industry division, data also might be pre sented for most or all of the major groups (2-digit SIC codes) for several groups (3-digit SIC codes), and perhaps for some individual in dustries (4-digit SIC codes). In addition to na tionwide presentation, data for all or some of the industry segments generally are shown separately on a regional basis, and frequently on a metropolitan-nonmetropolitan area basis. In area studies, which include most industry divisions, data obtained are tabulated for all the industries in the area combined, and then separately for manufacturing and nonmanu facturing industries. Where sufficient data are available, tabulations are presented for selected industry groups as well. Earnings are tabulated to show the distribu tion of employees by intervals of 5- and 10-cent average hourly earnings. Hours are tabulated to show the distribution of employees by weekly hours of work. The total number of employees, their average hourly earnings and average weekly hours, of course, also are shown. Data also may be presented to show average weekly earnings and cross-tabulations of average hourly earnings by weekly hours of work. The text accompanying the tabulations gives Uses and Limitations The design of these employee earnings and hours distributions studies makes them par ticularly useful in the analysis of Federal minimum wage and maximum hours legislation — in analyzing the effects of legislation, in con sidering new legislation, and in formulating wage and hours policy. The information is used by the executive and legislative branches of government, organized labor, business,, aca demicians, etc. Special tabulations are prepared for the U.S. Department of Labor’s Employ ment Standards Administration, for use in the Secretary of Labor’s annual report to the Con gress required under section 4(d) of the Fair Labor Standards Act. In addition, the data are used in collective bargaining, wage setting, broad economic analysis of labor area and in dustry wage structures, comparisons of wage levels in various parts of the country, and to show trends in employment, earings, and hours. Employment estimates are subject to some error, and in smaller groups this error may be relatively large. Therefore, these estimates fre quently are provided to serve only as general guides to the size of the labor force included in the survey and as reference points from which to measure the direction and general magnitude of employment changes. The average earnings reported are straighttime, and any premium pay for overtime or late shift work is not reflected. Similarly, dif ferences in prevailing supplementary compen sation practices ( “ fringe benefits” ) among establishments, industries, and areas are not considered. EMPLOYEE EARNINGS AND HOURS BLS 3010b O ffic e o f Managem ent and Budget No. 4 4 -S -7 1 0 0 6 Approval expires 1 2 -31 -7 1 U.S. DEPARTMENT OF LABOR B U R E A U O F L A B O R S T A T IS T IC S W A S H IN G T O N , D.C. 20212 Report Number * r Y our report will be held in confidence n Keep this copy for your company tile. T h e d a ta , e x c e p t fo r Ite m 1 w h ic h r e la te s t o th e e n t ir e c o m p a n y , s h o u l d c o v e r t h e e s t a b l i s h m e n t s ) in t h e l o c a t i o n d e s i g n a t e d t o t h e le f t . l_ SURVEY OF INDIVIDUAL HOURS AND EARNINGS OF NONSUPERVISORY EMPLOYEES (Check appropriate box.) (2) (O ANNUAL GROSS SALES OR RECEIPTS FOR THE COMPANY, ENTERPRISE, OR INSTITUTION 1 (3) 1 1 1 1 ----------------1 Under $250,000 to $1,000,000 or more $250,000 $1,000,000 Check the box which indicates the annual gross volume of sales or receipts (exclusive of excise taxes) from all related activities of the company, enterprise or institution, including receipts from the establishment(s) covered by this report. Use the last calendar or fiscal year. 2. ESTABLISHMENT INFORMATION: Please enter the information requested in the columns below for each separate establishment covered by this report. Each place of business, warehouse, or central office in a separate location is considered a separate establishment for the purpose of this survey. (a) Location: Identify each establishment by its street address and city. (b) Type o f Activity: Describe the activity of each establishment listed, for example, gas station; drug store; department store; women’s apparel store; shoe store; dime store; hotel; motel; dance hall; bowling alley; race track; amusement park; movie theatre; etc. (c) Employment: Report all full-time, part-time, seasonal, and casual employees who received pay for any part of the payroll period including May 12, 1971. Include those paid for sick leave, holidays, vacations, etc. DO NOT INCLUDE proprietors, m em bers o f u n in co rp o rated firm s, o r unpaid fam ily workers. Total - Enter total number of employees. Include all classes of employees— executive, administrative and professional, as well as nonsupervisory employees. Nonsupervisorv - Enter number o f workers below the supervisory level. Include employees such as inside salespersons, shipping and receiving clerks, routemen, laborers, warehousemen, repairmen, installers, room clerks, waiters and waitresses, nurses, office clerks, janitors, watchmen, etc. DO NOT INCLUDE driver salesmen, outside salesmen, executive, professional, and supervisory personnel. (d) Collective Bargaining: Are a majority of nonsupervisory employees covered by union-management agreements? (Check appro- (e) Annual Gross Sales or Receipts fo r the Establishment: Check the column which indicates the annual gross volume of sales or priate column). receipts (exclusive of excise taxes). (b ) (a ) (c) E m p lo y m e n t f o r p a y r o ll p e r io d in c lu d in g L o ca tio n M ay 12, T y p e o f A c t iv it y (s tre e t a d d ress N on s u p e rv iso ry s a le s A g reem en t ( A r e m a jo r it y ( w e r e la s t y e a r ’s s a le s covered ?) $ 2 5 0 ,0 0 0 o r m o r e ? ) 1971 a n d c ity ) T o ta l (d ) (e ) U n io n -m a n a g e m e n t G r o s s e s ta b lis h m e n t Yes No Y es No 0 ) (2 ) to (2 ) (f ) Fo r BL S On iy i 2 3 4 5 6 7 8 9 (0 How many hours per day and days per week are the majority o f your nonsupervisory employees scheduled to work? (g) After how many hours per day would your nonsupervisory employees receive overtime premium pay? A fter______hours per day or. I I we do not pay daily overtime. Hours per day | [ Days per week [ | Check here if you want a copy o f the Bureau’s report on this survey. I E n te r h o u r s t o o n e d e c im al, e.g., 7 .5 I Name and title of person furnishing data (Please Type or P rint) Area code, Phone No. 142 BLS HANDBOOK OF METHODS Page 2 3. HOURS AND EARNINGS OF NONSUPERVISORY EMPLOYEES Definitions and Instructions (See also the examples below) Hours worked and paid for and straight-time earnings of all nonsupervisory employees are requested for the payroll period (weekly, biweekly, semimonthly, or monthly) which includes May 12, 1971. Please report and identify the data separately for each of the locations listed in item 2(a). • Nonsupervisory Employees are defined in item 2(c) on page 1 of this questionnaire. Remember to EXCLUDE outside sales men, driver salesmen, executive, administrative, professional, and supervisory personnel. • Hours should include all hours worked and hours paid for sick leave, holidays, vacations, etc. Also include the number of over time hours actually worked. • Straight-time Earnings should relate to the hours worked and paid for, including straight-time pay for overtime, holidays, week ends, and late shifts. Also include commission and bonus earnings. Exclude, however, all premium payments. (For example, if overtime is paid at time and one-half report only two-thirds of this pay). • These data are requested separately fo r each nonsupervisory employee. However, one entry may be made if two or more emplovees worked identical hours and received identical straight-time pay (see example A below). To avoid correspondence, please do not report aggregate hours and earnings for several employees. • Detailed instructions and examples for reporting the necessary data in each column are listed below. Column (2 ) - Use a separate line for each employee and enter “ 1,” unless two or more employees worked the same number of hours and received identical hourly or salary rates (see example A). Please check to see that the sum of the employees ALL NONSUPERVISORY EMPLOYEES, COMPLETE COLUMNS (2) and (3) in this column equals the number of nonsupervisory employees reported in item 2(c) on page 1. Column (3 ) - Enter the number of hours paid for during the week of .May 9-15, 1971 regardless of the length of the payroll period. Include hours paid for sick leave, holidays, vacations, etc. Count the number of overtime hours actually worked. (Not hours “ boosted” to reflect premium pay.) EMPLOYEES FOR WHOM STRAIGHT-TIME HOURLY EARNINGS ARE AVAILABLE, USE COLUMN (4) Column (4) - Enter the average straight-time hourly earnings. Exclude premium payments for overtime and for weekend, holiday, and late shift work. (You should use this column to report earnings of employees even when paid on a salary basis if average straight-time hourly earnings are available). If the hourly earnings exclude commissions or bonuses, also complete columns 7 and 8 (see example D). Column (5 ) - Enter the straight-time salary for the salary period (weekly, bi- EMPLOYEES PAID ON A SALARY BASIS (IF HOURLY EARNINGS ARE NOT AVAILABLE), USE COLUMNS (5) and (6) EMPLOYEES RECEIVING COMMISSION AND/OR BONUS PAY, USE COLUMNS (7) and (8) weekly, semimonthly or monthly) which includes May 12, 1971. Include straight-time pay for overtime but exclude overtime premium. Do not include “draws” against commission as salary. For employeespaid a commission or bonus in addition to a salary, also complete columns 7 and 8 (see example E). Column (6 ) - Enter the number of hours paid for during the payroll period for which earnings were reported in column 5 to permit accurate calculation of aver age straight-time hourly earnings. Column (7 ) - Enter for each employee the total commission and/or bonus pay, including “PM’s,” “Stims,” or any special bonuses based on sales paid quarterly or oftener. This pay is to be reported for the commission or bonus period includ ing May 12, 1971. If the commissions earned during that pay period are not representative of normal commission earnings, a longer period may be used. If employees receive both commission and bonus payments for an identical period of time, report the combined figure (see example D). If bonus payments cover a period longer than the commission period, add only the prorated am ount of the bonus to the commission earnings that correspond to the commission period (see example E). Column (8) - Enter the number of hours worked during the commission or bonus period. (The hours should refer to the total hours worked during the period and not necessarily only to those hours during which commissions or bonuses were earned.) For employees paid an hourly rate or salary in addition to commissions or bonuses, it is also necessary to complete column 4, or columns 5 and 6 (see examples D and E). EXAMPLES (See illustrations on next page) A. Two employees each worked 36 3/4 hours during the survey week and each was paid a straight-time hourly rate of $2.20. B. One employee, who worked 40.0 hours during the survey week, received a salary of $168.50 (exclusive of any premium pay) for 88.0 hours worked during the salary period (!4 month). C. One employee, who worked 32.5 hours during the survey week, was paid on a straight commission basis and received $361.20 for 168 hours worked during the commission period (1 month). D. One employee, who worked 40.0 hours during the survey week, was paid a straight-time hourly rate of $1.95 and also received $35.00 in commissions and $7.50 in “PMY" for 173.6 hours worked during the commission period (1 month). E. One employee, who worked 37.5 hours during the survey week, was paid a weekly salary of 575.00, received commissions of $102.00 for 162.0 hours worked during a 1-month period, and $150.00 in bonuses earned during a 3-month period. Only 1/3 of the bonus ($50.00) is reported so that the bonus period corresponds to the commission period. 143 EMPLOYEE EARNINGS AND HOURS 3. HOURS AND E A R N IN G S O F N O N S U P E R V IS O R Y C o m p le te th e s e co lu m n s for a ll n o n su p e rv isory e m p lo y e e s EM PLOYEES U s e th is colum n if U s e th e s e co lu m n s for n o n su p e rv isory e m p lo y e e s paid hourly e a rn in gs other than on an hourly b a s is . a v a ila b le Num ber L in e N o. of e m p lo y e e s H ou rs p a id for during the w eek o f M ay 9 15, S alary C o m m is s io n s a n d /o r B o n u s S tra ig h t-tim e s a la r y for 1971 during in c lu d in g M a y 1 2 , 1971 hourly ea rn in gs H ou rs p a id for sa la ry p eriod S traigh t-tim e s a la r y period T o ta l H ou rs p a id for co m m is s io n s a n d /o r co m m issio n during b on us pay p eriod ( N e a r e s t tenth) (i) (N e a r e s t c en t) ( N e a r e s t cen t) (N e a r e s t tenth) ( N e a r e s t c en t) (N e a r e s t tenth) (3) (4) (5) (6 ) (7 ) (8) (2 ) I llu s tr a tio n s o f e x a m p le s on p a g e 2. A. 2 3 6 .8 B. 1 4 0 .0 C. 1 3 2 .5 D. 1 4 0 .0 E. 1 3 7 .5 $ 2 .2 0 $ $ 1 6 8 .5 0 8 8 .0 3 6 1 .2 0 7 5 .0 0 37. 5 1 6 8 .0 4 2 .5 0 1 .9 5 17 3 .6 1 5 2 .0 0 1 6 2 .0 DATA FOR EACH ESTABLISHMENT SHOULD BE REPORTED SEPARATELY AND THE ESTABLISHMENT IDENTIFIED. 01 02 03 04 05 06 07 08 09 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 36 144 BLS HANDBOOK OF METHODS 3. HOURS AND EARNINGS OF NONSUPERVISORY EMPLOYEES C o m p le te th e s e co lu m n s for a ll n o n su p e rv isory e m p lo y e e s U s e th is colum n if U s e th e s e co lu m n s for n o n su p e rv iso r y e m p lo y e e s p a id h ourly e a rn in gs oth er than on an h ourly b a s i s . a v a ila b le Salary H ou rs p a id for Number during the of L in e N o. w eek o f e m p lo y e e s M ay 15, 12- (i) 01 02 03 04 05 06 07 08 09 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 37 38 39 40 41 42 (2 ) s a la r y for during in c lu d in g s a la r y p eriod M ay $ $ H ou rs p a id fo r c o m m is s io n s during a n d /o r c o m m issio n b o n u s pay (6 ) $ ( N e a r e s t tenth) (7 ) ( N e a r e s t tenth) pe rio d ( N e a r e s t cen t) 1971 (5) (4 ) (3 ) 12, ( N e a r e s t c en t) ( N e a r e s t c en t) T o ta l H ou rs p a id fo r s a la r y period S tra ig h t-tim e hourly e a rn in gs 1971 ( N e a r e s t ten th ) C o m m is s io n s a n d /o r B o n u s Stra ig h t-tim e (8) 145 EMPLOYEE EARNINGS AND HOURS Technical References Number 1. U.S. Department of Labor, Bureau of Labor Statistics: Employee Earnings and Hours in Eight Metropolitan Areas of the South, June 1965 (Bulletin 1533, 1966). 2 . ______Employee Earnings and Hours in Retail Trade, June 1966 (Bulletin 1584, 1968). 3 . ______Earnings in Wholesale Trade, June 1958, (Bulletin 1253, 1959). 4 . ______ Factory Workers’ Earnings, May 1958 (Bulletin 1252, 1959). 5 . ______ Employee Earnings in Nonmetropolitan Areas of the South and North Central Re gions, June 1962 (Bulletin 1416, 1964). 6 . ______Employee Earnings in Retail Trade, June 1962 (Bulletin 1380, 1963). Each of these bulletins contains a detailed description of the method used in the survey. — A l v in B a u m a n Chapter 16. Union Wage Rates Background and Description o f Survey Annual studies of union wage rates and hours are conducted in four industries: build ing construction, local transit, local trucking, and printing.1 Union wage rates and hours are those agreed on through collective bargaining between employers and trade unions; they are defined as (1) the basic (minimum) wage rates (excluding holiday, vacation, or other benefit payments regularly made or credited to the worker each pay period) and (2) the maximum number of hours per week at straight-time rates. Rates in excess of the negotiated min imum, which may be paid for special qualifica tions or other reasons, are excluded. The use of union agreements or other union records in studies of occupational wages is practicable in industries that are characterized by a high degree of organization and in which (1) defined craft groupings persist, as in build ing construction or printing, or (2) key oc cupations can be clearly delineated, as in local transit. The Bureau’s annual union wage studies began in 1907. Originally, information was ob tained for 39 cities, but the number was ex panded gradually until in 1948, 82 cities were covered.1 That number was reduced to 77 in 2 1949 and to 52 in 1953. The studies were ex panded again, after the 1960 Census of Popula tion, to the present coverage of 68 cities having 100,000 inhabitants or more. The scope of the information for individual industries has also been expanded. For example, 24 journeymen crafts and nine helper and laborer classifica tions in the building trades are covered current ly, in place of the 13 journeymen and 7 helper and laborer classifications in the initial studies. The study of union wage rates and hours in the building trades includes virtually all journeymen and helper and laborer classifica tions. Indexes and other data are shown for each important trade as well as for all trades combined.3 The trucking study embraces drivers and helpers engaged in local trucking. Over-theroad drivers and local city drivers paid on a mileage or commission basis are excluded. All data, including indexes, are presented for the two classifications indicated. Union wages and hours in the local-transit industry are limited to operating employees. Data are shown separately for operators of surface cars and buses, and elevated and sub way lines, except that indexes are shown only for the industry as a whole. In the printing industry, 12 book and job trades, 8 newspaper trades and 6 lithography trades are studied, and for the newspaper trades, separate data are shown for day and nightwork. Indexes and other data are pre sented separately, by type of printing, (except lithography) for each trade and for all trades combined. Data Sources and Collection Methods The union wage studies are designed to in clude all local unions in the covered industries in the selected cities. Periodic checks are made with central labor unions, district councils, and other authoritative bodies to identify new local unions that should be included in the studies. Information is collected by mail from local unions and when necessary from international unions and regional union organizations. Per sonal visits are made to unions that do not re spond to the mail questionnaire. Before 1947, all data relative to union wage studies were collected directly from local union officials (generally the secretaries or business agents) by Bureau representatives and entered on forms designed specifically for this purpose. 1 The coverage at various times also included barbers, line men, longshoremen, and workers engaged in breweries, laundries, metal trades, millwork, restaurants, soft-drink production, theaters, and bakeries. The Bureau plans to conduct the first in a series of biennial studies of union wage rates for grocery store employees in July 1971. 2 In these studies, data relate to individual cities and con tiguous suburban areas, rather than to the much broader Standard Metropolitan Statistical Areas which are used in most other Bureau surveys. 3 In addition to the annual studies in the building trades, a quarterly survey of 7 major construction trades is conducted in 104 cities. Estimated average hourly wage rates for all trades combined and for each surveyed trade are presented, together with the estimated change during the quarter and the year. 147 148 BLS HANDBOOK OF METHODS Information requested relates to July 1 for all industries. This date was adopted, after numerous changes, because most new agree ments in these industries have been negotiated by that time each year. In order to maintain year-to-year comparability, wage rate, hours, and membership data for the previous year are transcribed onto the forms before they are sent out. Union officials are requested to check the previous year’s data and revise any figures which may have been incorrectly reported, and to insert current data. Copies of union agree ments also are requested from union officials for the purpose of (1) checking the data en tered on the schedules with the terms of the agreements, and (2) building up the files of union agreements maintained by the Bureau of Labor Statistics.4 The reporting form used for the building trades survey is reproduced on pages 149-151. Sampling and Estimating Procedures The current series is designed to reflect union wage rates and hours in all cities of 100,000 inhabitants or more, excluding Honolulu. All cities of 500,000 inhabitants or more are in cluded, as are most cities in the 250,000 to 500.000 group. The cities in the 100,000 to 250.000 group selected for study are distributed widely throughout the United States. Data for some of the cities included in the study are weighted to compensate for cities not surveyed. To provide appropriate representation in the combination of data, each region is considered separately when city weights are assigned.5 Rates An over-all average hourly rate is computed for each of the industries included in the union wage studies. In addition, averages are pre sented by industry branch, trade, city, and region in building construction and printing; and by city and region, in local transit and local trucking. Average union rates are calculated by weighting each quotation for the current year by the reported membership.6 These averages are levels designed to provide comparisons among trades and cities at a given time. They do not measure the trend of union rates, the function served by the index series. Indexes Chain indexes are calculated for each of the four industries to portray the trend of union wage rates and weekly hours. In calculating these indexes, the percent change in aggregates is computed from quotations for all identical classifications in the industry for 2 successive years. To obtain the aggregates, the rates and hours for both the previous and current years are weighted by the membership in the par ticular classification for the current year. The index for the current year is computed by multiplying the index for the preceding year by the ratio of the aggregate change. In the 1970 study of building trades, the rate aggre gate for all quotations increased 11.6 percent over the previous year. The July 1, 1970, index of union hourly wage rates for all building trades (128.8) is the result of multiplying the July 1, 1969, index (115.4) by the ratio of the aggregates (111.6). This method of index cal culation minimizes the influence of year-to-year changes in membership. Indexes of union hourly wage rates and weekly hours are computed for each classifica tion as well as for all classifications combined in the building construction and printing in dustries. In the local trucking and local transit industries, indexes are provided only for all classifications combined. Irregular hours of work for operating employees in many of the covered cities prevent the computation of an index for union weekly hours in the local transit industry. The base period for the indexes of union wage rates and weekly hours is the 1967 aver age. The series for the building trades and printing industry date back to 1907, for local 4 See chapter 20. “ Collective Bargaining Agreements.” 5 The cities in the sample were not selected on a probability basis. Continuity of city data has taken precedence over sampling procedure during periods of contraction and expan sion in the program. 6 Reported membership, as used in this study, is defined as members working or immediately available for work. 149 UNION WAGE RATES B L S 1150.1 (R e v . 1971) B ud get B ureau N o . 4 4 -R 7 3 8 .1 2 . A p p ro v a l e x p ir e s 3 -3 1 -7 2 . U.S. DEPARTMENT OF LABOR BU R EA U OF L A B O R ST A T IST IC S r n L J Dear Sir: The Bureau of Labor Statistics is now conducting its annual survey of union wage scales and hours in the building trades. The continued success of these surveys, begun in 1907, depends largely upon your cooperation and we are asking that you furnish information requested on this schedule form The completed form should be signed and returned to the Bureau in the enclosed . self-addressed envelope which requires no postage. W would appreciate your returning the e completed form promptly. Thank you very m uch for your cooperation. Very truly yours, Regional Director. INSTRUCTIONS FOR COMPLETING THIS FORM P A Answ general inform art : er ation questions in this p of schedule. art P art B Check an revise, if necessary, inform : d ation reported for July 1, 1970. If any of you r trade classifications h ave been om itted, insert th an th negotiated scales in th appropriate em d e e colum ns. Sh ould you have m re th one negotiated scale for a trade o classification, rep o an r ort separate inform ation for each scale. For each trade or occupation group show in colum (2), n n enter in colum (5) the w n age scale in effect on July 1, 1971, and in colum (10), the n m er of n u b w eekly h rs before overtim rate is effective. The n m ou e u ber of union m bers at each scale em should b entered in colum (15), an th n m er of apprentices, as of July 1, 1971, should be e n d e u b entered in colum (17) for each trade. n Copy of A greem ent: Please provide a copy of each of you agreem r ents to assist us in o r analy u sis of collective bargaining provisions. If you have only one copy avaiblale, w shall be glad e to m e a duplicate an retu the original prom ak d rn ptly. If requested, th agreem will b kept in e ent e confidence and m aterial used only for general analysis w hich will not reveal the n e of either am party to th agreem e ent. 150 BLS HANDBOOK OF METHODS UNION SCALES OF WAGES AND HOURS IN THE BUILDING TRADES (A nnual Survey) PART A.-GENERAL INFORMATION: Please attach a copy of you agreem in effect on July lf 1971, an answ th follow questions: r ent d er e ing 1. W en did you agreem go into effect?_________________________________________________________________ h r ent h W en m it be reopened?__________________________ W en does it expire?_________________________________ h ay 2. W at is the m u n m er of h rs th can b w h axim m u b ou at e orked each day before overtim rate is effective?_______________ e 3 Does you agreem provide for a health an insurance plan (life insurance, hospitalization, m * r ent d edical, surgical, an d other sim types of health an w ilar d elfare program financed— s) Entirely by em ployer? Q Yes □ N o In p by em art ployer? □ Yes Q N o A ou of em m nt ployer contribution________________ per H r □ Shift Q W ou eek Q M th Q on (c e n ts or p e rce n t) 4. Does you agreem provide for a pension plan financed— r ent Entirely by em ployer? Q Yes □ N o In p by em art ployer? □ Yes N o ou eek Q M th Q on A ou of em m nt ployer contribution________________ .per H r Q] Shift Q W (c e n ts o f p e rce n t) 5 Does you agreem specifically provide for a paid vacition, financed by em * r ent ployer paym ents— (A) To a vacation fu d □ Yes Q N n o FU PAYM ND ENTS DEDUCTED FR M PAYROLL UPON W O RJTTEN IM , PLIED, O CONTRACTURAL AUTHOR R IZATION O W R ER ARE NOT CO SIDERED EM F OK S N PLOYER PAYM ENTS (B) To w orker each pay period, as p of negotiated scale □ Yes □ N art o (C) To w er each pay period, in addition to negotiated scale Q Yes Q N ork o (D O ) ther— (Explain)____________ ________________________________________________________________ If answ is Yes to any of above, indicate am n of em er ou t ployer payment__________________________________ per h r. ou (c e n ts or p e rce n t) 6. Does you agreem provide for em r ent ployer paym ents to oth funds, such as holiday, educational, prom er otional, u em n ploym benefits? Q Yes Q N ent o If YES, list below each type of contribution separately an am n of em d ou t ployer paym ents (cents o percent). r ow an en?_____________ H m are ow any 7. W at is th total m bership of you union?_____________ H m y are journeym h e em r apprentices?______________ H m y are helpers an laborers?_______________ ow an d 8. Betw een July 1, 1970, a d July 1, 1971, h m y apprentices com n ow an pleted their apprenticeship? (If n e enter zero.)_______________ on 9. Betw een July 1, 1970, an July 1, 1971, h m y jou eym becam unavailable for w because of death, perm d ow an rn en e ork an ent disability, o retirem r ent? (If n e enter zero.)_______________________________________________ ________ on P le ase sign your name here________________________________________ T itle _________________________ D a te ______________________ , 1971. A ddress_____________________________________________________________ C ity, State, and Z i p C o d e ___________________________________________ U. S. DEPARTMENT OF LABOR Bureau o f Labor Statistics Union Scales of Wages and Hours in the Building Trades (Annual Survey) Part B. Scale and membership information by trade: Please enter the data requested below. Do not use columns marked ‘ Code.’ Employer payments to insurance, pension, vacation, or other funds should be excluded from the hourly scales below, and shown as cents or percent of scale in the spaces pro vided in part A. Payments specifically designated as being in the negotiated wage scale for any benefit should be excluded from hourly scales reported below even though they are made or credited to the worker each pay period. Membership information will be kept in confidence and used only to compute average wage rates. Schedule No.______ Trade or occu p a tion Page_____ of Number o f union mem bers working or im m ed ia tely a vailable for work at each rate on July 1— Weekly hours before overtim e rate is e ffe ctiv e on July 1---- W age scale for each trade or occupation in e ffe ct on July 1— Number o f apprentices in ea ch trade on July 1— (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (ID (12) (13) (14) (15) (16) (17) (18) C od e T itle 1969 1970 1971 Code Code 1969 1970 1971 C ode C ode 1969 1970 1971 1970 1971 Code UNION WAGE RATES 151 152 BLS HANDBOOK OF METHODS transit to 1929, and for local trucking to 1936. Although data for the latter two industries were collected for years before the dates of the index series, indexes were not constructed because of inadequacies in the available data. Analysis and Presentation The averages and indexes mentioned to gether with other summary data are contained in the bulletins published annually for the in dividual studies. Included among the informa tion shown for individual trade classifications is the proportion of union members having hourly rates at different levels, as well as the proportion of union members having, since the previous study, wage rate increases of specified amounts in terms of cents per hour and per cent. The increase registered by the trade is shown also. In addition, the union rates of wages and hours in effect on the date of the survey, as reported by union officials, for both the previ ous and current years are published for each classification by city. These furnish a direct comparison of union rates between the 2 years for each of the industries studied. The rates of wages are indicated as hourly rates and the hours as the weekly hours of work before over time rates are applicable. The current studies also present data on employer payments for insurance (health and welfare) and pension payments; in addition employer payments for vacation and other funds (except those for apprenticeship) are shown for the building trades. These payments are expressed in terms of cents per hour or as percent of rate. 7 Membership (used for weighting purposes) relates only to active members in the city and contiguous suburban areas. It does not reflect the total jurisdiction of local unions, which may extend beyond these limits, and it does not necessarily reflect metropolitan area rates. Uses and Limitations The Bureau’s union-wage series provide a means of determining intercity wage differ ences for comparable work, and the relation ships between rates applicable to workers in occupations requiring varying degrees of skill. The data are used in wage negotiations by both management and labor. The wage rates of building-trades workers are especially impor tant in estimating construction costs, because labor expenditures constitute an important element in the total cost of building construc tion. The index series derived from these studies provide barometers of year-to-year changes in rates of wages and hours in the industries covered. Average union rates provide comparisons of wage rates among industries, trades, and cities at a given time. Unlike the indexes, they are not an accurate measurement of year-to-year changes because of fluctuations in membership and other factors. Membership figures for the various trades or classifications do not remain constant and changes may have a marked effect on average rates. For example, if organiza tional drives in cities having relatively lower rates of wages result in sharp increases in membership, the movement of the rate levels for the affected trades as a whole is naturally retarded. Conversely, increases in membership in cities having high wage rates accelerate the upward movement of averages.7 The union rates, are not necessarily the actual rates paid to all workers, and the union hours are not necessarily the hours actually worked. Workers with above average exper ience and skill may be employed at rates above the union wage rates, especially during pros perous times when a tight job market creates competitive bidding for the better workmen. During periods of depressed business activity, actual hours worked often are less than hours specified in the union agreement. — T homas C. M obley Chapter 17. Current Wage Developments Background Since January 1948, the Bureau of Labor Statistics has issued a monthly report listing general wage changes and changes in supple mentary benefits agreed to in selected collective bargaining situations, identifying the situa tions by company and union name. The scope of the listing has varied somewhat from time to time, but since 1953, it has been limited to agreements affecting approximately 1,000 or more production and related workers in manu facturing and selected nonmanufacturing in dustries. Current Wage Developments was initiated because of the rapid increase in wage rates and prices in the early post World War II period, the interest in determining the extent to which settlement patterns spread from industry to industry, and the discontinuance of an index of wage rates that had been initiated during World War II. Interest in the listing was stim ulated by the Korean emergency when the Wage Stabilization Board needed data on the extent to which wages and benefits were being changed. In 1949, and again in 1951 and 1952, sta tistical summaries of wage changes were pre pared to supplement the listing, but regular preparation of a statistical summary began in 1954. These quarterly statistical summaries show the distribution of settlements and (since 1955) of workers by the size of the general wage changes agreed to and the frequency with which various types of supplementary benefits were introduced or changed. Beginning in 1959, another statistical sum mary was instituted. It is limited to manu 1 The listing, as contrasted with these summaries, provides a much more detailed account of negotiated wage and benefit changes than can be presented in a tabular summary. When available, information on changes for large groups of non union workers, including professional, white-collar, and pro duction employees, also is presented. 1 Only changes in benefits that represent changes in costs are included. 3 Prior to 1966, the construction service trades and finance industries are also excluded. 4 General wage changes are defined as changes affecting at least one-tenth of the workers at any one time or all workers In an occupation. Changes resulting from promotions, merit increases, etc., are excluded. facturing, but includes information on general wage changes for nonunion and small union situations, as well as for large collective bar gaining situations.1 From 1959 through 1970, this summary also included information on changes in supplementary benefits. Description of Series The summary of major collective bargaining situations hereafter is referred to as the “ major” series, and the summary that is based on changes in wages in manufacturing firms of all kinds is described as the “ manufactur ing” series. The major series describes general wage changes and changes in benefits1 in all collec 2 tive bargaining settlements involving 1,000 or more production and related workers in manu facturing and 1,000 or more nonsupervisory workers in the nonmanufacturing sector, ex cluding Government.3 Supervisory or professional employees are excluded. Large units of technicians are in cluded even though they are part of a bargain ing unit that is predominantly professional. Contracts covering multiplant firms are included if the agreement as a whole covers 1,000 or more workers even though each indi vidual plant employs fewer. Also included are contracts with trade associations or with groups of firms that bargain jointly with a union or unions, even though the firms are not associated formally and each has fewer than the minimum number of workers covered by the series. Situations in which two or more unions, together representing more than 1,000 workers but individually accounting for fewer, negotiate essentially identical contracts with one firm or a group of firms, are tabulated as one bargaining unit. The summary for manufacturing as a whole represents all establishments with four or more employees that adjust wages by means of gen eral wage changes,4 regardless of whether the workers are represented by a union. Wage change data are presented in cents per 153 154 BLS HANDBOOK OF METHODS hour and, since 1959, as a percentage of aver age straight-time hourly earnings, adjusted to exclude premium pay for overtime work. Two types of information are presented on wage changes: (1) information on changes that go into effect within 12 months of the date on which they are decided or negotiated, thus reflecting the economic climate at the time the changes are decided upon; and (2) all changes effective during the period being summarized whether or not they were (a) decided upon during that period, (b) were deferred— that is, decided upon earlier— or (c) resulted from operation of cost-of-living clauses. In distribu tions of workers by size of wage change, all workers in an establishment or collective bar gaining situation are distributed according to the average wage increase in the establishment or situation. The number of workers affected by changes in supplementary benefits includes all production and related workers in the situa tions where the benefit is changed, whether or not all are affected immediately. For example, if a fourth week of vacation is added for work ers with 20 years’ service in an establishment employing 1,000 workers, a vacation change would be recorded for 1,000 workers, even though only a relatively small proportion would benefit from the change immediately. Data Sources and Collection Methods The statistical summary of the major series is compiled from the summaries of collective bargaining settlements presented in the monthly Current Wage Developments listing which, in turn, is derived primarily from sec ondary sources, including general circulation newspapers and periodicals, as well as union, management, and trade publications. Other important sources of information are the file of union contracts maintained by the BLS and the U.S. Department of Labor’s files of pension and health and welfare agreements, maintained by the Office of Labor-Management and Wel fare-Pension Reports.5 By the end of the year, 6 the BLS contacts, almost entirely by mail, either management or labor representatives in any situation for which these other sources have not yielded information on wage and benefit changes during the year. Information for nonunion and small union ized firms is gathered quarterly (semiannually in 1965 and 1966) by a questionnaire mailed to participating establishments. The information on general wage changes is supplemented by the contract file (unionized establishments) and from newspaper clippings, purchased from a commercial clipping service. At the end of the year, BLS field representatives contact, pri marily by telephone, a sample of firms that have failed to respond to the mail questionnaire or that have provided incomplete or unclear infor mation. Sampling and Estimating Procedures As indicated earlier, all bargaining situa tions with 1,000 or more workers in manu facturing and nonmanufacturing industries are included in the major series. It is believed that the current list of nearly 2,500 such situations, built up since Current Wage Developments was started in 1948, is very nearly complete. After a bargaining situation is added to the universe, it is withdrawn only if it ceases to be within the scope of the survey (e.g., because of a change in business to one outside the scope of the survey, a change to nonunion from union, or because of an apparently permanent drop in employment to substantially below 1,000). The sample for manufacturing is derived from State unemployment insurance listings (UI) which show reporting units with four or more employees by location, number of em ployees, and industry classification.6 The sam ple is a highly stratified probability design, with sampling ratios varying from 1 out of 150 establishments with 4 to 19 employees to all of those with 1,000 or more employees.7 5 Information from contracts supplied on a confidential basis is used only in the statistical summaries, not for the monthly listing. 6 See appendix B, "Industrial Classification.” For a more detailed description of unemployment insurance data, see p. 15, chapter 2. 7 In the case of a few companies with large numbers of establishments each with 1,000 workers or more, a sample of plants is chosen. 155 CURRENT WAGE DEVELOPMENTS U.S. D EPA RTM EN T OF LABOR B L S 2675d O f f ic e o f M anagem ent and B u d g e t N o . 4 4 -R 1 1 3 5 BUREAU OF LABOR S T A T I S T I C S W A S H IN G TO N , D .C . A p p r o v a l e x p ir e s 3 /3 1 /7 2 20212 Wage Developments in Manufacturing, 1971 T he B u reau w ill h o ld is h e d stric t by o f L a b o r S ta tis tic s a ll in fo r m a tio n th e fu r n resp on d en t in c o n fid e n c e . I d e n t ific a t io n or lo c a t io n o f e s ta b lis h m e n t fo r w h ich in form ation is re q u e s t e d , if d iffe re n t from m ailin g a d d r e s s . ( C h a n g e i f i n c o r r e c t , i n c lu d e Z I P c o d e .) I. What w a s the m ajor p rod u ct (in term s o f s a l e s v a lu e ) o f th is plan t during 1970? II. P le a s e p ro v id e em p loy m en t and p a y ro ll in form a tion for the p a y r o ll p e rio d in clu d in g January 12, 1971. S e e p a g e 4 for e x p la n a tio n . A. A ll e m p lo y e e s 1. N um ber III. B. A ll p r o d u c tio n and r e la te d w ork ers 1. ~~| | N um ber 2. ~| |$ P a y r o ll 3. ~| | D o c o l l e c t i v e b arg a in in g a g reem en ts c o v e r a m a jority o f you r p ro d u c tio n and re la te d w o rk e rs ? M an-hours Y es [^ j ~ No Q If " N o , ” p le a s e s k ip to s e c t io n V II. _____________________________________________________________________________________________________ If " Y e s , ” p le a s e a n sw er a ll q u e s tio n s e x c e p t X . F O R UNION FIRM S O N L Y IV . U nion and A greem en t I d e n t ific a t io n : A. With what u n ion or u n ion s d o y ou h ave a c o l l e c t i v e b a rg a in in g a greem en t? B. A re you a party to an a greem en t s ig n e d by an em p lo y e r a s s o c i a t i o n w ith th is u n io n (s )? Y es | No Q \ If s o , what is the a s s o c i a t i o n ’ s n a m e ? ______________________________________________________________________________________ ,__________ V. A greem en t E x p ira tion D a te: When d o e $ you r c o l l e c t i v e b a rg a in in g a g re e m e n t(s ) e x p ire or b e c o m e s u b je c t to r e o p e n in g on w a g e s ? (S p a ce ha s b een p ro v id e d for tw o e n trie s s in c e the d a te may ch a n g e during the y e a r .) V I. N ew or R e v is e d A g reem en t: P le a s e mark ap p rop ria te b o x e s b e lo w to brin g you r rep ort up to d a te fo r — J a n .—Mar. A. A p r .—June 1971 During the q uarter— 1971 1971 O c t .—D e c . 1971 D id you n e g o tia te a new or r e v is e d c o l l e c t i v e b a rga in in g c o n t r a c t (s ) for your Y es p ro d u ctio n and re la te d w o r k e r s ? ............................................... ........................□ B. J u ly —S ept. D id you a gree on an im m ediate or d e fe rre d ch a n g e in w a g e s ? ......................................................... .................. □ No □ □ Y es No Y es No Y es No □ □ □ □ □ □ □ □ □ □ □ □ PLEASE TURN TO NEXT PAGE. 156 BLS HANDBOOK OF METHODS FOR UNION AND NONUNION FIRMS V II. W a g e-R ate C h a n g e s for P r o d u ctio n and R e la t e d W orkers, 1971: P le a s e rep ort any g e n e ra l w a g e -ra te c h a n g e s y ou h a ve put in to e f f e c t fo r you r p ro d u c tio n and re la te d w o rk e rs in the p re v io u s quarter. (F o o d - p r o c e s s i n g c o m p a n ie s s h o u ld a l s o rep ort w a g e -r a te c h a n g e s a ffe c t in g d r iv e r -s a le s m e n .) In c lu d e : E x c lu d e: (1 ) A ll c h a n g e s a ffe c t in g eith er (a ) 10 p e rce n t or m ore o f you r p ro d u c tio n and r e la te d w ork ers at any on e tim e, or (b ) a ll w ork ers c o v e r e d b y a s in g le a g reem en t, e v e n i f the a g r e e m ent a p p lie s to fe w e r than 10 p e rce n t o f the w o r k e rs . (1 ) I n c r e a s e s to in d iv id u a ls re s u ltin g from p ro m o tio n s , a u to m a tic i n c r e a s e s w ith len gth o f s e r v i c e , or p r o g r e s s io n w ith in an e s t a b lis h e d rate ra n g e. (2 ) I n c r e a s e s d e c id e d on in 1971 but s c h e d u le d to g o (2 ) A ny ch a n g e in y ou r p a y s c a l e s e v e n though no w ork e rs r e c e i v e d im m ed iate p ay i n c r e a s e s a s a re s u lt o f th is ch a n g e . (3 ) A ny c o s t -o f - l i v i n g e s c a la t o r a d ju stm en ts w h eth er or not they are part o f you r perm anent rate stru ctu re. (4 ) I n c r e a s e s d e c id e d on in e a rlie r y e a r s but g o in g in to e f f e c t in 197 1 . (5 ) C h a n g e s in h ou rly ra te s r e s u ltin g from c h a n g e s in hou rs w ith ou t c o r r e s p o n d in g c h a n g e s in w e e k ly or d a ily p a y . A. (3 ) (T h e s e s h o u ld b e lis t e d T h e c o s t o f any c h a n g e s in su p p lem en ta ry b e n e fit s . H ave you put in to e f f e c t a n y s u ch g e n e ra l w a g e c h a n g e s during the quarter? Jan .-M ar. 1971 Y e s ............................................................................................. . . . . Q N o ............................................................................................. . . . . Q If you r a n s w e r is " N o ” , p r o c e e d to s e c t io n V III. and lis t in s u b s e c t io n C : A p r.-J u n e 1971 J u ly -S e p t. 1971 □ □ □ □ O c t .- D e c . 1971 □ □ If you r a n sw e r is " Y e s ” , in d ic a te b e lo w the form o f the w a g e c h a n g e (s ) (1 ) U niform c e n ts p er h o u r ........................................... . . . Q (2 ) U niform p e rce n ta g e c h a n g e ................. - ................ . . . Q (3 ) H igh er c e n t s p er hour for s k ille d w ork ers - -, . . . Q (4 ) C. in to e f f e c t in la ter y e a r s . in s e c t io n V III.) O ther ( s p e c i f y in s e c t io n X II. " R e m a r k s ” ) - -. . . . Q □ □ □ □ □ □ □ □ □ □ □ □ If a ll w ork ers d id not r e c e iv e the sam e am ount (e ith e r the sam e num ber o f c e n ts or the sam e p e r c e n ta g e ) lis t c h a n g e s for e a c h grou p on a s e p a ra te lin e with the a p p roxim ate number a f f e c t e d . F o r e x a m p le , if there w a s a uniform a c r o s s -t h e board ch a n g e p lu s a d d ed c h a n g e s for s om e w o r k e rs , lis t the uniform ch a n g e firs t and sh ow a d d itio n a l c h a n g e s b e lo w . If a c o s t -o f - l i v i n g e s c a la t o r ad ju stm en t w ent in to e f f e c t at the sam e tim e a s a n oth er in c r e a s e , l i s t it s e p a r a t e ly . In rep ortin g in form a tion for in c e n tiv e w ork ers i n c l u d e , i f p o s s i b l e , e s tim a te d e f f e c t s o f w a g e -r a te c h a n g e s on in c e n tiv e w o r k e r s ’ e a r n in g s . (F o r e x a m p le , if b a s e ra tes for in c e n tiv e w o rk e rs w ere r a is e d 5 c e n ts and th is in c r e a s e d th e ir h ou rly ea rn in g s a bou t 7 c e n t s , report 7 c e n t s .) If any c h a n g e s in s c a l e s w ere m ade that d id n ot a ffe c t any w o rk e rs im m e d ia te ly , in d ic a te the a p p ro xim a te num ber to be ______________________________________________ a ffe c t e d by the en d o f the y ea r. S e e p a g e 4 for e x a m p le o f In d ica te w h eth er ch a n g e w a s g iv e n in p e rce n ta g e or c e n ts term s. re p o rtin g g e n e ra l w a g e c h a n g e s . Approxi mate number Effec receiving wage tive adjust date ments Check if this was: Classes of production and related workers or jobs affected Deferred (effective in 1971 but decided earlier) Hourly change for workers (+ or -) % t % % % % Auto matic cost-ofliving escalator 157 CURRENT WAGE DEVELOPMENTS V III. Future W a g e-R a te C h a n g e s : L i s t any g e n e ra l w a g e -r a te c h a n g e s a lre a d y d e c id e d upon fo r you r p ro d u c tio n and r e la te d w o rk e rs but s c h e d u le d to g o in to e f f e c t in the future. Amount of change for W orkers a f f e c t e d E ffe c t iv e d a te (m on th , d a te , and y e a r) workers whose rates E stim a ted N um ber C l a s s e s or jo b s will be changed _________ L % _________ L % i % % IX . C o s t -o f-L i v i n g E s c a la t io n : D o you h a v e a c o s t -o f - l i v i n g e s c a la t o r p o l ic y w h ereb y w a g e s are c h a n g e d a u to m a tic a lly w ith s p e c i f i e d c h a n g e s in a p r ic e in d e x ? Y es | j No [^ ] If s o , p le a s e l is t the m onths in w h ich w a g e c h a n g e s w ill g o in to e f f e c t i f the p rice in d ex w a r r a n t s .______________________________________________________________________________________________________ ________________ F O R N ON U N IO N FIRM S O N L Y X. P o l i c y R eg a rd in g W age C h a n g e s : W hether or not you c h a n g e d w a g e s durin g the firs t quarter o f 197 1 , w h at i s you r p o l ic y re g a rd in g g e n e ra l w a g e c h a n g e s ? 1. W ag es are n orm a lly ch a n g e d o n ly on an in d iv id u a l b a s i s . 2. X I. X II. G en era l w a g e c h a n g e s are s o m e tim e s m ad e. [[ [ D o y ou w ant a c o p y o f the B u rea u ’ s sum mary on th is s u r v e y ? ..................................... Y e s | | No R em a rk s: Name and t itle o f p e rs o n fu rn ish in g d ata ( P l e a s e T y p e or Print) Area c o d e , Phon e No. 158 BLS HANDBOOK OF METHODS 4 Explanations for Report on Wage Developments in Manufacturing, 1971 Section 11-A A ll employees—o ta l t number on the p a y r o ll o f the p la n t c o v e r e d b y th is rep ort w h o w o rk e d fu ll-tim e or p art-tim e or r e c e iv e d p ay for any part o f the p e rio d re p o rte d . I n clu d e p e r s o n s on p a id v a c a t io n s and s ic k l e a v e . E x c lu d e p e r s o n s on le a v e w ith ou t com p a n y p ay the e n tire p e rio d as w e ll a s p e n s io n e r s and m em bers o f the Arm ed F o r c e s n ot w ork in g during the p e rio d re p o rte d . Section ll-B Production and related workers. In c lu d e w ork in g forem en and a ll n o n s u p e r v is o r y w o rk e rs e n g a g e d in fa b r ic a tin g , p r o c e s s in g , a s s e m b lin g , in s p e c t in g , r e c e iv in g , s to r in g , h a n d lin g , p a c k a g in g , w a re h o u s in g , s h ip p in g , tru ck in g , h a u lin g , m a in te n a n c e , rep a ir, ja n it o r ia l, w a tch m en s e r v i c e s , p rod u ct d e v e lo p m e n t, a u x ilia ry p ro d u c tio n for p la n t’ s ow n u s e ( e . g . , p o w e r -p la n t), re c o r d k e e p in g , and oth er s e r v i c e s c l o s e l y a s s o c i a t e d w ith t h e s e p r o d u c tio n o p e r a t io n s . E x c lu d e e m p lo y e e s e n g a g e d in e x e c u t iv e , p u rc h a s in g , fin a n c e , a c c o u n t in g , p e r s o n n e l, c a f e t e r ia , p r o f e s s i o n a l , and t e c h n ic a l a c t iv i t i e s ; s a l e s , a d v e r t is in g ; c o l l e c t i o n ; in s t a lla t io n , and s e r v ic in g o f p r o d u c ts ; rou te o f f i c e fu n c t io n s ; fa c t o r y s u p e r v is io n a b o v e the w ork in g forem en l e v e l ; and e m p lo y e e s on you r p a y r o ll e n g a g e d in c o n s t r u c tio n o f m ajor a d d itio n s or a lte r a tio n s to the p la n t. Number. In clu d e both fu ll-tim e and p art-tim e p r o d u c tio n and r e la te d w ork ers on you r p a y r o ll—w h eth er w a g e or s a la r ie d —w h o w ork ed during or r e c e iv e d p a y for any part o f the p a y r o ll p e r io d r e p o rte d . In clu d e p e r s o n s on p a id s ic k le a v e , p a id h o lid a y s , and p a id v a c a t io n s . Payroll. In clu d e p a y e a rn ed durin g the p a y r o ll p e rio d by p ro d u c tio n and r e la te d w ork ers re p o rte d in the p r e c e d in g b o x . P a y r o ll s h o u ld b e re p o rte d b e fo r e d e d u ctio n fdr o ld -a g e and u n em p loym en t in s u r a n c e , g ro u p in s u r a n c e , w ith h o ld in g ta x , b o n d s , and un ion d u e s . Man-hours. In clu d e c o s t -o f - l i v i n g a l lo w a n c e s , p a y for o v e r tim e , h o lid a y s , v a c a t io n s , and s ic k le a v e . In clu d e a ll h ou rs w ork ed , not s c h e d u le d h o u rs , durin g the p a y r o ll p e rio d b y the p r o d u c tio n and r e la te d w ork ers rep orted in the firs t b o x p lu s h ou rs p a id for sta n d -b y or rep o rtin g tim e and h o lid a y s , and m an -hours e q u iv a le n t to p a y r e c e iv e d by e m p lo y e e d ir e c t ly from you r firm for s ic k le a v e and for h o lid a y s and v a c a t io n s for th is p a y r o ll p e rio d . Section Vll-C E a ch p e rio d rep ort the gen era l w a g e -r a te c h a n g e s y o u h a v e put in to e f f e c t fo r y ou r p ro d u c tio n and r e la te d w o rk e rs s in c e you r p r e v io u s rep ort t o u s . for d r iv e r -s a le s m e n .) E ffe c tiv e d a te E s ta b lis h m e n ts in the f o o d - p r o c e s s i n g in d u s trie s s h o u ld a l s o rep ort w a g e -r a te c h a n g e s C h e c k i f th is w a s : A p p r o x i m ate number r e c e iv in g w age a d ju s t m ents 1 /5 1,0 0 0 1 /5 9 50 2 /1 0 (N o te : E n trie s o f v a riou s ty p e s o f w a g e c h a n g e s ca n be illu s tr a te d a s f o l l o w s : 1 ,0 0 0 C la s s e s o f p r o d u c tio n and re la te d w ork ers or jo b s a ffe c t e d D e fe rre d ( e f f e c t i v e A u to in 1971 m a tic c o s t-o fbut d e c id e d liv in g e a r lie r ) e s c a la t o r H ou rly ch a n g e for w ork ers (+ or - ) A ll + 8 A ll e x c e p t com m on la b or A vg. A ll A ll 3 + 2 4^ — A vg. 0 <t % + ^ to 5 % 0 % 8 0 6 0 X % 6 /1 5 1,000 6 /1 5 100 M ain ten a n ce + 9 add. 0 % 8 /5 4 00 T im e +6 0 % 8 /5 600 In ce n tiv e + 7 0 % A ll - 1 t % P ow erp la n t and m a in ten a n ce + 10 0 % 1 0 /1 1 ,0 0 0 1 2 /5 125 X CURRENT WAGE DEVELOPMENTS The ratios are uniform for all industries. Since data are available from secondary sources for all unionized situations with at least 1,000 production and related workers, data for all establishments meeting this criteria also are included in the summary for manufacturing. The sample selected from the UI listings is compared with this list of establishments for which information already is available; since data for these sample members are obtained from secondary sources, these establishments are not sent questionnaires. Approximately 6,000 establishments are left for the question naire survey. Although the sampling design yields a sample in which large firms are relatively over represented, this bias is overcome by the esti mating procedure. Each establishment in the sample is assigned a weight which is the re ciprocal of the sampling ratio in the stratum from which it was selected. An establishment selected from a stratum from which 1 out of 4 establishments is chosen is assigned a weight of 4, so that it represents itself and three other establishments. Information for each establish ment is multiplied by the weight assigned to the establishment. Thus, all establishments, re gardless of size, are represented appropriately in the final estimates. An establishment in the subsample of non respondents followed up by a visit is weighted to represent all nonrespondents in the stratum. It is assigned a new weight— the product of the original weight and the inverse of the subsampling fraction. Thus, 1 out of 3 non respondent establishments subsampled from a group originally sampled at the rate of 1 out of 2 would be assigned a weight of 6. If an estab lishment included in the sample with certainty fails to respond, another similar establishment would be weighted to represent it. To the estimates derived from the weighting of the sample questionnaire are added the data from secondary sources— the numbers of major bargaining situations. The totals thus obtained are further adjusted to reduce the hazards of sampling and to take account of opening or closing of establishments ■See chapter 2. 159 between compilation of the State unemploy ment insurance listing from which the sample is chosen and the date of the survey. Adjust ments are made to employment levels for prod uction workers in the 2-digit Standard Indus trial Classification manufacturing industry groups, as reported in the monthly employment series of the Bureau of Labor Statistics for the period covered by the summary of wage changes.8 For example, if the estimate of prod uction worker employment in an industry group derived from the sample is 100,000 but the Bureau’s estimate of employment in that industry group was 110,000 workers, each em110,000 ployment count would be multiplied by ^qq qqq or 1.1 The resulting industry group estimates would be added to provide the estimates for all manufacturing. The major series for manufacturing and nonmanufacturing combined is not adjusted in this fashion, since it is presumed to be all inclusive. A new sample of nonunion and small union ized plants in manufacturing usually is selected every 3 years. Establishments with fewer than four workers are omitted because in many States they are not covered by these unemploy ment compensation programs. After the initial contact, establishments of any size that indi cate that they have a policy of adjusting wages on an individual basis, rather than by means of general wage changes, are omitted from further survey. Presentation and Analysis Preliminary information on the “ package cost” and general wage changes resulting from collective bargaining settlements involving the major situations is issued in press releases about 3 weeks after every quarter and the in formation is also summarized in Current Wage Developments (See Chapter 23 for a descrip tion of the package cost program ). Final infor mation on wage and benefit changes is not available until the end of the year. It is pre sented in Current Wage Developments and the Monthly Labor Review. Yearend summaries also present information on total effective 160 BLS HANDBOOK OF METHODS wage changes— those negotiated in the period plus deferred and cost-of-living changes re sulting from settlements in prior years but going into effect in the current year. Informa tion is presented for manufacturing industries, for nonmanufacturing industries, and for both combined. Quarterly summaries of the manufacturing data are published in Current Wage Develop ments and the article covering the full year also appears in the Monthly Labor Review. The 3-, 6-, and 9-month summaries, which are based on preliminary data, stress wage changes re sulting from settlements or management de cisions made during the period, while the yearend article, which is based on final data, also analyzes trends in the size, frequency and type of wage changes, and the prevalence and results of wage escalation policies. Because it is based on data for both large and small union ized and nonunionized establishments, the manufacturing analysis can make many other useful comparisons of its components. Uses and Limitations The data are used extensively by labor, man agement, and the Federal Mediation and Con ciliation Service in collective bargaining; by private institutions and universities in studies of industries or groups of industries; and by local and Federal Government agencies in terested in the current economic picture to determine trends in wage and benefit changes as well as for wage, income, and gross national product forecasts. Since the sample is relatively small, data are not presented for individual industries. — G eorge R u b e n Chapter 18. Employer Expenditures for Employee Compensation Background The measurement of employer expenditures for employee compensation and the composition of payroll hours was undertaken by the Bureau of Labor Statistics (BLS) to fill a large gap in the statistics of employee compensation and hours paid for. Prior to World War II, compensation for American labor consisted mainly of wages and salaries for time worked or units produced. Under the New Deal, however, additional pay ments were required under various social in surance programs, and, later, during the years of the Second World War, employers were en couraged by the policies of the War Labor Board to grant wage supplements instead of wage increases, e.g., vacations, hospitalization. Shortly after the war, the NLRB ruled that pension plans were within the purview of col lectively bargained agreements.1 Expenditures for these and other compensation elements, in addition to. pay for working time, began to comprise a substantial portion of the total com pensation of labor. As early as 1875 the American Express Com pany instituted a private pension plan.1 In 2 1929, a private study3 indicated that there were almost 400 such plans, and by 1968 there were about 34,100 pension plans and 157,700 private pension or welfare plans in America.4 Paid vacations and holidays also have a rela tively short history for most workers. Paid vacations were fairly well established for salaried workers by the middle of the nine teenth century. Industrial workers, however, first started to receive paid vacations around the turn of the century, and not until after World War I did the principle of paid vacations begin to assume importance in the development of labor policy; paid holidays generally were not found in industry until World War II, al though it had been customary for salaried workers to receive pay for time not worked on designated holidays. By 1968, approximately 7 percent of the production worker hours paid for in manufacturing industries were leave hours, almost all of which were vacation and holiday hours. The importance of recent changes in the structure of compensation may be illustrated by examining those that occurred for manu facturing production workers between 1959 and 1968. During that period pay for working time increased from $2.23 to $3.02 an hour or about 35 percent. At the same time, employee expenditures for all other elements of compen sation increased from 38 cents to 69 cents an hour or about 76 percent. Thus, the relative importance of pay for working time decreased from 85.4 percent of total compensation in 1959 to 81.8 percent in 1968. The Bureau has for many years recognized the necessity of studying outlays for employee compensation. Early attempts, however, were limited to exploratory work on methodology and the availability of data.5 By 1959, many of the technical and conceptual problems had been sufficiently resolved to permit the initia tion of a regular program. The first survey in the program, 1959 ex penditures in manufacturing, was followed by a 1960 mining study; a 1961 finance, insurance, and real estate survey, and another manufac turing industry study in 1962. The 1963 study of expenditures for salaried (white-collar) workers, which covered most nonagricultural industries in the private sector, represented the first shift in program emphasis from an indus try to an economy-wide orientation. Since then, the program has been redesigned to cover all employees in the private nonfarm sector and to cover all significant items of employee com pensation. 1 Inland Steel vs. National Labor Relations Board, 170 Fed eral Reports, Second Series 247 (1948), 251 (1949). 2 This was the first recorded private pension plan in America. “ Latimer, Murray Webb. Industrial Pension Systems in the United States and Canada, Industrial Relations Counselors, Inc., New York, 1932. 4 Total active plans covering 26 employees or more for which descriptions had been filed with the U.S. Department of Labor by April 1, 1968 under the provisions of the Welfare and Pension Plans Disclosure Act, as amended (P.L. 85-836 as amended by P.L. 87-420). 5 Problems in Measurement of Expenditures on Selected Items of Supplementary Employee Remuneration, Manu facturing Establishments, 1953 (BLS Bulletin 1186, 1956); Wage Structure; Basic Iron and Steel, January 1951 (BLS Series 2, No. 81, 1951); Wage Structure: Electric and Gas Utilities, September 1957 (BLS Report 135, 1958). 161 162 BLS HANDBOOK OF METHODS Description o f Survey The survey relates to employee compensation practices, employer expenditures arising from these practices, and to all hours for which pay ment is made— hours worked, paid hours of vacation, holiday, sick leave, and civic and personal leave. The program is designed to provide data biennially for the entire private nonfarm sec tor; separate information is given for manu facturing and nonmanufacturing industries. In the intervening years a number of individual manufacturing and nonmanufacturing indus tries are studied. Both types of studies provide data for all employees and for office and non office employees separately. Survey coverage extends to the 50 States and the District of Columbia and separate data are provided for broad economic regions. The data relate to cash disbursements of employers and the hours-paid-for during a calendar year. Data for a lesser period of time do not completely reflect the outlays made by employers and the hours-paid-for. Paid leave time, for example, usually is spread unequally throughout the calendar year; similarly, ex penditures for many of the legally required insurance programs stop after a specified maximum amount is earned by each worker during the year. These practices result in wide variations between expenditures in the early part of the year and in the latter part of the year. Employee compensation is increasing contin ually in complexity and magnitude. Practices differ by industry group and new supplemen tary practices are being added. Each survey is designed to show the individual characteristics of compensation practices in particular indus tries without impairing comparability with others. This objective is accomplished by re taining the same basic reference framework. Expenditures The expenditures studied are considered to constitute the major elements of employee compensation in American industry. The ex penditures, and therefore their measurement, fall into two broad groups determined by the way they are paid: payments made directly to the workers and payments made to a third party on behalf of the workers. Direct payments include pay for working hours; pay for hours not worked such as va cations, holidays, sick leave, and civic and personal leave; premium pay for overtime, weekend, holiday, and shift work; nonproduc tion bonuses; and severance pay. The total of these payments constitutes the gross payroll. Indirect payments are those made by the employer on behalf of the worker to funds, trustees, insurance companies or Government agencies which may make a payment to the worker at a later date or provide full or partial economic security against a future contingency (i.e. unemployment, retirement, medical ex penses etc.). The programs from which these expenditures arise are either legally required or voluntary. Legally required programs in clude social security, unemployment insurance, workmen’s compensation, and State temporary disability insurance. Voluntary plans studied are life, accident, and health insurance; retire ment plans; vacation and holiday funds; severence and supplemental unemployment benefit funds; and savings and thrift plans. Payroll Hours The payroll hours studied are all hours for which the workers receive pay. These hours consist of plant or working hours, and vaca tion, holiday, sick and other hours of paid leave. Although an hour normally is defined as 60 minutes of elapsed time, a payroll hour does not necessarily consist of 60 minutes. For ex ample, hours worked on a day that would other wise have been a paid holiday are paid for twice— once as a paid holiday, and once as working hours. Therefore, an 8-hour holiday worked for which 16 hours of payment was made is counted as 16 hours— half of which are holiday hours and half are plant-hours. Conversely, some hours of leave are paid for at less than the regular rate and only the equiva lent hours are counted. Each overtime hour worked at premium rate is counted as 1 planthour. EMPLOYER EXPENDITURES FOR EMPLOYEE COMPENSATION Establishment Policies Although the data on company policies are used primarily in the review and analysis of the expenditure and hours data collected in the survey, these policies also have significance in their own right, and often signal changes in supplementary practices before actual expendi tures are incurred. In addition, certain charac teristics of American industry are measured on a national basis and the relationships among these characteristics studied. Examples are the actual distribution of workers by amount of vacation and the degrees of unionization. Data Sources and Collection Methods The data are obtained from establishment records. Generally, no single record is sufficient and several record sources must be summarized to arrive at annual totals. The data are entered by the employer on preprinted forms in accord ance with detailed instructions. Not all companies keep records in the detail requested and approximations in these cases may be accepted. In general, two types of ap proximations are used. First, if the establish ment records are kept for a broader grouping of employees than are being studied, the pro rated share for the workers included in the survey is computed on the basis of employment, man-hours, or payroll, whichever is most ap propriate. Second, by using collateral data, estimates are made where records are not kept but the practice is observed. For example, the expenditures for holiday pay may be approx imated by multiplying the number of hours paid for holiday leave by average straight-time hourly earnings. Errors occurring from the use of these approximations would have to be in the same direction in substantially all the cases (overstatement or understatement of the actual values) to have a material effect on the ac curacy of the results. Data are collected primarily by mail, al though personal visits are made to many of the large employers and to a sample of the estab lishments that have not responded to a second mailing of the questionnaire. A questionnaire form used in the expenditure study is repro duced on pages 164-167. 163 Sampling Procedure The surveys are conducted on the basis of a highly stratified probability sample of estab lishments selected by industry, location, and employment size. The samples generally are designed to yield reliable data for an industry division at the national levels, in four broad economic regions, and for major industry groups. The lists of establishments from which the samples are selected are those maintained by the State agencies administering the employ ment insurance laws. These lists show the em ployment, industry classification, and location of all establishments covered by those laws in each State. Since some States do not cover establishments with fewer than four employees under the unemployment insurance (UI) law, the samples exclude establishments in that size group. (See method of estimation for treat ment of the employment in such establish ments.) Some establishments in particular industries are exempted from the UI laws even though they employ more than four workers. The data used in sampling these establishments are obtained from lists compiled by regulatory Government agencies, trade associations, and other sources. Within each industry, the sample is selected to yield the most accurate estimates possible with the resources available— the principle of optimum allocation. This is done by including in the sample a greater proportion of large establishments than of small. In general, an establishment’s chance of selection is roughly proportionate to its employment size. A subsample of establishments failing to re ply to the mail inquires is selected to represent all nonrespondents, following the same general plan as is used in the original sample. Estab lishments in this subsample are visited per sonally, instead of being solicited again by mail. Estimating Procedures Data for each sample establishment are weighted in accordance with the probability of selection of that establishment. In the individ- BLS HANDBOOK OF METHODS 164 U.S. DEPARTMENT OF LABOR BLS 2868 (R e v . '69) B udget B ureau N c. 4 4 -R 1 300 A p p ro v al ex p ires 10-31-71 BUREAU OF LABOR STATISTICS W a s h in g t o n , D.C. r i L 20212 J L o c a tio n of u n it for w h ic h d a ta a re req u e sted . Expenditures for Employee Compensation, 1970 G en tlem en : The B ureau of L ab or S ta tistic s is conducting an im p ortant su rvey of how m uch co m p an ies spend fo r em p loyee co m p en sation— fo r w ages and s a la r ie s and for le g a lly requ ired and private p ro g ra m s w hich provide for e m p lo y e e s ' health and w e lfa r e . The G overn m en t has u sed such in form ation fro m e a r lie r su rv e y s to aid in fo r m u lating e con o m ic policy and ask s that you a s s i s t it by com p letin g this fo r m . In form ation fr o m the su rvey w ill be of value to your com pany a ls o , since it w ill enable you to com p are your expenditures for e m p lo yee com p en sation with those m ade by in du stry in ge n e ra l. K eep a copy of your rep ort— an ex tra fo rm is e n c lo se d . W hen the B u r e a u 's re p ort on the su rvey is issu e d , we w ill send you a copy and show you how to co m p are data fo r your com pany with the national a v e r a g e s . Y ou r rep ort w ill be held in confidence and w ill be seen dnly by sw orn e m p lo y e e s of the B u reau . Nothing w ill be re le a s e d relatin g to individual co m p a n ie s. P le a s e co m p lete the fo rm within 3 w eeks and retu rn it in the e n clo se d en velop e. If you have questions about the in form ation req u ested , or if you need a ssista n c e in co m p letin g the fo r m , phone the B ureau c o lle c t at a r e a code 2 02, 9 6 1 -4 0 1 9 or 9 6 1 -4 0 3 0 . Thank you fo r your co op eration. C o m m is s io n e r I. Company official to contact if there are questions about this report: N am e and t it le D. (P lease p rin t o r ty p e ) A rea co d e, phone no. Units covered by this report: Is th is q u e stio n n a ire b e in g c o m p le te d fo r th e unit(s) d e s ig n a te d ab o v e? . I W hat is th e p rin c ip a l p ro d u ct, se rv ic e , or a c tiv ity of th is u n i t ? _________________________________________ __ * □ Yes □ No, o u r reco rd s m ak e it im p o ssib le to rep o rt se p a ra te ly for th e u n it d e sig n a te d ab o v e. for w h ich d a ta are req u e sted a re in clu d e d in th is rep o rt. U nits in a d d itio n to th e one P lea se complete Item VII at end o f questionnaire to describe the units covered. III. Average 1970 employment in units covered by this report: P lease e n te r th e a v e ra g e n u m b er o f e m p lo y e e s in e a c h c a te g o ry d u rin g 1970. T ypes o f e m p lo y e e s in e a c h c a te g o ry are d e scrib ed b elo w . In clu d e f u ll - an d p a r t- tim e e m p lo y e e s. A. O ffic e e m p lo y e e s B. N o n o ffice e m p lo y e e s C. T o ta l 1970 a v e ra g e e m p lo y m e n t_____________________________ _ ________________________________ _ OFFICE EMPLOYEES— In clu d e a ll e m p lo y e e s in e x e c u tiv e , a d m in istra tiv e , and m a n a g e m e n t positions, ab o v e th e w orking supervisor le v e l. A lso in clu d e supervisory and nonsupervisory pro fessio n al e m p lo y e e s an d th e ir t e c h n ic a l assistants; e m p lo y e e s en g a g ed in o ffic e c le r ic a l o p e ra tio n s; and a ll salespersons whose sales a c tiv itie s are p rim a rily p e rfo rm e d o u tsid e o f th e e sta b lish m en t ( e . g . , r e a l e sta te sa le sm e n , and d o o r-to -d o o r sa le sm e n ). NONOFFICE EMPLOYEES— In clu d e a ll e m p lo y e e s, e x c e p t o ffic e e m p lo y e e s as d e fin e d a b o v e , in nonsupervisory, n o nprofessional positions. In clu d e e m p lo y e e s en g ag ed in f a b ric a tin g , processing, or assem bling; b u ild in g or e x c a v a tin g ; m in in g , d rillin g , or p um ping; m a in ta in in g o r re p a irin g ; sh ip p in g , re c e iv in g , h an d lin g , w arehousing, p a c k in g , o r tru c k in g ; r e ta il sales; o p e ra tin g or w orking on m o v in g v e h ic le s (buses, b o ats, e t c . ); ja n ito ria l work; guard or w a tc h m a n work; and s im ila r a c tiv itie s . TO TA L EMPLOYMENT— Is th e sum o f o ffic e plus non o ffice e m p lo y e e s. P ro p rieto rs, m em b ers o f u n in co rp o rate d firm s, and w orkers are n o t c o n sid e red to be e m p lo y ees and a re e x c lu d e d fro m th e survey. u n p a id f a m ily EMPLOYER EXPENDITURES FOR EMPLOYEE COMPENSATION 2 Instructions for Specific Items Item IV. A. G r o s s p ay roll— total of w a g e s, s a la r i e s , and other paym ents m ade during 1970 b e fo re any dedu ction s. The am ount should equal w ages rep orted on Internal Revenue S e r v ic e F o r m s W -2 as subject to F e d e r a l withholding ta x e s , or total rem u n eration rep orted on IRS F o r m 9 4 0 , E m p lo y e r 's Annual F e d e r a l U n em ploym en t Tax R eturn 1970, Schedule B , Item 1. A -2 . Pay for ov ertim e and w eekend and h o l i d a y w ork— for o v e r tim e , re p ort the s tr a ig h t-tim e pay for w ork b e y o n d the n o rm a l w orkw eek in A - 2 a and p rem iu m pay in A - 2 b . F or e x a m p le : If o v ertim e is paid at tim e and o n e -h a lf, rep ort tw o th ird s of this pay in A - 2 a and o n e -th ird in A - 2 b . F o r w ork on w e e k e n d s , h o lid a y s, or during vacation p eriod s— rep ort the r e g u lar pay fo r w ork in A - 2 a , and the p rem iu m pay for w ork on w eek en d s, h o lid a y s, or during vacation p eriod s in A - 2 b . P aym en ts m ade in lieu of tim e off for h olidays or va ca tion s should be excluded fro m A - 2 and included in A - 4 . F or e x a m p le : An e m ployee w orked on a h oliday. He re ce iv e d h is reg u la r pay fo r w orking; plus o n e -h a lf his re g u la r pay as a p rem iu m for having to w ork on a h olid ay; plus his re g u la r h oli day pay. R eport h is reg u la r pay fo r w o rk ing in A - 2 a ; the p rem iu m in A -2 b ; and the re g u la r holiday pay in A -4 b . A - 3 . Shift d iffe re n tia ls— total expen ditu res fo r pay above reg u la r d a y -s h ift r a te s for w ork on late sh ifts . Include pay for hou rs not w ork ed . F o r e x a m p le , . if la t e -s h ift e m p lo y e e s w ork l 1 h hours a day but r e ceive pay for 8 hours re p ort the total of the o n e -h a lf hour paym en ts. A-4. Pay for lea v e — only paym ents that w ere part of the e s ta b lis h m e n t's g r o s s p a y ro ll. E xclude p a y m e n t s to fu n ds, tr u s t e e s , e tc . (R eport th ese in C - 3 . ) If an em p lo ye e re c e iv e d both "p a y in lieu of tim e o f f " and "p a y for w o r k " include only "p a y in lieu of tim e o f f " h e re . B. L e g a l l y r e q u i r e d in su ran ce— net lia b ility in cu rred during 1970 under the p ro v isio n s of State and F e d e ra l law s for each p ro g ra m . Exclude p aym ents m ade in 1970 for 1909, but include p aym ents m ade in 1971 fo r 1970. Include paym ents to g o v ernm ent a g e n c ie s , insurance co m p a n ie s, and to e m p lo y e e s under s e lf-in s u r e d plans. E xclude paym ents m ade by or withheld fr o m e m p lo y e e s. R e q u ire d P ay m en ts— S o c ia l S ecu rity : In 1970 th e e m p lo y e r's p a y m e n t was 4 .8 p e rc e n t o f th e first $ 7 ,8 0 0 p a id e a c h e m p lo y e e , or a m a x i m u m o f $ 3 74. 40 p er e m p lo y e e . See Form 9 4 1 —E m p lo y e r's, Q u a rte rly F ed eral T a x R e tu rn w h ich you file d in A p ril, July, an d O cto b er 1970 an d January 1971. R eport th e sum o f o n e h a lf th e FICA T a x re p o rte d on lin e 5 for th e 4 q u a rte rs. U n e m p lo y m e n t in su ran ce— F ed e ra l: In 1970 th e e m p lo y e r's p a y m e n t was 0.5 p e rc e n t o f th e first $ 3 ,0 0 0 p a id e a c h e m p lo y ee or a m a x im u m o f $15.00 per em ployee. See IRS Form 940 for 1970, S ch ed u le A , Item 16. R ailro ad s in c lu d e to ta l p a y m e n ts under R a ilro a d U n e m p lo y m e n t Insurance A ct. S ta te : In m o st S ta te s th e p a y m e n t was a t v a ry in g rates on th e first $ 3 ,0 0 0 p a id e a c h e m p lo y e e . See IRS Form 940 for 1970, S c h e d u le A , C o lu m n 9. C. P r i v a t e w e l f a r e plans— net pay m e n ts (after deduction of refu n d s, r e b a te s, and dividends) m ade during 1970 by the esta b lish m en t to funds (including u n ionm an agem en t funds), tr u s te e s , insurance c o m p a n ie s, and paym ents m ade under the p ro v isio n s of s e lf-in s u r e d plans to e m p lo y e e s or th eir b e n e fic ia r ie s . Include p ay m e n ts fo r cu rren t e m p lo y e e s , e m p lo y e es on la y o ff, r e tire d e m p lo y e e s and their d e pendents. E xclude em p loyee contributions and a ll ad m in istra tiv e c o s ts incu rred by the e sta b lish m en t. A ls o exclude paym ents m ade by funds, tr u s te e s , and insurance c a r r ie r s to your e m p lo y e e s or th eir b en e fic ia r ie s . Life, a c c id e n t, an d h e a lth in su ra n c e: L ife, a c c id e n ta l d e a th an d d ism e m b e rm e n t, sickness an d a c c id e n t, w age a n d salary c o n tin u a n c e in su ra n c e, an d d e a th b e n e fits; a n d h o s p ita liz a tio n , s u rg ic a l, m e d ic a l, d e n ta l, o p tic a l, an d drug p lan s. Exclude ex p e n d itu re s for in -p la n t m e d ic a l ca re and v isitin g nurses or p h y sic ia n s. Pension an d r e tire m e n t p lans: D irec t p ay m en ts to pensioners un d er a p a y -a s - y o u -g o p en sio n p la n , p a y m e n ts u n d er p ro fitsh arin g plans d e fe rre d u n til r e tire m e n t, an d p a y m e n ts for past an d c u rre n t lia b ilitie s un d er fu nded p lan s. Item V. A - 5 . Nonproduction bon u ses— total am ount paid for nonproduction bon u ses including lu m p -s u m paym ents under p r o fit-s h a r in g p la n s, and o t h e r ir re g u la r or s e a so n a l bon u ses (such as attendance, C h r is tm a s , or yearen d b o n u ses). P ro c e e d s of p r o fit sharing plans which are paid into savin gs and thrift funds or re tire m e n t plans should be rep orted as expen ditu res for a private w elfa re plan ( I V -C ) . A-6. S ev eran ce pay— total of a ll paym ents m ade by the e stab lish m en t to e m p lo y e es b eca u se of te m p o r a r y or perm anent s e v era n ce of e m p lo ym en t. E xclude paym ents to fu n ds, and to p en sion ers under the p r o v is io n s of p a y -a s -y o u -g o pension p la n s. R eport these in I V - C . T otal num ber of hours paid fo r— a ll hours w orked at straigh t tim e , a ll hours worked at o v e r tim e , a ll paid leave h ou rs, and hours equivalent to paym ents m ade by the e sta b lish m en t d ire ctly to the w o rk e rs for other hours not w orked but paid fo r . B . N um ber of o v e r tim e h ou rs— all hours for w ork beyond the n o rm a l workday or w orkw eek for which s tr a ig h t-tim e or p r e m iu m paym ents w ere m ade during 1970. C . N um ber of leave hou rs— rep ort hours equivalent to the paym ents m ade and not the actual tim e taken off. F o r e xam p le: 3 hours paid for at tw o -th ir d s the regu lar rate should be rep orted as 2 h ou rs. In clude leave hours fo r which paym ent was m ade instead of tim e off. 165 166 BLS HANDBOOK OF METHODS 3 G e n e ra l In structions If your reco rd s fo r an ite m c o m b in e d a ta for o ffic e and n o n o ffice e m p lo y e e s, p lea se p ro ra te th e c o m b in e d d a ta b e tw e e n th e tw o e m p lo y e e groups in th e m ost a p p ro p riate m a n n e r. If it is n o t possible to p ro rate th e c o m b in e d d a ta , e n te r th e to ta l fig u re un d er o ffic e and in d ic a te " c o m b in e d " u n d er n o n o ffic e. If your reco rd s c o m b in e d a ta for s e v era l ite m s, p ro ra te th e c o m b in e d fig u re a m o n g th e ite m s to w h ic h it r e la te d in th e m ost a p p ro p ria te m an n e r O R rep o rt th e c o m b in ed fig u re and c le a r ly in d ic a te to w h ic h ite m s it re la te s . If it is n o t possibe to m a k e a n e s tim a te of ex p en d itu res or hours for an ite m , p lea se e n te r "n o t a v a ila b le " on th e a p p ro p riate lin e . P lease c o m p le te a ll ite m s. If th e re w ere no e x p e n d itu re s or hours for an ite m , e n te r "0 . " For e a c h ite m , e n te r to ta ls for th e y e a r 1970 O ffic e N o n o ffice IV. Total compensation in 1970: A. Gross p a y r o l l -------------- -----------------------------------------------------------1. S tra ig h t- tim e p a y -------------------------------------------------------------- 2. Pay for o v e rtim e and w e e k e n d and h o lid a y w o r k --------------a . S tra ig h t- tim e pay for o v e rtim e a n d w ee k e n d an d h o lid a y w ork -------------------------------------------. b . P rem iu m p a y for o v e rtim e a n d w e e k e n d and h o lid a y w o r k ---------------------------------------------------------------- 3. S hift d i f f e r e n t i a l s --------------------------------------------------------------- 4. Pay for l e a v e ---------------------------------------------------------------------a. H o l i d a y s ------------------------------------------------------------------------ c. Sick l e a v e ---------------------------------------------------------------------- d. C iv ic and p e rso n al le a v e ---------------------------------------------- 5. N o n p ro d u c tio n b o n u s e s ------------------------------------------------------- 6. B. S e v e ra n c e p a y -------------------------------------------------------------------- S o c ia l s e c u rity o r r a ilro a d r e t i r e m e n t -------------------------------- 2. U n e m p lo y m e n t in s u r a n c e ---------------------------------------------------a. F e d e ra l or ra ilro a d u n e m p lo y m e n t--------------------------------- b. S ta te ----------------------------------------------------------------------------- 3. W o rk m e n 's c o m p e n sa tio n and p ay m en ts u n d er F e d e ra l E m p lo y e r's L ia b ility A c t ------------------------------------------------ 4. O th e r, e .g ., S ta te tem p o rary d is a b ility in su ran ce (sp e c ify ): E m plo y er E x penditures for P riv ate W e lfa re Plans 1. L ife, a c c id e n t, an d h e a lth in s u r a n c e ---------------------------------- 2. Pension and re tire m e n t p la n s ---------------------------------------------- 3. V a c a tio n an d h o lid a y f u n d s ------------------------------------------------ 4. S e v e ra n c e pay and s u p p le m e n ta l u n e m p lo y m en t b e n e fit f u n d s ------------------------------------------------------------------ 5. Savings and th rift plans ----------------------------------------------------- 6. O th e r p riv a te w e lfa re plans (sp ecify ): V. Total number of hours paid for in 1970 A. N um b er of s tr a ig h t- tim e hours w orked — B. N um ber o f o v e rtim e hours w orked -------- C. ................................ E m plo y er E xp en d itu res for L eg ally R e q u ire d Insurance 1. C. ■■■■■■■ V a c a tio n s ---------------------------------------------------------------------- b. wmmmmmmmm N u m b e r of le a v e hours p a id f o r ------------1. V a c a ti o n -------------------------------------------- 2. H o l i d a y ---------------------------------------------- 3. Sick le a v e ----------------------------------------- 4. C iv ic and p e rso n al le a v e Hours Hours 167 EMPLOYER EXPENDITURES FOR EMPLOYEE COMPENSATION VI. Establishment practices and policies: A. Paid v a c a tio n s. R ep o rt th e n u m b er of e m p lo y e e s who r e c e iv e d v a c a tio n pay d u rin g 1970 d ire c tly fro m th e e sta b lish m en t ac co rd in g to th e am o u n t o f pay. N u m b er of e m p lo y e e s r e c e iv in g — Em ployees U nder 1 w e e k 's pay No v a c a tio n pay 1 and under 2 w eek s' pay 2 and un d er 3 w eek s' pay 3 and u nder 4 w eek s' pay 4 and un d er 5 w eek s' pay 5 w eek s' pay or m ore O f f i c e ---------------------------N o n o f f ic e ---------------------B. Paid h olidays. Enter the n u m b er o f days per e m p lo y e e . If m ore th a n o n e p ra c tic e e x iste d for an e m p lo y e e group, rep o rt th a t w h ic h a p p lie d to th e g re a te st n u m b er in th e group. If th e g re a te st n u m b er o f e m p lo y e e s re c e iv e d no p a id h o lid ay s, e n te r "0 . " Fu ll days H alf days O ffic e e m p lo y e e s ---- -----N o n o ffice e m p lo y e e s------ C. Sick le a v e . Did th e e s ta b lish m en t h a v e a p ra c tic e or a p o lic y (ev e n th o u g h th e re m ay h av e b e e n no ex p e n d itu re ) for providing p a id sick le a v e for a n y — Yes D. C iv ic and perso n al le a v e . D id th e e sta b lish m en t hav e a p ra c tic e o r a p o lic y (e v e n th o u g h th e re m ay h av e b e e n no ex p e n d itu re ) for p ro v id in g p a id c iv ic le a v e (m ilita r y , ju ry , w itn ess, v o tin g , e tc . ) o r personal le a v e (su ch as for d e a th in fa m ily ) for any — O ffic e e m p lo y e e s --------------------------------------------------------------------------------------------------------------------------N o n o ffice e m p lo y e e s - ------------------------------------------------------------------------------------------------------------------- E. No B B O ffic e e m p lo y e e s -------------------------------------------------------------------------------------------------------------------N o n o ffice e m p lo y e e s -------------------------------------------------------------------------------------------------------------- B B L ife, a c c id e n t, and h e a lth In su ran ce. O ffic e e m p lo y e e s -------N o n o ffice e m p lo y e e s— Did the e stab lish m en t finance any of the follow ing insurance p lan s for— L ife S ic k n ess and accid e n t H o sp ita liz a tio n or m edical Y es No | | Yes No | | Y es No Q“] Yes No Y es No Y es f~^] No D id em p lo y e e s pay for p art of an y of th e se in su ran ce plans (answ er NO if payment was only for additional b e n e fits or c o v e ra g e for d e p e n d en ts) L ife O ffic e e m p lo y e e s--------- Y es Q j N o n o ffice e m p lo y e e s--- Y es F. Pension and r e tire m e n t plans. S i c k n e s s a n d a c c id e n t No □ No QJ] Y es L i ] Y es [ ^ ] H o s p it a l iz a t i o n or m e d ic a l Y es C7] Y es Q- ] N o C71 No No |— | No Q- ] D id th e e sta b lish m en t fin a n c e such a p lan fo r— O ffic e e m p lo y e e s ------------------------------------------------------------------------------------------------------------------N o n o ffice e m p lo y e e s -------------------------------------------------------------------------------------------------------------- B B D id e m p lo y e e s p ay fo r p art of any of th ese plans (answ er NO if p a y m e n t w as o n ly for a d d itio n a l be n e fits) O f f i c e e m p l o y e e s ------------------------------------------------------------------------------------------------------------------------------ N o n o ffice e m p lo y e e s -------------------------------------------------------------------------------------------------------------G. C o lle c tiv e b a rg a in in g . O ffic e e m p lo y e e s-------------------------------------------------------------------------------------------------------------------N o n o ffice e m p lo y e e s -------------------------------------------------------------------------------------------------------------H. B B Did u n io n -m a n a g e m e n t a g re e m e n ts co v er a m a jo rity of th e — B B R e g u la r w ork w eek , 1970. How m an y hours ( e .g ., 44, 40, 3 7 .5 , e t c . ) w ere n o rm a lly w orked e a c h w eek by th e m a jo rity of th e — O ffic e e m p lo y e e s------------------------------------------------------------------------------------------------------------------N o n o ffice e m p lo y e e s -------------------------------------------------------------------------------------------------------------- Hours p er w eek Hours p er w eek VII. Units included in report ( if d iffe re n t from th a t req u e sted in address b o x ): If this rep o rt re la te s to units in a d d itio n to th e o n e d e sig n a te d a t th e to p of p ag e 1, p le a se p rovide th e fo llo w in g in fo rm a tio n for e a c h u n it in clu d e d in the rep o rt. A v erag e 1970 e m p lo y m e n t L o catio n O ffic e N o n o ffice R rincipal product, s e rv ic e , o r a c tiv ity 168 BLS HANDBOOK OF METHODS ual industry studies the selection is based on establishment size strata. Thus, a reporting unit which is in a stratum at which the selection probability was set, for example, 1 out of 5 establishments, will be given a weight of 5, representing itself and four other establish ments in this same stratum. In the biennial studies of the entire private nonfarm sector, the probability of selection is proportionate to establishment employment size. Thus, a report ing unit employing 1000 workers, in a sample where the employment size probability base was set at 10,000, will be given a weight of 10, representing itself and other establishments having an aggregate employment of 10,000 workers. Under both procedures all establish ments over a certain size are included. The sample of nonrespondents for which data are collected by Bureau field representa tives is weighted appropriately to represent all nonrespondents. In the event that usable data cannot be ob tained from any unit visited in person, whether among the followup of nonrespondents or among large units often selected in the sample with certainty, its weight is assigned to units in the sample with the most similar industrysize-location characteristics. All estimated totals derived from such weighting procedures are adjusted further by the level of total employment or other appro priate measure for the survey year, based on data from the Bureau’s monthly establishment employment statistics program, in each of four broad economic regions. For instance, if the level of the aggregates, as derived from the weighting procedures, is 40,000,000 in an in dustry-region class and the corresponding level as shown by the employment statistics program is 44,000,000, the totals of the survey items would be multiplied by 1.1. The adjusted data represent all establishments, including those having fewer than four employees, in the in dustries studied.6 Some improvisation is necessary in the con struction of such annual benchmark totals. The monthly employment series provides data for only one pay period each month, and the esti mate of annual totals is made by multiplying by the average number of weeks in a year (52.14). Information from other sources, wherein a detailed breakdown by State or region is shown, is used as a basis for prorating the current employment (or hours) estimates into regional aggregates. Such sources include the Census of Manufactures and County Business Patterns (based on Social Security establish ment data.)7 Presentation The expenditure data on the individual ele ments of compensation are combined to give a measure of total employee compensation. The expenditure data for each individual element and for groups of elements are presented as a percent of total compensation, in cents per paid hour, and in cents per working hour. These measures are shown for all establishments, as well as for only those establishments that had an actual expenditure for a particular practice during the reference year. Hours data for working hours and for paid leave hours are presented as a percent of all paid hours. A. The expenditure ratios are calculated as follows: 1. Expenditures as a percent of total compensa tion for all establishments^: Aggregate expenditures for the practice -------------- ~ ~ ~Aggregate compensation in all a ~ --------r----------------n--------------------- X 100 establishments 2. Expenditures as a percent of total compensa tion for establishments reporting expendi tures = Aggregate expenditures for the practice ^ Aggregate compensation in establishments reporting expenditures for the practice. B. The expenditure rates are calculated as follows: 1. Expenditures in cents per paid hour for all establishments^: Aggregate expenditures for the practice Aggregate paid hours 2. Expenditures in cents per hour of work for all establishments^ Aggregate expenditures for the practice Aggregate paid hours minus aggregate paid leave hours 8 See chapter 2, “Employment, Hours and Earnings” . 7 U.S. Department of Commerce, Bureau of the Census, County Business Patterns (various years). EMPLOYER EXPENDITURES FOR EMPLOYEE COMPENSATION 3. Expenditures in cents per paid hour for es tablishments reporting expenditures= Aggregate expenditures for the practice Aggregate paid hours in establishments reporting expenditures for the practice. 4. Expenditures in cents per paid working hour for establishments reporting expenditures= Aggregate expenditures for the practice Aggregate paid hours minus paid aggregate leave hours in establish ments reporting expenditures for the practice. The distribution of workers by establishment expenditure ratios and rates is published, as well as the averages of ratios and rates. Ex penditures also are shown by selected es tablishment characteristics such as size, compensation level, unionization, and, for cer tain surveys, by area. Analysis The expenditure data describing the payroll or nonpayroll elements of compensation are presented in summary by this characteristic. However, the analysis of the data is related to the benefit function of each element. Thus, for analytical purposes, elements of compen sation that provide similar or interchangeable benefits are grouped together. The following groups of compensation elements are studied:1 1. Pay for working time; straight time pay, and premiums for overtime, weekend, holiday, and shift work. 2. Pay for leave time; vacations, holidays, miscel laneous leave of absence, and payments to vaca tion and holiday funds. 3. Payments for retirement programs; social se curity and private retirement plans. 4. Payments for health and related programs; life, accident, and health insurance, sick leave, and workmen’s compensation. 5. Payments for unemployment benefit programs; unemployment insurance, severance pay, and severance pay funds and supplemental unem ployment benefit funds. 6. Nonproduction bonuses. 7. Savings and thrift plans. Data are presented on the importance of various types of paid hours relative to all paid hours. Information is also published on the 169 number of paid holidays and number of weeks of paid vacation received by workers. Uses and Limitations Data from the surveys are used by employers in comparing their expenditure and hours prac tices with the averages for their industry and with those of other establishments having sim ilar or dissimilar characteristics (industry, size, location, union status, and average earn ings levels of workers). Labor and manage ment use the data in collective bargaining; and Government uses the statistics in the formulation of public policy, in producing esti mates of industry output per man-hour, and in making international comparisons. They also are used in deriving estimates of the amount and type of labor compensation and the nature of the hours for which compensation is re ceived by workers. As indicated earlier, the expenditures studied comprise the significant elements of employee compensation in American industry. The ag gregate of the expenditures studied represents total employee compensation. It does not, how ever, represent total labor cost which is a more encompassing concept and includes factors such as the cost of recruiting and training labor, the administrative expenses incurred in admin istering benefit programs, and many other ex penditures resulting from the use of labor as a factor of production. Some of these expendi tures may be important in particular estab lishments. The expenditures and hours data are subject to both sampling and reporting errors, the precise magnitude and direction of which are not known. Nevertheless, the errors resulting from sampling generally are considered to fall within acceptable confidence ranges; and re porting errors, to have a material effect on the accuracy of the results, would have to be in the same direction in substantially all of the cases. The omission of establishments with fewer than four employees from the samples may result in some bias, but it is very small since less than 1 percent of the workers in most industries are employed by such establish ments. 170 BLS HANDBOOK OF METHODS Technical References Number 1. U.S. Department of Labor, Bureau of Labor Statistics. Problems in Measurement of Expend itures on Selected Items of Supplementary Employee Remuneration, Manufacturing Estab lishments, 1953 (Bulletin 1186, 1956). A study of the availability of records, willingness and ability of industry to provide data, the quality of expenditure data, and other matters of methodology and definition. 2 . ______Employer Expenditures for Selected Supplementary Remuneration Practices for Pro duction Workers in Manufacturing Industries, 1959 (Bulletin 1308,1962). 3 . ______Production Workers in Mining Industries, 1960 (Bulletin 1332, 1963). 4 . ______Employer Expenditures for Selected Supplementary Remuneration Practices in F i nance, Insurance, and Real Estate Industries, 1961 (Bulletin 1419, 1964). 5 . ______Employee Compensation in the Private Nonfarm Economy, 1966 (Bulletin 1627, 1969). 6 . ______ 1967 Industry studies titled “ Employee Compensation and Payroll Hours.” Seven re ports (1 969); Banks (3 6 2 ); Research Laboratories (3 6 3 ); Confectionery (3 6 4 ); Struc tural Steel (3 6 5 ); Hotels & Motels (366) ; Laundries. Cleaning & Dyeing Plants (3 6 7 ); Men’s & Boys Shirts (368). 7 . ______Employee Compensation in the Private, Nonfarm Economy, 1968 to be published in 1971. Each expenditure bulletin contains descriptive information on the detailed procedures and techniques used in the study. 8 . ______Bauman, Alvin, “ Measuring Employee Compensation in U.S. Industry,” Monthly La bor Review, October 1970, pp. 17-23. A l v in B a u m a n Chapter 19. Work Stoppages Background Work stoppage statistics are compiled by the Bureau of Labor Statistics to provide a quantitative measure of the extent to which disputes between labor and management result in strikes or lockouts and of the immediate economic disruption resulting from such stop pages.1 When considered along with general economic measures, these statistics also serve at times as a broad indicator of the state of industrial unrest. The first attempt by any Federal agency to compile statistics on strikes was made in 1880,1 when the Bureau of the Census sent 2 questionnaires to employers and workers in volved in all disputes which were noted in the public press during the year. Information was received on 762 situations. Some data were obtained on the causes of strikes and their re sults, but not on the number of workers in volved or resultant man-days of idleness. The next collection of strike statistics was undertaken in 1887, when the Bureau of Labor, then in the Department of the Interior, ex amined files of newspapers, trade journals, and commercial periodicals for references to strikes for all years from 1881 to 1886. Staff members visited the areas where strikes were reported and obtained detailed information about each strike from every available person or source. Studies utilizing basically the same procedures subsequently were made in 1894, 1901, and 1906. As a consequence of these efforts, data were published for the 1881-1905 period on the number of strikes and workers involved, with breakdowns by industry and State; the number of establishments involved; and the percentage of strikes involving labor organi zations. No Federal agency collected national infor mation on stoppages occurring during the 1906-13 period. In 1914, relying exclusively on printed sources, the Bureau of Labor Sta tistics attempted to compile a record of all strikes and lockouts during the year. In the following year, the Bureau inaugurated a method for the collection of strike and lockout material which has been followed, with modifi cations, since that time. Briefly stated, the procedure was to send questionnaires to the parties involved in work stoppages, following receipt from the press and other sources of notices relating to these situations. Improvements in the program in 1927, in particular the procurement of data on the number of workers involved in all stoppages and the computation of man-days of idleness, inaugurated the modern series of monthly and annual strike data.3 Description o f the Series The present series on work stoppages covers all strikes and lockouts known to the Bureau of Labor Statistics and its cooperating agencies. It covers all that continue for 1 full day or shift or longer and involve six workers or more. These limitations on size and duration, somewhat arbitrary but of long standing, are necessary for reasons of efficiency, and, in part, because of the difficulty involved in defining, identifying, and securing information on strikes that last a few hours or less. The Bureau defines a strike as a temporary stoppage of work by a group of employees to express a grievance or enforce a demand. A lockout is defined as a temporary withholding of work by an employer (or a group of employ ers) to enforce terms of employment on a group of employees. Since 1922, the Bureau has made no attempt to distinguish between strikes and lockouts in its statistics; both types are included in the term “ work stoppages” and, for the sake of convenience in writing, in the term “ strikes.” 1 Throughout this chapter, the terms “work stoppage” and “ strike” are used interchangeably; both terms, unless other wise noted, also include lockouts. The definitions, terms, and classifications used by the Bureau in compiling work stop page data were adopted for statistical and research purposes and have no legal significance. 2 On the State level, the Bureau of Labor Statistics of Mas sachusetts, issued a report in 1880 on strikes in that State from 1825. In 1881, the Bureau of Industrial Statistics of Pennsylvania issued a report on strikes in that State from 1835. 3 For additional information on the early history of the work stoppage statistics program, see BLS Bulletin 651, Strikes in the United States, 1880 to 1936 (1938). 171 172 BLS HANDBOOK OF METHODS Although an employer-employee dispute is implicit in these definitions, some inclusions in the Bureau’s series relate only indirectly to this concept. For example, jurisdictional strikes and rival union disputes between two unions or more often have the employer on the sidelines. In a sympathy strike, the issue of the stoppage does not usually involve the immediate em ployer. Moreover, protest strikes against the actions of governmental agencies are not the result of a dispute between an employer and his employees. All stoppages, whether or not authorized by the union, legal or illegal, are counted. On the other hand, the Bureau’s series excludes strikes of American seamen or other workers in for eign ports and strikes of foreign crews in American ports. Also excluded are so-called slowdowns, where employees continue at work but at deliberately reduced production speed, and those instances in which workers report an hour or two late each day as a protest gesture or quit work several hours before clos ing time to attend rallies or mass meetings. The number of work stoppages occurring during a given period provides a measure of the frequency of disputes; the severity and effect of such actions are measured by the number of workers involved, duration, and the resultant man-days of idleness. The basic sta tistical unit in the Bureau’s series is the indi vidual strike or lockout. If groups of employees (regardless of their number and type and loca tion of employment) join in a work stoppage for a common objective, their action is classed as a single strike. The figure used for the number of workers involved in a strike or lockout is the maximum number actually made idle in the establishment or establishments directly involved. No dis tinction is made in arriving at this figure be tween the active participants in the strike, the number of union members or workers covered by an agreement, and those sent home by the employer when a stoppage by one group or department prevents plant operation. Man-days of idleness, like the number of workers involved, are based on the idleness at the establishment or establishments directly involved. Workers involved multiplied by work days lost equal total man-days idle. In instances where the number of workers idle varies dur ing the period of the stoppage, appropriate adjustments are made in this calculation. Al lowance is made in these computations for holidays and days not normally worked. Data Sources and Collection Methods The task of collecting strike data has two basic elements: (1) to learn of work stoppages when and wherever they occur, and (2) to ob tain the necessary facts regarding each situa tion as quickly as possible. Information about the existence of stoppages currently is obtained from various sources, in cluding: (1) clippings from daily and weekly newspapers throughout the country provided by commercial clipping services; (2) notices received directly from the Federal Mediation and Conciliation Service; (3) a periodic com pilation by the local offices of the State employ ment security agencies, provided through the U.S. Training and Employment Service of the U.S. Department of Labor; (4) a number of other State agencies, such as State mediation boards and labor departments; (5) various employers and employer associations; (6) in ternational unions and their publications; (7) firms under contract to the Atomic Energy Commission; and (8) other Federal agencies and commissions. Aside from the clippings from newspapers and other publications, most of these sources have been developed over a period of years. As a general rule, expansion in the Bureau’s “ lead” sources brings an increase in the num ber of stoppages reported, but has little effect on the total number of workers and man-days of idleness, because the added stoppages tend to be small. After the receipt of notices regarding the existence of work stoppages, the Bureau mails questionnaires to the parties involved to secure direct information on each stoppage. Should a reply not be received within 3 weeks, a sec ond questionnaire is mailed, and, in the case of continued nonresponse, a mailogram or tele gram may be sent, or an effort made to secure the necessary data by telephone. In some in 173 WORK STOPPAGES stances of nonresponse, field representatives of the Bureau secure the necessary data; in others, cooperating State agencies may be asked to contact the parties. The types of information sought by the Bu reau through its questionnaire have changed over the years, partly in response to changing needs. The primary function of these reports is to compile statistics, not to keep records on the strike activity of individual firms and unions. The separate questionnaires currently used for private and public sector disputes are shown on pages 174-176.4 Although strikes, by their very nature, are usually matters of public knowledge and of reporting by newspapers and other publica tions, the Bureau holds confidential the indi vidual reports submitted by employers and unions, as well as supplementary data collected through State or Federal agencies. The rules of confidentiality observed here are similar to those followed in other Bureau surveys. Estimating Procedures Since the Bureau is able to obtain informa tion on virtually all work stoppages within the scope of its definition, estimating is necessary only in the preparation of its monthly reports on the level of strike activity in the United States as a whole. The availability of reason ably accurate data on the larger stoppages at the time these estimates are prepared— ap proximately 4 weeks after the end of the month of reference— assures approximate conformity to the final statistics which are based almost exclusively on the parties’ replies. Monthly estimates are prepared on the num ber of stoppages, the number of workers in volved, and man-days of idleness. As there is a lag between the occurrence and reporting of a number of relatively small strikes, the number of stoppages beginning during a given month is estimated by increasing the number of strikes on which leads have been received by a percentage which is fixed for each calendar month. An estimate of the total number of stoppages in effect during the month is ob tained by supplementing the latter estimate by a percentage of the stoppages in effect during the prior month. In estimating the number of workers in volved and total idleness, efforts are made to obtain as much preliminary information as pos sible on the size and duration of individual large stoppages— those of at least 500 workers or 5,000 man-days of idleness. To the known figures for these large stoppages is added the product of the estimated number of smaller strikes and the average number of workers (or man-days) that previous experience indicates for such stoppages. In its preliminary reports, as well as in its final reports, the Bureau relates the man-days of idleness to the total estimated working time of all workers. The “ total economy” measure of strike idleness, which was instituted in 1967, includes government and agricultural employees as well as private nonfarm workers in its employment count as well as in the computation of idleness ratios. Before 1967, the BLS series excluded government and agricul tural workers from employment totals, but ac counted for time lost by these workers while on strike. This reevaluation of methods has improved the calculations of idleness and made the Bureau’s measurement of work stoppage intensity national in scope. Analysis and Interpretation The data presented in the parties’ reports are analyzed and classified according to a num ber of significant factors, briefly described here: (1) Each strike is assigned an industrial classification in accordance with the Standard Industrial Classification Manual prepared by the Bureau of the Budget.5 In those cases in which a stoppage affects workers in more than one industry, one of two procedures may be followed. If the stoppage is small, the strike is classified in the industry in which it was initiated; in large interindustry stoppages, a 4 A modified form of this questionnaire is used in the case of most jurisdictional disputes and those in coal mining. In the case of prolonged strikes, a less detailed question naire is sent to the parties periodically to determine the status of the stoppage. 5 See appendix B. 174 WORK STOPPAGES Form A pproved B udget Bureau N o. 4 4-R 2 1 0 B L S 817 (R e v . 1963) U. S. DEPARTMENT OF LABOR BUREAU OF LABOR STATISTICS W A S H I N G T O N , D . C . 20212 WORK STOPPAGE REPORT Your report w ill b e held in c o n fid e n ce P le a s e answ er a ll q u estion s T h is req u est for inform ation re la te s to: r L i . E m ployer Name and m ailing address 2 E s ta b lis h ments in v o lv e d in stop p a ge Number o f esta b lish m en ts d irectly in v olv e d or in w hich workers ob se rv e d p ick e t lin e s If m ore than one esta b lish m en t, u se rev erse s id e ; if one, enter b e lo w ; a. L o c a t i o n _______ __ ________________________________________________________ _________ _ b . Industry (In d ica te m ajor type o f a ctiv ity and a ls o p rin cip a l products or s e r v ic e s , e .g ., M ining-bitum inous c o a l; C o n stru ctio n -h ig h w a y s and stre e ts ; M a n u factu rin g-w ood en up h olstered furniture; W holesale trade - plum bing s u p p lie s ; T ra n sp orta tion -m otor freig h t.) Name 3. Union in v olv ed AFL CIO A d dress L o c a l N o. l D Other 4. D ates o f stop p a ge S toppage began on 5. Number o f workers a ffe cte d E m p loy ees returned to work on T O T A L workers id le d at lea st one full sh ift or day Settlem ent was rea ch ed on □ Settlem ent was ra tifie d on ( i f a p p lica b le) workers (IMPORTANT — Include a ll w orkers d irectly in v o lv e d in the sto p p a g e and w orkers made id le by la ck o f work in the same esta b lish m en ts or by o b se rv a n ce o f p ick e t lin e s . If exa ct figu res are not a v a ila b le , p le a s e p rovid e e s tim a te .) D id the number id le ch an ge s ig n ifica n tly during the sto p p a g e ? Y e s Q No Q (If “ y e s , ” p le a s e en ter ch a n g es in number id le and d a tes o f ch a n g es on r e v e r s e s id e o f s h e e t .) ........................... ............................................. d a ys 6. Normal w orkw eek prior to s t o p p a g e ..................................................................................................... 7. C ontract status (c h e c k o n e ) Stoppage occu rred Q In n eg otia tion o f first con tract or in obtaining union re co gn itio n Q I n ren egotiatin g con tract terms (expiration or reop en in g) Q During term o f con tract (ch an ge in con tract terms not in v o lv e d ) □ Other ( s p e c i f y ) . ....... - ...... - — ■■ 8 . MAJOR is s u e s in d isp u te (lis t in order o f im portance) 9. D id the agreem ent to return to work in clu d e a p rocedu re for handling any u n se ttle d m ajor is s u e s in v o lv e d in the sto p p a ge ( e .g ., by subm itting is s u e s to a rbitration)? Y es Q No Q If “ y e s ” , in d ica te the is s u e s and the p rocedu re agreed upon. 10. D id a F ed era l, State, or lo c a l governm ent a g e n c y , or a private m ediator, m ediate in this d isp u te or a s s is t in arranging the return to work? (C h eck more than one if a p p lic a b le ). F ed era l Q State Q L ocal Q P riv ate Q N one Q P le a s e id entify a s s is t in g governm ent a g en cy , if any (D ate) (Signature and title o f person making report) (Com pany or organ ization ) U se re v e rse s id e for any cla rify in g rem arks. BLS HANDBOOK OF METHODS 175 Supplementary Information for Items 2 and 5: If the stoppage involved more than one establishment or if idleness varied from period to period during the stoppage, please use the following space to indicate the number idle in each establishment and the variation in idleness at different dates. Include both workers directly concerned and those made idle because of dispute in the same establishment. IF E X A C T FIGURES A R E NOT A V A IL A B L E , P L E A S E FU R N ISH ESTIM A TE S. Industry or principal product Establishment involved and location (City, County, State) Approximate number Dates this number was idle of workers idle a full shift or more a full shift or more ' REM ARKS: * U. S. GOVERNM ENT PR INTING O F F IC E : 1 9 6 9 - 3 6 1 - 7 14 176 WORK STOPPAGES Budget Bureau No. 44-R 1397 A pproval expires 6 /3 0 /7 1 BLS 3006 U.S. DEPARTMENT OF LABOR WORK STOPPAGES REPORT BUREAU OF LABOR STATISTICS Government W a sh in gton , D.C. 20212 No. _____________ T h i s r e q u e s t L r in f o r m a t io n r e l a t e s t o : 1. Government Department, Agency, or Installation N am e: Address: F a c ility w here stoppage occu rred (n am e if d ifferen t from above):* Level (c h e c k one): [^ F e d e ra l j | S tate Function (check one): □ A d m in istrativ e services □ W elfare services □ Law en fo rcem en t and c o rrectio n 1 | Fire pro tectio n □ □ □ □ P □ Address: M u n ic ip a lity C ounty □ S a n ita tio n services Education Streets and highways Parks, re c re a tio n , lib ra rie s, e tc . School d istric t O ther (specify) ------------------------------- □ H ospitals and h e a lth services □ T ran sp o rtatio n and a llie d fa c ilitie s I j O ther u tilitie s j | O ther ( s p e c i f y ) -------------------- --------- 2. Union or Association A ffiliatio n : □ AFL-CIO 1 | O ther union I I E m ployee asso ciatio n 3. Dates of Stoppage D id th e organization c a ll or support th e work stoppa ge ? □ Yes □ No □ No in fo rm atio n Does this org an izatio n h av e o ffic ia l re c o g n i tio n ? □ Yes □ No and Workweek Stoppage b egan on: S ettle m e n t reach ed on: S e ttle m e n t ra tifie d on: Em ployees retu rn ed to work on: Sch ed u led w orkw eek prior to stoppage— Days Hours 4. Employees Affected T o ta l em ployees id led a t lea st one fu ll shift or day (IM PORTANT - In clu d e a ll em ployees d ire c tly involved in th e stoppage and em ployees m ad e id le by la c k of work in th e sam e fa c ilitie s or by observ an ce of p ick e t lin es. If f r t' •/ D id the num ber id led ch ange sig n ific a n tly during the stoppage? \ j Yes | [ No (If "yes" please e n te r changes in num ber id le and d ates o f changes on reverse side of this f o r m .) O c c u p a tio n a l c la ssifica tio n (ch e c k one or m ore): | | T eachers | | Nurses □ O ther professional and te c h n ic a l em p lo y ees □ C le ric a l □ P o lic e m en □ F ire m e n Q ] S a n ita tio n m en | j C raftsm en (specify) I I O ther blu e c o lla r and m anual □ O ther (specify) 5. Agreement Information Stoppage occurred (ch e c k on e): □ □ In a tte m p tin g to o b tain reco g n itio n In ne g o tia tin g first ag re e m en t □ □ D uring a g re e m en t term (ch an g e in term s not involved) In ren e g o tia tin g ag re e m en t (ex p iratio n or reopening) □ □ No fo rm al a g re e m en t involved O th er (specify) M ajor issues in dispute in order of im p o rtan ce: Did em ployees retu rn to work Q ] v o lu n ta rily , or Q under th e term s of a court o rd er or in ju n ctio n ? Did ag re e m en t to retu rn to work in clu d e a procedure for han d lin g any u n settled m ajo r issues involved in th e stoppage (e . g . , by su b m itta l to a rb itra tio n or factfin d in g )? | | Yes Q No If yes, note issues and procedures agreed upon on reverse side of this form . Did a governm ent ag en cy , or p riv ate in d iv id u al, or org an izatio n assist in arran g in g th e retu rn to work? (C heck one or m ore): F ed eral | ) S tate Local j j Private Please id en tify gov ern m en t agency Signature of person m ak in g report: T itle : D ep artm en t or organization: * If m ore th an one fa c ility was in v o lv ed , p lease en ter in fo rm atio n on reverse side of this form . c la rify in g rem a rk s, p a rtic u la rly on n atu re of stoppage (mass sick le a v e , or resig n atio n s, e tc ). I | None D ate: Also use reverse side for 177 WORK STOPPAGES stoppage is recorded for each industry affected, and the approximate numbers of workers and idleness are allocated to each. (2) The duration of each stoppage is taken as the number of calendar days from its be ginning to end. For stoppages which begin at a definite time and are terminated by a formal agreement, no problem arises in the determina tion of duration. However, some stoppages, for a variety of reasons, are never settled formally. These range from situations in which the workers gradually return to their jobs without a settlement to those in which the employer decides to go out of business. In cases of the former variety, the details of each individual situation are studied before a stoppage is terminated for statistical purposes; in the lat ter instances, the stoppage is terminated with the employer’s announcement of his decision to discontinue operations. On occasion, if actual settlement is reached later, the statistical rec ord of the stoppage is adjusted accordingly. (3) Number of establishments involved. The standard definition of establishment is used. (See appendix B.) An establishment is a single workplace, for example a factory, mine, or store. In a widespread strike of intercity bus drivers, truckdrivers, or railroad workers, the establishment is regarded as the terminal out of which the employees work; in a strike of seamen, the ship is the establishment; and in a strike of dockworkers, the individual dock or loading place is regarded as the place of work. (4) Geographical classification of stoppages followed State and city boundary lines, through 1951. Beginning in 1952, the compilation of data by Standard Metropolitan Statistical Areas superseded city boundary lines.6 In interarea stoppages, a stoppage is recorded in each area affected, and workers and man-days of idleness are allocated proportionately. (5) The issues in dispute in most strikes are many and varied, and do not always lend them selves readily to immediate and exact classifica tion. Stoppages are classified by major issue into the following broad groupings: (a) wages, hours, and supplementary benefits; (b) union organization and security; (c) job security; (d) plant administration; and (e) inter- or intra-union matters. Each of these groups is sub-divided into more specific categories.7 (6) Stoppages are classified by the con tractual relationship existing between the parties involved. The following four situations apply: (a) negotiation of the initial agreement; (b) renegotiation of an agreement; (c) agree ment in effect (new contract terms not in volved) ; and (d) no contractual relationship. (7) The union involved is another major classification. For this purpose, the union is the organization whose contract was involved or which has taken active leadership in the stop page. Disputes involving more than one union are classified as jurisdictional or rival union disputes or as involving cooperating unions. If unorganized workers strike, a separate classi fication is used. For publication purposes, union information is presented by major affilia tion of the union, i.e., AFL-CIO, or non affiliation such as “ Independent,” “ single firm,” or “ no union.” (8) The assistance of mediators, either gov ernmental or private, in the resolution of in dustrial disputes is recorded. (9) The manner in which stoppages are settled involves classification into the following categories: (a) those ending with a formal settlement; (b) those terminating without a formal settlement, and those in which work is resumed with either the old or new workers; and (c) those concluded by the employer’s decision to go out of business. (10) A classification also is made of the manner in which unsettled issues are to be resolved in those situations where strikes are terminated with the understanding that such matters will be handled after the resumption of normal operations. Presentation Publications in the area of work stoppages include monthly preliminary estimates, annual 8 See appendix C. 7 When a major change in the classification of issues was introduced in 1961, the Bureau included in its annual report for that year a method of linking the new classifications with the old. 178 BLS HANDBOOK OF METHODS reports, and special reports which are issued irregularly. Monthly preliminary estimates are issued approximately 30 days after the end of the month of reference in the form of press re leases. Such reports presently are prepared for the first 10 months of each year, with a pre liminary estimate of total strike activity dur ing the year being issued a few days after the close of the calendar year. Selected final tabulations of strike activity are presented in a release which usually is is sued in June. An annual bulletin,8 containing detailed information on the characteristics of work stoppages during the prior year, is pub lished each winter. Special reports containing historical work stoppage data by industry and area, or an analysis of a particular aspect of strikes, are issued irregularly. This latter category also may include Bureau reports of a nonstatistical nature, including chronologies of “ national emergency” disputes arising under the terms of the Taft-Hartley Act. Uses and Limitations The use of strike statistics as an indicator of industrial unrest has been the traditional rea son for their compilation in the United States and in other industrialized countries.9 Whether they serve this elementary purpose today is open to question.1 In any event, some qualifica 0 tions must be taken into consideration in this use. The willingness of workers to strike as a protest against existing conditions may be en couraged or deterred by outside influences, such as the employment situation, the state of the business cycle, and possible political or public reaction. Within the plant, the strength of the union or of employer opposition may influence both the willingness of the workers to start a strike or to extend its duration. Whether as a measure of industrial unrest or the state of labor-management relations, strike statistics are necessary for Federal, State, and municipal government agencies, particularly those concerned with labor affairs. Unions, em ployers, and employer associations use strike data to assess their own experiences, and busi ness and civic organizations are concerned with their community promotional possibilities. Schools, particularly those teaching courses in industrial relations, and industrial relations counselors also find strike statistics useful. Finally, the press is interested, since strikes and information about them are newsworthy. Although it is virtually certain that the Bu reau is able to locate, and obtain information on, the larger work stoppages, some small strikes undoubtedly escape notice each year. While these omissions do affect data on number of strikes, the statistics on workers and mandays of idleness are virtually complete. As has been noted, the addition of new sources of in formation has not materially changed these latter figures, but these new sources have acted to lessen slightly the degree of comparability in the number of strikes reported from period to period. It follows that the narrower the classification of strike data, the greater is the chance of a significant omission. For example, while the figures for individual States may be taken as reasonably complete, the figures for a specific industry group within a State may be appreciably affected by the ommission of one strike. Secondary idleness is not measured; that is, the figures do not cover those employees made idle in other establishments or industries as a result of material or service shortages result ing from a work stoppage. At times, the idle ness of employees directly involved in a strike may be considerably less than the idleness of other workers brought about indirectly. No satisfactory method, however, has been found to measure or estimate such indirect effects adequately. The Bureau does not attempt to measure the cost of strikes in terms of the amount of pro duction and wages lost. The calculation of cost 8 Since 1949, these bulletins have been titled Analysis of Work Stoppages (year). 9 For a discussion of the methods used in compiling strike statistics in other countries, see Methods of Compiling Statis tics of Industrial Disputes, Geneva, International Labour Office, Studies and Reports, Series N, No. 9 (1926). See also A. M. Ross and P. T. Hartman, Changing Patterns of In dustrial Conflict, New York, John Wiley & Sons (1960). Data on work stoppages in other countries are published annually in the Year Book of Labour Statistics, Geneva, In ternational Labour Office. 1 See article by Joseph W . Bloch, “ The Strike and Discon 0 tent,” Monthly Labor Review, June 1963, pp. 645-651. 179 WORK STOPPAGES involves many complex and interrelated factors for which information is not readily available, including such matters as production schedules before and after the stoppage, diversion of out put or services to other plants or employers, the flow of raw materials, and the amount of overtime worked before and after the strike.1 1 The problem is magnified beyond statistical control if secondary costs are to be accounted for. 1 1 A framework for the measurement of the cost of strikes is developed by Neil W. Chamberlain and Jane Metzger Schilling in The Impact of Strikes: Their Social and Economic Costs, New York, Harper & Bros. (1954). Technical References Number 1. International Labour Office, Methods of Compiling Statistics of Industrial Disputes, Studies and Reports, Series N (Statistics), No. 9, Geneva (1926). Compares the methods used to compile statistics on industrial disputes, and outlines standards by which some degree of international comparability may be secured. 2. Peterson, Florence, “ Methods Used in Strike Statistics,” Journal of the American Statistical Association, pp. 90-95, 1937. Summarizes the definitions and methodology utilized by the Bureau of Labor Statistics in its work stoppage statistics program. 3. U.S. Department of Labor, Bureau of Labor Statistics, Analysis of Work Stoppages (year). The Bureau of Labor Statistics issues annually a detailed statistical bulletin analyzing work stoppages in the United States. 4 . ______Strikes in the United States, 1880 to 1936 (Bulletin 651, 1938). Contains a history of statistics on strikes and lockouts in the United States and major statistical data available from the earliest recorded date through 1936. —E dward D. Onanian Chapter 20. Collective Bargaining Agreements Background Collective bargaining agreements and related documents setting forth the provisions of em ployee-benefit plans provide a valuable source of information on industry wage practices, supplementary benefits, job and union security, the timing of wage negotiations, the nature of plant operations and working conditions, and many of the day-to-day aspects of employeremployee and union-management relationships. The Bureau of Labor Statistics has utilized these basic industrial relations materials for public and Government informational purposes in three major ways, by: (1) maintaining a file of current agreements open to public inspection and inquiry; (2) preparing reports which re produce representative agreement provisions or the variety of provisions relating to similar problems, or digests of selected identified agreements or benefit plans; and (3) by pre paring studies measuring the prevalence and characteristics of specific types of agreement and benefit plan provisions or of other aspects of collective bargaining such as multiemployer bargaining. The development of industrial relations prac tices that are now widely prevalent is reflected in the Bureau’s studies over the years. The Bu reau’s interest in the collection and analysis of union agreements dates back almost 70 years.1 Systematic efforts to collect agreements began in 1912. The first of a number of BLS bulletins devoted entirely to the subject of collective bargaining agreements appeared in 1925. A large number of reports and bulletins on a wide variety of industrial relations subjects have since been published. The Bureau’s responsibility in the field of agreement collection and analysis received additional sanction and guidance in the Labor Management Relations (Taft-Hartley) Act, 1947, section 211, which reads as follows: Sec. 211. (a) For the guidance and information of interested representatives of employers, employees, and the general public, the Bureau of Labor Statistics of the Department of Labor shall maintain a file of copies of all available collective bargaining agreements and other available agreements and actions thereunder settling or adjusting labor disputes. Such file shall be open to inspection under appropriate conditions pre scribed by the Secretary of Labor, except that no specific information submitted in confidence shall be disclosed. (b) The Bureau of Labor Statistics in the Depart ment of Labor is authorized to furnish upon request of the [Federal Mediation and Conciliation] Service, or employers, employees, or their representatives, all available data and factual information which may aid in the settlement of any labor dispute, except that no specific information submitted in confidence shall be disclosed. Concepts and Scope Although the substance of collective bargain ing rests partly upon a foundation of unwritten industry, company, and union practices, and upon various legal requirements, the basic unit in agreement collection and analysis is the writ ten agreement itself. The agreement may cover a single plant, a number of plants of a multi plant company, or a number of companies, in some cases over a thousand, bound together formally or informally in an association for collective bargaining purposes. It may express conditions of employment in simple terms, and leave much of the administrative details and other matters to the day-to-day relationships between the parties, or it may attempt to cover all details and, thus, leave as little as possible to later bargaining or haggling. Agreements vary in size from a few sheets to over 300 pages of a pocket-sized booklet. They reflect the diver sity of employment conditions among indus tries and companies and of the scope of the issues over which bargaining takes place, as well as differences in the degree of precision sought and the language used. Estimates of the number of agreements cur rently in effect range up to 155,000. The num ber of workers covered by agreements is estimated at about 20 million. The Bureau presently maintains a file of approximately 1 A bulletin of the Department of Labor, Number 42, Sep tember 1902, included this note: “ It is the purpose of this Department to publish from time to time important agree ments between large bodies of employers and employees with regard to wages, hours of labor, etc. The Department would be pleased to receive copies of such agreements whenever made.” (p. 1057) Between 1888 and 1903, the Bureau of Labor (now the Bureau of Labor Statistics) had independent status as a Department of Labor, under the direction of a commissioner. 181 182 BLS HANDBOOK OF METHODS 6,500 current agreements in the private sector, covering about 9.8 million workers.2 All in dustries are represented in the file except rail roads and airlines. Since railroads and airlines are required to submit copies of agreements to the National Mediation Board, the Bureau does not attempt to collect these agreements. As a result of the recent growth of collective bargaining in the public sector, the file has been expanded to include a variety of docu ments covering employees of the Federal Gov ernment, States, counties, cities, and special jurisdictions. These documents range from traditional collective bargaining agreements through memoranda of agreement to executive orders, administrative promulgations, and leg islative actions, which are clearly the result of bilateral negotiations. The public file is growing daily, presently exceeds 1,000 docu ments at all government levels, and involves more than 1 million public employees. The Bureau’s quantitative analysis of agree ment provisions covers virtually the entire range of issues dealt with in collective bar gaining.3 The basic assumption underlying such analysis is that the variety of subjects can be defined, classified, and counted. In its general analysis of agreements, as dis tinct from special industry studies, the Bureau is concerned with these major objectives: (1) the presentation of data by industry group and for manufacturing and nonmanufacturing as a whole, (2) the presentation of data by region or union affiliation if the subject requires it, (3) a realistic measure of representativeness in the agreements studied, and (4) the study of practices which are (as yet) relatively uncom mon in collective bargaining agreements. For agreements covering public employees, data are presented by the type of government ac tivity (police department, veterans administra tion, etc.) and for cities, by population size. Methods o f Collection and Analysis Collection o f Agreements The selection of agreements for the file is currently based on two guides: To enlarge to the fullest the opportunities for public and governmental use of the file,4 and to provide a diversified collection of agreements for special reports, which the Bureau occasionally is called upon to prepare. The extent to which these objectives are fulfilled is obviously affected by the size of the file. A third guide— that of con structing a file which truly represents all public and private agreements and thus provides a firmer basis for sound generalizations on all agreements— has long been a goal of the Bureau. The maintenance of a current file of agree ments is a continuous undertaking because of two factors: (1) The typical agreement has a fixed duration, after which it is no longer considered current (unless notice of renewal without change has been received), and (2) submission of agreements to the Bureau is voluntary on the part of employers or unions. To allow for the ratification and the printing of new agreements, requests for copies are mailed about 2 or 3 months after the expira tion date indicated in the previous agreement or upon other notice of contract change. As in other phases of the Bureau’s work, the volun tary cooperation of employers and unions is of utmost importance. Any restrictions imposed by respondents on the public use of agreements are observed scrupulously by the Bureau. To facilitate the use of the file in accordance with the types of requests customarily made, each agreement received is coded for a series of identifying features, which include union, loca tion, number of workers covered, industry, and effective and expiration dates. For public em ployee documents, the level of government and the government activity also are coded. Agreement Analysis The Bureau’s utilization of the private sector agreements it collects has moved through dif 2 During the early postwar period, the number of agree ments on file exceeded 12,000. See Program Note by Joseph W. Bloch, “ Union Contracts— A New Series of Studies,” Monthly Labor Review, October 1964, pp. 1184-1185. 4 The agreements file is located in the Washington Office of the Bureau’s Division of Industrial Relations. Agreements submitted to the Bureau with a stipulated limitation on public use are not available for inspection. Requests for informa tion concerning specific agreements or agreement clauses are accommodated, depending upon the nature of the request, within the limits of staff resources. COLLECTIVE BARGAINING AGREEMENTS ferent stages over the years, in pace with, or controlled by, the increasing prevalence and maturity of collective bargaining. During the early years, significant agreements were re produced in their entirety. As collective bar gaining spread, and the size and representative character of the Bureau’s file increased, atten tion was directed towards reproducing and analyzing the variety of agreement clauses re lating to similar subjects, culled from a large number of agreements. Although illustrative clauses continue to be utilized in most of the Bureau’s agreement studies, major emphasis during recent years has been devoted to meas uring the prevalence and characteristics of particular provisions and of types and levels of benefits. The Bureau’s Bulletin 1425 series represents its most comprehensive efforts to date. The Bureau also has undertaken a broader, more extensive rather than intensive analysis, in which it searches the agreement to measure the prevalence, but not the detailed characteristics of more than 100 different col lective bargaining provisions. In these kinds of analysis, problems relating to techniques of coding and analysis come to the fore. In 1948 and 1949, when the Bureau’s file consisted of more than 12,000 agreements, it was decided that a sample of 3,000 agreements would be feasible. The selection of specific agreements was based on a number of factors, including industry, agreement coverage, loca tion, union representation, and bargaining practices. Limited data upon which to base a representative selection of agreements were compensated for, at least in part, by extensive experience with collective bargaining practices on the part of the sample selectors. During subsequent years, however, available staff resources were not sufficient to deal with a sample of this size. The feasibility of recon stituting a sample of 1,500 to 2,000 agreements, which had become the maximum work load, and of assuring appropriate safeguards against deterioration, was rejected as being beyond the resources of the staff and the available data. The most advantageous alternative was to base the private industry agreement studies on all agreements covering 1,000 workers or more and, thus, to avoid sampling. The Bureau’s file 183 already contained almost all of these; the Bu reau’s monthly report, Current Wage Develop ments, was a ready source of information on those that were not included. The total number of workers covered by agreements of this size (now between 1,700 and 1,900) is about 7.7 million and represents a very substantial worker coverage in agreement studies. The number of establishments covered is not known.5 A key analysis list containing all private in dustry agreements covering 1,000 or more workers, although perhaps not the ideal cover age, has definite advantages: (1) It achieves maximum worker coverage in the studies for a given investment of staff resources; (2) it provides a simple, objective measure of the coverage of the studies; (3) it permits the presentation of various combinations and breakdowns of the data without the necessity of complicated weighting schemes and without the bias resulting from the lack of proper weighting; (4) it is safeguarded against obso lescence, since the Bureau is best able to keep abreast of changes in agreements of this size; and (5) it has a significant meaning to users of these studies. The distinguishing feature of agreement analysis is that it deals mainly with legal lan guage, which requires interpretation, rather than with numbers or other universal, sharply defined attributes. The process of analysis con sists of interpreting provisions, reducing them to numbers (codes), aggregating the numbers, and converting the aggregates back to types of provisions and prevalence. Data are presented in terms of number of agreements and number of workers covered. Since agreement provisions on the whole are notable more for their variety of expression and details than for their uni formity, the process of analysis becomes a simplification process by which some of the original content and variety is lost. Under such circumstances, the planning of studies acquires 5 The distinction between size of agreement (employees cov ered) and size of establishment is important. A substantial proportion of these agreements are association negotiated and cover a large number of small establishments. Two agree ments, for example, involving the United Mine Workers ( I n d . ) , cover most of the anthracite and bituminous coal mines in the country. Some association agreements in New York State cover more than a thousand firms. 184 BLS HANDBOOK OF METHODS a special importance if significant differences are not to be buried. The keystone of agreement analysis study is obviously the interpretation of the agreement and the assignment of the predetermined code number. For some subjects, an agreement must be read in its entirety; for others, only a por tion. Long and legal provisions must be reduced to their essentials. Since the interpretation of agreement provisions is often a troublesome matter for the parties themselves (as reflected in the widespread adoption of provisions for arbitration), misinterpretations undoubtedly occur. These are kept to a minimum by a staff experienced in agreement analysis and by con tinuous efforts to assure consistency of in terpretations. Uses and Limitations The studies of agreement provisions are of practical use to public and private employers and unions engaged in collective bargaining, to arbitrators and factfinding boards, to admin istrators of company wage and industrial relations programs, and to legislators and Gov ernment officials. Persons not directly involved in collective bargaining or in related admin istrative activities (e.g., teachers and students of labor problems, writers for newspapers and trade and technical journals, and foreign ob servers) find value in the broader aspects of employer-employee relationships revealed in these studies. The limitations of these studies of agreement provisions and employee-benefit plans are de termined, in large part, by their application. For studies of paid holiday provisions or other supplementary benefits, the fact that these studies cover only the area of collective bar gaining may constitute a limitation on gen eralizations applying to all workers but not necessarily on their uses in collective bargain ing or in wage and employee administration. On the other hand, these studies do not show locality practice, which may reduce their use fulness for some collective bargaining purposes but not for broad generalizations relating to workers under collective bargaining. Additional limitations of agreement pro vision studies are inherent in the selection of agreements for study— the exclusion of rail road and airline agreements and of agreements covering fewer than 1,000 workers— and in the technique of analysis, as indicated previously. Limitations are also connected with the par ticular subjects studied, which are pointed out in each study. A fundamental limitation which must be borne in mind in connection with certain studies (e.g., grievance procedure) is that they relate to written policy rather than actual practice. Practices which are not provided for in the agreements but are, instead, matters of company policy going beyond the agreement, traditional industry policy, informal acceptance by management and unions, or arbitrators’ de cisions, can be neither detected nor measured in agreement analysis. 185 COLLECTIVE BARGAINING AGREEMENTS Technical References Number 1. Bloch, Joseph W . Union Contracts— A New Series of Studies. Monthly Labor Review, October 1964, p p . 1 1 8 4-1 1 8 5 . 2. U.S. Department of Labor, Bureau of Labor Statistics, Major Collective Bargaining Agree ments (Bulletin 1425 Series). 1425-1, 1964: Grievance Procedures 1425-2, 1965: Severance Pay and Layoff Benefit Plans 1425-3, 1965: Supplemental Unemployment Benefit Plans and Wage-Employment Guarantees 1425-4, 1966: Deferred Wage Increase and Escalator Clauses 1425-5, 1966: Management Rights and Union-Management Cooperation 1425-6, 1966: Arbitration Procedures 1425-7, 1969: Training and Retraining Provisions 1425-8, 1969: Subcontracting 1425-9, 1969: Paid Vacation and Holiday Provisions 1425-10, 1969: Plant Movement, Transfer, and Relocation Allowances 1425-11, 1970: Seniority in Promotion and Transfer Provisions 1425-12, 1970: Administration of Negotiated Pension, Health, and Insurance Plans 3 . ______ Bulletin 1353, Major Union Contracts in the United States, 1961 (1962). 4 . ---------- Bulletin 1686, Characteristics of Agreements Covering 5,000 Workers or More (1970). ■Jo s e p h W . B loch Chapter 21. Union Membership Background The Bureau of Labor Statistics’ interest in union membership as a significant social and economic indicator is of long standing. The first BLS publication listing membership fig ures for national and international labor unions in the United States was probably the Hand book of American Trade-Unions, issued in 1926 (BLS Bulletin 420), followed by a similar pub lication in 1929 (BLS Bulletin 506) and an extensive revision in 1936 (BLS Bulletin 618). The Handbooks, however, devoted little space to overall membership figures or trends; their main emphasis was on the origins and early history of particular unions, their government and structure, trade jurisdiction, and types of beneficial activities. In 1939, on a modest scale, the Bureau began to publish an annual trade union directory, but it was not until 1948 (Bulletin 937) that the listing of union officers, headquarters’ ad dresses, etc., was supplemented with an entry for each union on membership and number of local unions, and a compilation of total mem bership. In subsequent years the information sought from national and international unions has expanded considerably. Data on women members first appeared in the 1953 Directory, and separate tabulations for areas outside the United States (Canada, Puerto Rico, etc.), in the 1955 edition. Since 1957, each Directory has carried information on the number and proportion of members who are white-collar workers and on those in major industry groups. State figures were introduced on a limited basis in 1959, when AFL-CIO central bodies were asked to furnish estimates on the number en rolled by Federation affiliates. In its 1965 Di rectory, the Bureau showed State figures as reported by national and international unions. Many of the items referred to have been re fined since they were first introduced, and the accumulated information now permits analysis 1 For a membership survey of these unions see Unaffiliated Intrastate and Single-Employer Unions, 1967 (BLS Bulletin 1640, 1969). 2 The names of all reporting unions appear in the Depart ment’s Register of Reporting Labor Organizations, last issued in 1968. of trends in total membership and several of its components. Data Sources and Collection For part of the Bureau’s biennial Directory of National and International Labor Unions in the United States, unions meeting the criteria noted are asked to report the average number of dues-paying members or the number of members in good standing for the 2 most recent years, including members outside the United States; the proportion of white-collar and women members; breakdowns by major in dustry and by State, and other membership information. (See copy of questionnaire on pp. 188-191.) To be included in the Directory, a union must be an affiliate of the AFL-CIO or, in the case of unaffiliated unions, a party to collective bargaining agreements with different employers in more than one State. Among the unions in the latter category are such un affiliated unions as the Teamsters, the Mine Workers, and the Auto Workers. In addition, the Directory accounts for all unions of Fed eral Government employees that have received “ exclusive recognition” as specified in Execu tive Order 11491. Thus, by definition, the Bu reau excludes from its Directory those unions whose activities are confined to a single locality or to a single employer.1 Although the Bureau sends questionnaires to all multilocal unions who have submitted reports to the U.S. Depart ment of Labor’s Office of Labor-Management and Welfare-Pension Reports, as required by the Labor-Management Reporting and Disclosuere Act,1 possibly some small unaffiliated 2 unions, interstate in scope, escape attention. These inadvertent omissions do not affect mem bership totals in any significant way. When some unions are unable to furnish in formation for one or more of the questionnaire items, estimates are derived from other sources on file, notably union periodicals, convention proceedings, financial statements, and collec tive bargaining agreements. No sampling pro cedures are used; the data are based on the 187 188 BLS HANDBOOK OF METHODS B u d g e t B u re a u N o. 4 4 *S 7 0 0 2 7 A p p r o v a l E x p ir e s D e c . 31, 1971 B L S 2441 U. S. D EP A R T M EN T OF LABO R Bureau of Labor Statistics Washington, D. C. 20212 DIRECTORY OF NATIONAL AND INTERNATIONAL LABOR UNIONS IN THE UNITED STATES, 1971 I. T . I.phono number U n io n a nd O ffic e r Id e n tific a tio n : and area code: 1. Affiliation □ ( Check appropriate box 2. President: (□ M r. 3. Secretary-Treasurer: □ AFL-CIO Mrs. 1 1 Mr. □ None □ Other (s p e c ify ) 1 | Miss Year first elected to this office: 1 1 Mrs. 4. Person in charge of organizing activities: □ Miss ( □ Mr. Year first elected to this office: [ □ Mrs. □ ] Miss ___________________________________________________________________ Tide:. Mailing address if different from headquarters (Street, City, State, Zip Code): 5. Research Director: I I Mr. I I Mrs. |_ |Miss ______________________ _ Mailing address if different from headquarters ( Street, City, State, Zip Code): 6. Education Director: Q M r. □ Mrs. □ M iss______________________ Mailing address if different from headquarters (Street, City, State, Zip Code): 7 . P e rs o n in ch a rge o f s o c ia l in su ra n ce (health, insurance, pension, e t c .) a c t iv i t ie s : □ Mr. □ Mrs. □ M i s s _________________________________ T itle :___ Mailing address if different from headquarters (Street, City, State, Zip Code): 8. Person in charge of legal activities: □ ] Mr. □ ) Mrs. ( □ Miss ___________________________________________________________________ T id e :____ Mailing address if different from headquarters (Street, City, State, Zip Code): 9. Person in charge of legislative activities: □ ] Mr. □ ) Mrs. [ □ Miss ------------------------------------------------------------------------------------------------------- Title: Mailing address if different from headquarters (Street, City, State, Zip Code): 10. Person in charge of public relations activities: ( □ Mr. ( □ Mrs. ___________________________________________________________________ □ ] Miss T i d e :___ Mailing address if different from headquarters (Street, City, State, Zip Code): 189 UNION MEMBERSHIP II. Conventions and Publications: 1. Frequency of conventions: [ / j Annual Semi-annual 2. Next convention:______________________ Month Day Year Other ( s p e c ify ) ___ ____________________________________________________ (C ity, State) 3. Name of official publication(s) III. Biennial How often published Editor Affiliated Bodies: Indicate number of locals in operation at the end of 1970. IV. Membership: 1. Indicate annual average dues-paying membership count for 1969 and 1970. If complete returns for 1970 are not available, use 9 or 10 month average: 1969 members 2. If your records do not permit an annual average dues-paying good standing or those carried on your rolls: 1969___________________ members 1970___________________ members membership count, please indicate the number ofmembers in 1970___________________ members 3. Indicate if retired members are included in 1 or 2 above: Q Yes Q No If yes, indicate number of retired members included: _____________________members* 1 V. Classification of 1970 membership: 1. Indicate approximate percentage of membership who are women:______________________________________ % ( i f none, enter zero) 2. Indicate approximate percentage of membership in the following “ white-collar” categories: Professional and/ortechnical C lerical------------------------------Sales-----------------------------------Total “ white-collar” - ------- __________ % % % % ( i f none, enter zero) 3. Industry composition of union membership. Indicate the approximate percentage of all union members working in establishments in each of the following industry groups: Manufacturing: Ordnance and a ccessories--------------------------------------------------------------------------------------------------Food and kindred products (incl. b e v e r a g e s ) --------------------------------------------------------- -----------Tobacco manufactures---------------------------------- r ------------------------------------------------------------------Textile mill products-------------------------------------- ------------------------------------------------------------------Apparel and other finished products made from fabrics and similar materials-------------------------Lumber and wood products, except furniture------------------------------. . . . . . ----------. . . . . . . . . ----Furniture and fixtures---------------------------------------------------------------------------------------------------------Paper and allied products-------------------------------------------------------------------------------------------------Printing, publishing and allied industries------- ------------------ . . . . . -----. . . . . . . . . . . . . . . . . . . . . . Chemicals and allied products---------------------------------------------- ---------------------------------------------Petroleum refining and related industries----------------------------------------------------------------------------Rubber and misc. plastics products------------------------------------------------------------------------------------Leather and leather products------------------- --------------------------------------------------------------------------Stone, clay, glass, and concrete products ----------------- ----------- --------------------------------------------Primary metals industries-------------------------------------------------------- -------------- -------------Fabricated metal products, except ordnance, machinery and transportation equipment ----------Machinery, except electrical...........- ------------------------------------------------ ------------- r -------------------Electrical machinery, equipment and supplies---------------------------------------------------------------------Transportation equipment---------------------------------------------------------------------------------------------------Professional, scientific and controlling instruments; photographic and optical goods, watches and clocks-------------- — ---------- ----------------------------------------------------------------------------------------Miscellaneous manufacturing industries----------------------------------------------------------------------------- 190 BLS HANDBOOK OF METHODS - 3- V. C la s s if i c a t i o n o f 1 9 7 0 m e m b e rsh ip : (Continued) 3. Indicate industry composition of union membership N o n m a n u fa c tu rin g : Mining and quarrying (including crude petroleum and natural gas production) ---------------------------------Contract construction (building and sp e c ia l trade) ---------------------------------------------------------------------------Transportation services (including railroads, air, bus, truck and water transportation, and allied ------------------- % ------------------- % s e r v i c e s ) ----------------------------------------------------------------------------------------------------------------------------------------- -------------- Telephone and telegraph---------------------------------------------------------------------------------------------------------Electric, gas, and sanitary services (including w ater) --------------------------------------------------------------------Wholesale and retail trad e--------- -------------------------- ---------------------------------------------------------------Finance, insurance and real estate-------------------------------------------------------------------------------------------Service industries (including hotels, laundries and other personal s e r v ic e s , repair s e r v ic e s , ------------- motion pictu res, amusements and related s e r v ic e s , hospitals, educational institutions, non profit membership organizations) ---------------------------------------------------------------------------------------------------- Agriculture and fishing------------------------------------------------------------------------------------------------------------Nonmanufacturing (classification not available) -------------------------------------------------------------------------------- -------------- ------------------------- -------------- -------------------------- G overnm ent: F ed era l----------------------------------------------------------------------------------------------------------------------------------State and lo c a l------------------------------------------------------------------------------------------------------------------------Total..................................... ------------------------.........100 4. State distribution of union membership. Indicate the approximate number or percentage of members in each of the 50 States. Alabama----Alaska -----Arizona — Arkansas - California Colorado -Connecticut Delaware Florida-----Georgia-----Hawaii-----Id aho------Illinois-----Indiana----Io w a -------■ K a n s a s-------- Kentucky---Louisiana -• Maine----------------Maryland—District of Columbia — Massachusetts---Michigan-----------Minnesota - --------Mississippi--------M issouri-----------Montana-----------N ebraska----------Nevada -------------New Hampshire New J e rs e y ------New M ex ico ------New York............. North Carolina North Dakota ---O h io----------------- Oklahoma------------------Oregon----------------------Pennsylvania-----------Rhode Isla n d -----------South Carolina----------South D akota-----------Tennessee ---------------Texas ....................... U tah............................. Vermont--------------------V irginia-------------------Washington---------------West Virginia------------Wisconsin-----------------Wyoming................... No. or percentage not accounted for in the U .S .................... 5. For any area outside the United States, please indicate the number of dues-paying members and the number of local unions in existence as of the end of 1970 or any other appropriate current period: Location Canada Puerto Rico Canal Zone Other (s p e c ify ) Approximate number of union members Number of local unions 191 UNION MEMBERSHIP - 4V I. 1. T e rm o f O ffic e r s: * Are officers required to retire at a specified age? National officers: I I Yes [_J No; If yes, age___________ Local officers: I I Yes 3 If yes, age___________ No; 2. Are the number of terms an officer may serve limited? National officers: 1. Yes 3] No; If yes, number of terms Local officers: V I I. [31 3] Yes 3 1 No; If yes, number of terms. C o lle c t iv e B a r g a in in g A g re e m e n ts: Indicate the number of basic collective bargaining agreements with employers. Exclude supplemental and pension, health and insurance agreements: Total_______________ agreements. United States_________________ agreements. 2. (a) Indicate number of different employers covered by collective bargaining agreements---------------------- ------------------------------.employers United States o n ly ------------------------------ -employers (b) If more than one employer, are the employers located in at least two States?------------------- ------------ 3 ] Yes 3 No 3. Indicate the number of workers covered by these agreements. Include nonmembers in the bargaining units: Total________________ workers. United States________________ workers. 4. Indicate the number of agreements in the United States that: Expired in 1970_________________agreements. Will expire in 1971_________________ agreements. 5. Indicate the number of agreements in the United States that provide a wage reopener in: 1970 _______________agreements. 1971 _______________agreements. V I I I . U n :on Staff: Indicate the number of full-time employees on the payroll of the national in the United States. Exclude elected officials and employees on the payrolls of local unions and intermediate bodies: Managers and administrators---------- --------- --------------- Professionals-------------------------------------- -- --------------- Clerical and secretarial-------------------------- --------------- Organizers and representatives---------------- --------------- Others ( s p e c i f y ) .....................- ..................... .................. May we have your comments regarding the present Directory and proposals for changes in future editions? Name of person reporting Title Date 192 BLS HANDBOOK OF METHODS entire universe of national and international unions, as defined. Presentation The data for each union and summaries are presented in the Directory and in articles in the Monthly Labor Review. Wherever appropriate, separate data are presented for AFL-CIO affiliates and unaffiliated unions and cover ques tionnaire items such as total membership for the last 2 years, members in and outside the United States, size of unions, women and white-collar members, and industry distribu tion. Gains and losses over the past decade are analyzed, and major unions are ranked accord ing to size at particular points in time. Of con siderable interest to users of the Directory are the changes of union membership related to changes in the total labor force and to em ployees in nonagricultural establishments. Data by industry and on women and white-collar members permit rough approximations on the extent of union penetration compared with earlier periods. In addition to statistical summaries, the Directory also contains appendixes which list, for each union, membership in areas outside the United States, number of women, propor tion of white-collar members, and proportion of members in major industry groups. Uses and Limitations The Bureau’s membership figures are pub lished on a regular basis, and serve as the principal indicator of gains and losses for par ticular unions and for the labor movement as a whole, particularly as a measure of inroads or declines in industry and occupational groups. Comparisons of labor relations policies and their economic effects can be made between organized and unorganized sectors. They are used by agencies of the Federal Government, State and local governments, by management personnel, union officials, students of the labor movement and economic affairs, and the gen eral public. The Directory also is distributed widely abroad, notably to international organi zations and labor unions. Difficulties in measuring union membership arise from (1) the variety of concepts and practices among unions as to the definition and reporting of membership, (2) the availability at union headquarters of the various data re quested, and (3) the willingness of the unions to make those data available to outsiders. In an attempt to achieve uniform reporting practices, the Bureau asks unions to report on the annual average number of dues-paying members. Although a worker when joining a union assumes an obligation to pay dues, uni form reporting practices do not result from applying this criterion alone. Unions define eligibility for membership in a variety of ways and payment of dues is only one of several criteria. Some unions set less than full dues requirements or waive the payment for work ers who are unemployed or on strike. Such exonerated Workers, however, usually remain members in good standing, with the same rights as full dues-paying members. Similar qualifications may apply to members who are apprentices, retired, or in the Armed Forces.3 In an attempt to determine union practices in reporting membership, the Bureau repeat edly has requested unions to indicate whether they include or exclude from membership re ports five specified groups: the unemployed; those involved on work stoppages; those in the Armed Forces; apprentices; and the retired. Moreover, unions were asked to furnish an :! In its 1963 Directory, the Bureau took a closer look at the relationship between dues submitted to the international union, the so-called per capita tax, and reported membership totals for particular unions. Per capita tax receipts were divided by the per capita tax rate. Briefly, the findings indi cated that while a number of unions use a “ per capita” receipt figure in their reports to the Bureau, such an ap proach was inappropriate in the case of other unions for reasons which the unions explained in detail. Some unions include a large number of seasonal employees whose dues payments are limited to several months during the year. Other unions questioned the use of a computed membership figure where for large groups of workers (sick, unemployed, those promoted out of the bargaining unit, etc.) only a partial per capita tax is paid, at times less than one-tenth of the amount required of other members. A few unions indi cated categories of members completely dues exempt (e.g., life members, 50-year members, etc.). Financial obligations also frequently are waived for workers recently organized and for local unions in economic difficulties. In addition, several unions set dues on a sliding scale proportionate to the income of members, a method which rules out computa tions of this sort. Thus, on examination it became clear that computing membership by dividing the tax rate into tax receipts could not be used as a uniform yardstick applicable for all unions. 193 UNION MEMBERSHIP estimated or actual figure on the number of members in “ excluded” categories. Ideally, if all unions could furnish such data, it would be possible to compute the total number of work ers who, at least in some way, are still attached to unions. The responses, however, have fallen far short of this goal. Only a small number of unions reported, in whole or in part, the prac tices they followed. An even smaller number of unions were able to furnish figures on the num ber of workers involved.4 Thus, no uniform answer exists to the de ceptively simple question: Who is a union member? The answer varies from union to union, as determined by its own policies and practices. Although financial statements may be of * At various times, suggestions have been made on the use of alternative sources for membership information, notably the use of “voting strength” of international unions at A F L CIO conventions, based upon average paid membership to the Federation. This, however, is not a reliable measure. Some unions, for reasons of prestige, tend to overpay, while others, as a matter of convenience, pay an arbitrary amount. Moreover, this method could not be used for unions outside the AFL-CIO. some help in arriving at membership approxi mations, they cannot be used to obtain data on various components of union membership, such as the proportion of white-collar and women members, those under contract in particular industries, etc. For such data, the Bureau must rely entirely on the cooperation of national unions. These, however, are not always able to furnish the information for the simple reason that it is not compiled at union headquarters. Many national unions are decentralized or ganizations and as a rule, do not seek reports on these items from their local unions. Also, some unions in possession of such information may be reluctant to make it public. Data presently compiled are submitted by officials of national and international unions. Perhaps more accurate data, and certainly more detailed data, could be obtained by con tacting local unions, as is done in Canada, but the task of soliciting responses from more than 70,000 organizations is beyond the present re sources of the Bureau. Technical References Number 1. 2. 3 4. 5. 6. 7. Troy, Leo. Distribution of Union Membership Among the States, 1939 and 1953. New York, National Bureau of Economic Research, 1957. A study measuring union growth by State and region, analyzing geographic and indus trial shifts in membership. Includes a discussion of sources and methods of measurement. U.S. Department of Labor, Bureau of Labor Statistics. “ Membership of Labor Unions, 18971950.” Handbook of Labor Statistics, 1950 ed., (Bulletin 1016, 1951), pp. 137-139. . ______Directories of Labor Unions in the United States; (Bulletins 937, 1948; 980, 1950; and 1127, 1953). --------- - Directories of National and International Labor Unions in the United States (Bulletins 1185, *1955; 1222, 1957; 1267, 1960; 1320, 1962; 1395, 1964; 1493, 1966; 1596, 1968; and 1665, 1970). In various articles in Monthly Labor Review. (Excerpted) In addition to membership data, the Directories also include a brief description of the methods used, and a copy of the questionnaire. Paschell, William, “ Limitations of Union Membership Data,” Monthly Labor Review, Novem ber 1955, pp. 1265-1269. Wolman, Leo. Ebb and Flow in Trade Unionism, New York, National Bureau of Economic Research (1936). Of invaluable aid to students in the field because of its extensive discussion of measure ment problems. Appendix tables present figures for the period 1900 to 1934 including data on the extent of organization by industry. U.S. Department of Commerce, Bureau of the Census, Current Population Reports, Series P 20, No. 216, “ Labor Union Membership in 1966” (1971). The first comprehensive study giving economic and demographic data for union and non union private wage and salary workers. — H arry P. Co h a n y Chapter 22. Annual Earnings and Employment Patterns of Private Nonagricultural Workers Background The purchasing power of workers depends largely on their annual earnings which are determined by the interaction of variables such as straight-time rates of pay, number of hours worked, and the hours worked at premium rates which depend on other variables like occu pation, union status, industry, and area. More over, some workers move into and out of the labor force during the year. In addition, a substantial portion work for more than one employer in the same industry and for one or more employers in different industries. Con sequently, occupational wages or hourly and weekly earnings data cannot be extrapolated to annual earnings estimates with any degree of precision. The Bureau’s program of annual earnings and employment patterns studies, initiated in the late 1960’s, is designed to fill the gap in our knowledge of annual wage and salary earnings from private nonagricultural employ ment. The first study 1 for 1964, was limited to wage and salary earnings covered by social security; the second,1 for 1965, included data 2 on wage and salary earnings covered under either the Social Security Act or the Railroad Retirement Act. Subsequent studies, like that for 1965, will include information on earnings covered under either social security or railroad retirement. Bulletins providing 2 years of data are expected to be issued biennially. Special analytical studies will be published intermittantly in the Monthly Labor Review. ticipate in social security and function like private firms, such as hospitals and schools. The studies provide distributions of median and mean earnings for the private nonagri cultural sector as a whole, for each industry di vision, for each major industry group at the two-digit SIC level, and for selected industry groups at the three-digit SIC level of industry classification. The data, though available only for white-collar and blue-collar workers com bined, are unique, because unlike annual earn ings data from other sources, they permit an analysis of the distribution of wage and salary earnings and employment patterns by industry and quarters of employment. The studies focus separately on earnings in the industry in which workers had greater earnings than in any other, and earnings in all industries. Some of these data and some of the employment patterns data, separately and with earnings data, also are presented for selected demographic characteristics. Earnings and employment patterns of workers who had covered wage and salary each quarter of the year are emphasized par ticularly. Information about earnings and em ployment patterns of four quarter workers is the closest to data for workers fully attached to the private sector work force that can be obtained from the source materials. However, some workers who have earnings in each quar ter are attached to the work force to only a limited extent. Source o f Data Description o f the Series The series covers earnings and employment patterns in the private, nonagricultural sector which, broadly defined, includes individuals who work for wages and salaries in employ ment covered by the Social Security Act and the Railroad Retirement Act.3 Excluded are earnings in agriculture, self-employment, and in government units other than those that par The data are developed by the Bureau of Labor Statistics from statistical information 1 Annual Earnings and Employment Patterns, Private Non agricultural Employment, 1964 (BLS Report 330 1969). 2 Annual Earnings and Employment Patterns of Private Nonagricultural Employees, 1965 (BLS Bulletin 1675, 1970). 3 For a discussion of the interrelationship of, and joint coverage under, the railroad retirement and social security systems, see U.S. Department of Health, Education, and W el fare, Social Security Administration, Social Security Hand book, 4th ed. (1969), and U.S. Railroad Retirement Board, Handbook on Railroad Retirement and Unemployment Insur ance Systems (1969). 195 196 BLS HANDBOOK OF METHODS obtained from the Social Security Administra tion and the Railroad Retirement Board. To preserve the confidentiality of the records, the data are provided to the Bureau of Labor Statistics without identification of individuals or employers. However, to combine data from various employments and to facilitate statistiscal processing, each individual and employer is assigned a permanent control number, differ ent from his social security or employer identi fication number. Each individual in the sample provides demographic information (race, sex, and year of birth) when he applies for a social security number. Each employer under social security from whom the individual receive any wages or salaries during a calendar quarter reports the amount of the wage payment in the quar ter; employers covered under railroad retire ment .report monthly. However, employers, cease to report wage and salary earnings after the worker has reached his taxable earnings limit in that employment situation. Employers report wages paid to the maximum annual limit under social security and to the maximum monthly limit under the railroad retirement system. Employer reports also indicate the industry and, except for employment covered by the Railroad Retirement Act (R R A ), the area in which the wages or salaries were earned. Employers subject to the RRA, also provide information about the occupational category the worker was employed in. Sample Design, and Sampling and Nonsampling Variability The sample, which includes 1 percent of all social security numbers, was selected on the basis of a multistage systematic cluster sam pling procedure. Social security numbers are used as the individual’s identification number in both the social security and railroad retire ment systems. An individual selected for the sample remains in it permanently.4 Since estimates in this study are based on a sample, they may differ from census figures. Moreover, the sample data are not adjusted to benchmark levels established by complete counts. In addition, the data are subject to nonsampling variability due to errors in re porting and classification and other possible error sources, that would be present in a com plete enumeration as well as in a sample. As a result, ratios established from the sample data are considered to be reasonable estimates of those existing within the population as a whole. Nevertheless, particular care should be exercised in interpreting medians and percents based on relatively small numbers of cases as well as small differences between figures.5 Ab solutes are subjet to large error which may vary from year to year. Estimating Procedures To estimate total wages of individuals, the Social Security Administration determines the quarter in which the taxable limit is reached ( “ limit quarter” ). Wages in the prior quarter equal to or greater than the “ limit quarter” wages are substituted for those in the “ limit quarter” and in all subsequent quarters. Limit quarter earnings, however, are used to estimate earnings in the limit and subsequent quarters if limit quarter earnings were higher than earn ings in previous quarters. The summation of the quarterly wages after these substitutions then becomes the estimated annual total. When the taxable limit is reached in the first quarter, the Social Security Administration imputes an estimated total. Employers covered by the Railroad Retire ment Act are required to provide information about the monthly earnings of each employee up to the maximum creditable limit subject to Railroad Retirement Act taxes. Hence, even earnings reported at the maximum level for each month aggregated to annual totals may be* 3 4 For a detailed discussion of the sampling procedure, re porting criteria, and coverage under the social security and railroad retirement systems, see U.S. Department of Health, Education, and Welfare, Social Security Administration, Workers Under Social Security, 1960 (1968) and Social Se curity Handbook, op. cit.; also see Handbook on Railroad Retirement and Unemployment Insurance Systems, op. cit. 3 For an indication of the order of magnitude of the sam pling errors and a fuller discussion of the sampling and non sampling variability to which the series is subject, see Workers Under Social Security, 1960, op. cit. ANNUAL EARNINGS & EMPLOYMENT PATTERNS OF PRIVATE NONAGRICULTURAL WORKERS substantially below the worker’s total earnings. The Railroad Retirement Board, however, col lects information from employer records about the total annual earnings of a sample of workers covered by the act. Factors for raising creditable compensation under the Railroad Re tirement Act to total railroad earnings are derived by comparing the total earnings data for individuals collected in the special study with the aggregated monthly earnings data for the same individuals. The incremental factors for workers in the same broad occupational categories then are averaged. The resulting factors, developed by the Railroad Retirement Board, are applied by the Bureau of Labor Statistics to the credited monthly earnings of each individual by taking into account his occupational category.6 Analysis and Presentation Distributions of workers by annual earnings, and mean and median earnings are presented for all major earners and for those with four quarters of earnings. In addition, data are provided showing the number of workers who had wage and salary earnings in the various industries and the proportion of these who had greater (major) earnings in that industry than in any other industry. Data on major earners by industry showing the proportion with earn ings in each quarter, their demographic char acteristics, what proportion had earnings within a single industry and the number of employers whom they worked for also are presented. Some earnings and employment pat terns data are presented for broad geographic areas. Terms used in the series and the methods used to classify workers by industry and region of major earnings and the industrial classifi cation scheme used in classifying nonpolicy governmental units in scope of the study pro gram are described below. Annual earnings are defined as gross wages, salaries, and other payments (such as bonuses) received by employees, before deductions, in employment covered under the Social Security Act or the Railroad Retirement Act. Such pay ments may be cash, cash equivalents, or other 197 media such as goods, clothing, board, or lodging. Self-employment earnings, payment for work in employment excluded from the coverage of the acts, and payment for work in agriculture, in covered governmental units engaged in public administration and for military service have been excluded from this study. Most pay ments by employers to or on behalf of em ployees, or for employees and their dependents for retirement, death, sickness or accidental disability, or medical and hospitalization ex pense under provisions of a plan or system meeting certain general criteria, and employer payments to a trust fund, such as a pension trust, exempt from tax under the Internal Revenue Code, are not counted as wages in this series.7 Workers with some earnings in the industry. Each individual who earned $1 or more in an industry during the year is counted in each industry in which he had any earnings. A worker who had some earnings in each of five three-digit industries, as defined in the Stand ard Industrial Classification Manual8 for ex ample, is counted for each of these industries as well as in each two-digit industry and in each division of which the three-digit industres are a part. Because a worker is counted in each three-digit industry, each two-digit industry and in each division in which he had $1 in cov ered wage and salary earnings or more, the aggregate count at each level is greater than the total number of workers at each broader industry level (two-digit, division, private nonagricultural economy). 8 Although the Railroad Retirement Board collects annual earnings data for a 1-percent sample of workers, the sampling criteria differ from those used by the Social Security A d ministration. To permit the combination and integration of data from the two systems, the Railroad Retirement Board provides the Bureau of Labor Statistics with data for a sample of workers selected according to the sample design established by the Social Security Administration. 7 Under certain circumstances tips are counted as wages for social security purposes and thus are included in the data file used in this study series. Payments to workers from tax exempt trust funds are not considered wages (except for wages paid by the fund to its employees) and thus are excluded. Employer payments to trust funds which are not tax exempt also are excluded. Depending on their nature and purpose, payments to workers from these funds may be counted for social security purposes and thus are included in the data file.— Such would be the case if the payment was a bonus or vacation payment. For a more detailed discussion of covered wages, see the Social Security Handbook, op. cit. 8 Issued by the U.S. Office of Management and Budget. 198 BLS HANDBOOK OF METHODS Major earners and industry of major earn ings. A major earner in this study, is a worker who earns more of his annual wages and salaries in a specific industry than in any other industry. This test to determine the worker’s industry of major earnings is based on a plurality earnings concept and is applied sepa rately at each industry level. Data for major earners are included in only one industry at each level in industry classifi cation (i.e., at the three-digit, the two-digit, and at the divisional levels). Thus, the number of workers who received the major proportion of their earnings in each industry, at any level of industry classification, is unduplicated. Hence, data for any level of classification at which all component industries are shown add to the total for the private nonagricultural sector. However, data at the three-digit level (if all three-digit industries were shown) might not add to the detail at the two-digit level, and data at the two-digit level might not add to the divisional level. Except for totals at the private nonagricultural sector, this non additive relationship exists because some workers shift employment several times during the year and bring about patterns like that shown in the following tabulation of a hypo thetical worker’s employment and quarterly earnings experience. Total (any quarter) Private non agricultural econom y_ _ Division A 2-digit group, A -l 3-digit group, A -ll _ .. $530 January to to March $80 July Aprilto Septem Juneber October to D ecem ber $170 $130 $150 150 60 90 150 60 90 50 90 150 Division B 2-digit group, B -l 3-digit group, B -l l 3-digit group, B-12 __ _ 180 60 50 60 50 70 130 10 50 70 50 50 Division C 2-digit group, C -l _ ___ 3-digit group, C -l l _ - __ 2-digit group, C-2 ______ 3-digit group, C-21 _ ___ 200 20 20 60 100 20 20 60 100 20 20 60 Quarters o f W ork 1 1 0 In this series, because workers are classified both by quarters worked 1 in their major in 0 dustry and in all wage and salary employment, quarters of work of the hypothetical worker whose employment and earnings were illus trated previously are as follows: Major industry Private Division 2 - digit 3 - digit nonagricultural __ C ________________ group, B -l ____ group, A -l l 4 4 3 2 All wage and salary em ploym ent 4 4 4 4 Employer. An employer in this study is an individual, partnership, or corporation recog nized under the law as a separate entity meet ing certain criteria.1 However, a firm which 1 separately incorporates at each of its locations may be considered a separate employer at each location. Thus, a worker transferred from one to another location that is incorporated sepa rately may have more than one employer in the same year even though he continued to work for the same firm. Industrial classification. Employment and earnings data based on the Social Security Ad 70 180 In this tabulation, the worker had greater earnings at the three-digit level in industry A - ll than in any other three-digit industry; at the two-digit industry he had greater earnings in industry B-l than in either A -l, C-l, or C-2; and at the divisional level he earned more in division C than in either A or B. Therefore, applying the plurality earnings test, this worker is assigned to industry A - ll at the three-digit level, B-l at the two-digit level, and to C at the division level.9 100 100 100 100 100 9 An unpublished tabulation, prepared by the Social Security Administration, indicated the industry of major earnings at the three-digit level for about 12 percent of all covered workers was part of a two-digit industry different from the workers two-digit industry of major earnings. 1 “ Quarters of work” and “ quarters of coverage” are not 0 synonymous. A quarter of coverage based on covered non agricultural employment is a calendar quarter in which the individual was paid $50 or more; a quarter of coverage also may be credited on other bases such as agricultural employ ment or self-employment. 1 For details, see the Social Security Handbook, op. cit. In 1 addition, because some workers work for more than one employer during the same week, data showing number of employers should be used with caution. ANNUAL EARNINGS & EMPLOYMENT PATTERNS OF PRIVATE NONAGRICULTURAL WORKERS ministration’s (SSA) data file are classified according’ to the Administration’s industrial classification system. This system differs slightly from that published in the Standard Industrial Classification Manual (SIC), and used in most statistical series in the assignment of industry codes to governmental units. Most statistical series classify governmental units into SIC Division I-Government. The SSA, however, classifies only policymaking govern mental units in Division I. All separable non policymaking units are assigned to nongovern mental SSA industry classification code appro priate to their activity. Thus, for example, all employment with policymaking boards of edu cation (classified by the SSA and SIC as gov ernment) has been excluded from this study. Schools, colleges, and other operating units cov ered under voluntary election provisions of the act, however, were treated as service industry employment, because the units were classified by SSA into SIC 82, educational services.1 2 Employment and earnings data based on the Railroad Retirement Board’s file are classified into the following industries as defined in the SIC Manual: Railroads, SIC 401; sleeping car companies, SIC 402; express companies, SIC 404; rental of rail cars companies, SIC 474; and other companies performing services rail road transportation and certain railway labor organizations, SIC 861 and 863. In each case the assignment is based on the industrial classi fication of the worker’s last employer who was covered under the Railroad Retirement Act. Data for workers who had earnings in more than one industry, all of which were covered 1 For detailed information, see U.S. Department of Health, 2 Education, and Welfare, Social Security Administration, Com parison of Social Security Administration and Standard In dustrial Classification Systems, 1963 (undated) and the guide prepared by the Administration in 1968, entitled “Industrial Codes in the Social Security Administration Continuous Work History Sample (CW HS), Data for 1957 through 1966.” 1 For all industries, except water transportation, employ 3 ment outside the 50 states or the District of Columbia is only a small fraction, if any, of the total number of the industry’s major earners. 1 Employment under the Railroad Retirement Act makes up 4 nearly all the major earners in the railroad industry (SIC 40) and a substantial proportion of those at the all transportation level, but only a very small fraction of the total employment in other industries or at the total private nonagricultural level. Thus, the convention adopted has a serious effect on the regional employment distributions and regional earnings picture at the all transportation level but little effect on other industries or at the total private nonagricultural level. 199 under the social security system or one of which was covered under the railroad retirement system, are classified and presented separately and in combination. Single and multi-industry workers-at each level of industry classification the employment experience of each sample member was ex amined to see if all of his earnings were in one or more than one industry. Those with earnings in more than one industry were classi fied as multi-industry workers. This conceptual approach may be seen in the case of a worker who was employed, as illustrated below, by an employer in each of two three-digit industries within the same two-digit industry. Industry level Number of employers Division A _____________ _ _____ _ ______ 2 2- digit group, A -l ...__________ ______ 2 3- digit group, A - l l 1 _________ ______ 3-digit group, A-12 ______________ ___ 1 Region of major earnings. The region in which workers had their major earnings is deter mined by a plurality earnings test similar to that described in detail in “Major earners and industry of major earnings.” The region in which he had greater earnings than any other is his region of major earnings. In a few cases, earnings in the industry and region of major earnings may not coincide. All data for major earnings are classified first by industry and then by region. A worker who earned 40 percent of his annual wages in in dustry A in the Northeast, 30 percent in in dustry B in the South, and the remaining 30 percent in industry C also in the South would be classified as a major earner in industry A and as having had his major earnings in the South. In this series, workers employed under the Social Security Act or the Railroad Retirement Act are divided into five regions: Four cover the 50 States and the District of Columbia; the fifth includes all employment in U.S. territories, on foreign soil, or aboard ocean-going vessels.1 3 Data in this series do not indicate where wages and salaries covered by the Railroad Re tirement Act were earned.1 Therefore, a con 4 vention was adopted ascribing all such em ployment to the North Central region where 200 BLS HANDBOOK OF METHODS many railroads and railroad-related organiza tions have their headquarters. Race. All workers in this series have been di vided into two groups, “ white” and “ Negro.” The white category includes all workers except Negroes. This convention, which is different from that used in most statistical series, was adopted to minimize the effect on those groups for whom the sample was not sufficiently large to present data separately and to maximize the analytical usefulness of the data. Uses and Limitations The data provide an insight into the answer to the question: “ How well do private nonagricultural workers do in their industry of major earnings and to what extent do they supplement these earnings by employment in other private sector industries.” Data are used in collective bargaining; in formulating public policy and in making inter-industry and inter national comparisons; in analyzing the distri bution and diversity of earnings, and varia tions in employment patterns among industries, regions, and between workers of different races and sexes. The data, however, have several substantial limitations. Some workers whose annual earn ings are included in the series also have earnings in self-employment, agriculture, or employment in governmental units excluded from the series. As a result, these workers appear to have low annual earnings. Probably most of these workers were attached to the employed work force only a very limited extent. The unavailability of data on hours or weeks worked or paid for, or occupational group (other than in railroading) seriously limits the analytic potential of the series. The 3-5 years lag between the reference period and the date of publication introduces another limiting factor. These delays result primarily from re porting requirements established by the law, the administrative requirements of the collect ing agencies, and the processing required to reduce the mass of micro data into statistical summaries. Nevertheless, since employment patterns usually change very slowly and rela tive earnings distributions generally are quite comparable from year to year even though the level of earnings moves upward, the relation ships shown are indicators of the current situation. Nothwithstanding the limitations, data from this series, unlike those from other sources,1 5 permit an analysis of the distribution of wage and salary earnings and employment patterns of workers by demographic characteristics, in dustry, and quarters of employment. Thus, they are uniquely useful to all concerned with the annual wage and salary income of individuals and the employment patterns of those in the private nonagricultural work force. 1 5 Dissimilarities in concept or method between the BLS annual earnings and employment pattern series and other series may result in important differences in sampling and nonsampling variability between series. Therefore, caution must be exercised in using data from the BLS annual earn ings series in conjunction with data from other statistical series. The major sources of other annual earnings data to gether with a brief explanation of the most important differ ence between them and the data in this series are noted below. The Social Security Administration (SSA) publishes some annual earnings data by industry. Their most recent report Workers Under Social Security, 1960 (1968), provides sta tistical information about employment, earnings, and insur ance status of workers under old-age, survivors, disability, and health insurance. The SSA also publishes selected sum mary data in the Social Security Bulletin. The industry attachment concept used by the SSA, however, is different from that used in this study. Further, the Social Security data do not include earnings in employment covered by the Railroad Retirement Act. The Railroad Retirement Board (RRB) annually publishes a “ research and statistics note” which provides information about the total railroad earnings of railroad employees. The RRB data, however, exclude earnings in employment not covered by the Railroad Retirement Act. Some annual earnings data at the all-industry level by occupational group are published by the Bureau of the Census, U.S. Department of Commerce, in “ Consumer In come,” Series P-60 of the Current Population Reports. This publication also provides a distribution of wage and salary earnings, at the all-industry level, by sex and race. In addi tion average earnings by sex are presented for selected in dustry divisions and for a few major industry (two-digit SIC) groups. The study, based on a household survey, does not provide distributions of wage and salary earnings by industry group and has different concepts of industry attach ment from those used in this study. The Office of Business Economics of the U.S. Department of Commerce also publishes estimates of the average annual earnings of “full-time employees” in its Survey of Current Business; these estimates do not reflect the effect on average earnings of workers who work less than a full year. — A rnold Strasser Chapter 23. Measuring Collective Bargaining Settlements Background The Bureau’s program of measuring the effects of collective bargaining settlements on hourly labor compensation is a reflection of two developments: One, the growing impor tance of fringe benefits as a proportion of employee compensation, and two, increased concern about the effects of collectively bar gained wage and benefit changes on the price level. Whereas in earlier years the economic terms of negotiated settlements could be equated largely with agreed-upon changes in wage rates, today, possible changes in a host of pay supplements must be considered— such as vari ous forms of premium pay, paid leave, cash bonuses, and employer contributions to funds providing pension or health and welfare bene fits. Although straight-time pay for working hours is still the major element of compensa tion, supplements are now a significant portion, accounting for about a fifth of total employer outlays for worker compensation.1 Moreover, growing concern during the 1960’s over the extent to which increased labor costs 1 See Alvin Bauman, “ Measuring Employee Compensation in U.S. Industry,” Monthly Labor Review, October 1970, p. 23. It is difficult both to measure the growth of supplements over the years and to quantify their current importance. The national income accounts provide one pertinent source of data. They show that supplements to wages and salaries rose as a percentage of total employee compensation from 1 percent in 1929 to 10 percent in 1970. The National Income and Product Accounts of the United States, 1929-1965: Sta tistical Tables, U.S. Department of Commerce, Office of Busi ness Economics (1966), p. 14; Survey of Current Business, January 1971, p. 10. These figures, however, do not reveal the relative importance of all supplements in either year, since many— such as premium payments, leave payments, and cash bonuses— appear as parts of wages and salaries. The figures therefore are not comparable to those found in sur veys of employer expenditures for supplementary compen sation. 2 Publication of first-year changes is a recent innovation. Before 1970, the Bureau published two measures of change over the life of the contract, the so-called equal timing and the time-weighted measures. The former assumed equal spacing of changes during the term of the contract; the latter took account of the actual effective dates of wage and benefit changes. The time-weighted measure has been discontinued, because it appeared to be of significance primarily for the analysis of individual settlements rather than for overall series of the type produced by the Bureau. Moreover, drop ping of the time-weighted measure and introduction of a series on first-year changes provides parallel statistics both on wage-rate changes alone and on wages and benefits combined. may contribute to inflation has heightened in terest in the size of collective bargaining settlements. Responding to these influences, the BLS be gan estimating the cost of wage and benefit (i.e., “ package” ) changes in a limited number of key settlements in 1964. The work was ex panded the following year and, since 1966, the Bureau has attempted to determine the price of all settlements affecting 5,000 workers or more in the private nonfarm sector. In addi tion, a separate series has been developed for the construction industry, covering settlements for 1,000 workers or more. Description o f Series At present, the Bureau publishes two sets of data on wage-benefit decisions. One shows the annual rates of increases in settlements reached in a given time period and scheduled to go into effect at any time during the term of the agreements. The other is limited to the changes set for the first 12 months of the agreements.1 Published data summarize settle 2 ments reached during individual quarters of a year, during full years, and during the first 6 and 9 months of each year. Frequency distributions are shown for work ers grouped by the size of their settlements. In these distributions, all workers affected by a given action are entered at the average for the bargaining unit. The sums of the individual settlements are averaged— both means and medians are presented— each settlement being weighted by the number of workers affected. However, the pricing of individual settlements is not disclosed. Averages for full years are available sep arately for manufacturing and nonmanufactur ing industries. Otherwise, no industry detail is published, except for the separate construction industry series. As indicated in the preceding paragraphs, these series relate to the pricing of decisions, i.e., they measure the effect of changes agreed on in a given period although, considering the general practices of negotiating multiyear 201 202 BLS HANDBOOK OF METHODS collective bargaining agreements containing provisions for annual (and sometimes more frequent) improvements, the changes may be introduced only at a subsequent date. Work has recently been completed on the development of data on the wage and benefit changes actually placed in effect in specified periods, whether as a result of current bargains, changes agreed upon earlier but with deferred effective dates, or the operation of cost-of-living wage es calator clauses. Data Sources and Collection Methods The terms of the settlements to be priced are obtained primarily from secondary sources, such as general circulation newspapers and periodicals and union, management, and trade publications. Collective bargaining agreements and documents on pension and health and wel fare plans also are consulted. When these sources are inadequate, direct requests for in formation are made to the companies and unions involved. Large quantities of statistical data, as well as the settlement terms, are required. These are needed both to determine existing em ployer outlays and to assay the effect on these expenditures of agreed wage and benefit changes. Efforts are made to use existing data. However, when these prove inadequate, the parties are requested to furnish data. Such re quests, it must be emphasized, are made to receive specific information from which the Bureau can price settlements; the requests are not made to receive the parties’ own evalua tions of the terms of their settlements. One of the major sources of information on current hourly earnings is the establishment information which BLS obtains through its monthly employment, payroll, and hours sur vey. Information on current outlays for pay supplements may be available from BLS sur veys of expenditures for such benefits. Perti nent information for estimating expenditures for some items may be found in industry wage surveys, e.g., extent of late-shift work and occupational employment distributions. Annual financial reports filed with the Department of Labor under the Welfare and Pension Plan Disclosure Act provide useful material. Not all the sources tapped are governmental; for in stance, information on workmen’s compensa tion insurance rates is reported by the National Council on Compensation Insurance. Sampling and Estimating Procedures As was noted earlier, the Bureau attempts to cover all settlements for 5,000 workers or more (1,000 or more in construction) in the private nonfarm sector. Substantially all such settle ments come to the attention of the Bureau and are included in its series, it is believed. Discussion of procedures for pricing individ ual settlements 3 may be centered around three questions: (1) What items in a collective bar gaining settlement are to be priced? (2) How are the costs of these items to be determined? (3) How are these costs to be expressed? Coverage. Little evidence is needed to demon strate that many terms of a union-management agreement, not merely the wage and benefit provisions, may affect directly or indirectly an employer’s costs. For example, one of the socalled “ noneconomic” terms of an agreement— seniority— may have a limited influence on costs through its effect on employee efficiency.4 How ever, such an item is essentially not meas urable. Consequently, the BLS program is confined to the wage-benefit component of collective bargaining agreements, i.e., to the effect of settlements on employer outlays for employee compensation. This component, clearly, is of major significance in its own right. Included are items such as changes in wage rates; modi 3 Production of statistical series merely entails grouping and averaging the pricing of individual settlements. Published distributions show the number of workers under known settlements for which data were insufficient to permit pricing. 4 Almost inevitably a tendency has developed to regard all contract changes as serving to increase employer payments. Although this undoubtedly is the common result, some changes, even when proposed by unions, may lower costs. Union-management cooperation schemes have at times pro vided examples. MEASURING COLLECTIVE BARGAINING SETTLEMENTS fications in premium pay, bonuses, paid leave, and severance pay; and adjustments in em ployer payments for pension, for health and welfare, and for supplemental unemployment benefits, excluding the costs of administering these benefits. Also included are changes in formal contract provisions specifying paid time for clothes change, washup, and lunch periods. Excluded are informal modifications of un written rest-period practices; items related to, but not normally considered part of, compensa tion— such as per diem payments, moving ex pense reimbursements, and payments for safety clothing; and provision of facilities or services such as parking lots and health units, the costs of which often are charged to capital rather than labor accounts. Determination of costs. Since a value is placed on settlements at the time they are reached, the costs attributed to them obviously are esti mates of outlays to be made in the future; they cannot be taken from employers’ accounting records. The estimates are made on the assump tion that conditions existing at the time the contract is negotiated will not change. For ex ample, estimators assume that methods of financing pensions will not change, and that expenditures for insurance will not change ex cept as a result of altered benefit provisions or modified participation because of changes in company contributions. They also assume that the composition of the labor force will not change. In this regard, except for any guaranteed increases, which are treated as deferred adjust ments, possible wage-rate changes as a result of cost-of-living escalator clauses are excluded because of difficulties in predicting movements of the Consumer Price Index for 3 years—the time span of many collective bargaining agree ments. Thus, the Bureau prices the wage and 5 The series on wage and benefit changes actually placed into effect includes escalator adjustments, since the issue of prediction of CPI movements does not arise. 8 The terms “roll-up” and “bulge,” among others, also are used to express the same idea. 7 It varies from this ratio to the extent that there are limits on earnings that are subject to social security taxes. 203 benefit changes that would go into effect if the price level were to remain stable.5 Nevertheless, package estimates do attempt to measure the costs associated with actual characteristics of the work force affected by the settlements, not the costs for some hypo thetical employee group. Attempts to base esti mates on the actual age, length of service, sex, and skill characteristics of the workers involved recognize that the choice in incorporating alternative benefit changes into contracts is affected by their costs, which, in turn, are affected by the character of the work force. For example, an extra week of vacation after 15 years of service will cost very little when only 10 percent of the workers have that much service, but will cost about 1 percent of straight-time hourly earnings when half of the workers have been employed for 15 years or more. As a rule, indirect effects of settlements are ignored: factors such as possible extension of settlement terms to nonunion workers in the same firm or to members of other bargaining units. Similarly, the cost of providing length ened vacations is measured by the wages and salaries paid for the additional time o ff; costs of hiring vacation replacements, if necessary, are not considered. Moreover, effects on unit labor costs, which involve consideration of em ployee efficiency as well as employer payments, are disregarded. However, “ creep” 6 is accounted for. Creep reflects the fact that an increase in wage rates will have a secondary effect on employer costs through its effect on outlays for benefits. A 20-cent-an-hour wage increase will affect not only straight-time wage rates but also supple mentary payments governed by wage rates— such as overtime premiums, leave payments, pension benefits related to salary level, and social security payroll taxes (if earnings are not at or above the statutory maximum tax base). Creep is taken into account by raising each wage increase provided by the new contract by a creep, or loading, factor. This factor is es sentially the ratio of current hourly expendi tures on benefits that vary with wage levels to current straight-time hourly earnings.7 204 BLS HANDBOOK OF METHODS Many items in a collective bargaining agree ment are priced without difficulty. This is particularly true when settlement terms are expressed as cents-per-hour adjustments, e.g., a 20-cent-an-hour general wage increase 8 or a 5-cent increase in employer contributions to a health and welfare fund. These stipulated cents-per-hour figures are utilized as the costs of the settlement provisions.9 Percentage wage adjustments are converted to cents-per-hour figures on the basis of current average straighttime hourly earnings. Although less direct, the cost of an additional holiday is estimated ade quately by prorating 8 hours’ average pay (if the normal workday is 8 hours) over the num ber of annual working hours per employee. The cost of an additional week of vacation for 25-year employees is estimated similarly, but one must know the number of employees with the required seniority. Other settlement terms are more difficult to price. For example, the cost of an unfunded severance pay plan depends on the frequency of layoffs as well as on plan provisions. Estimates of such frequencies are at best hazardous. Pension improvement costs are particularly difficult to estimate because of the considerable discretion employers often have in funding their obligations.1 The general ap 0 proach followed by the Bureau is to assume that a given pension benefit increase will raise existing expenditures for current service pro portionately. Since employer contributions for pensions frequently vary widely from year to year, outlays in several past years are examined to develop a measure of current pay ments. Under the BLS framework, estimates con cerning most provisions are of actual cash outlays to be made by employers. However, in the case of improved paid leave provisions, a change may entail time off for workers, but not additional cash payments by the employer. 8 Where appropriate, the effect of a general wage increase on incentive earnings is included in the measurement of the wage increase. 9 Since estimates are on a cents-per-hour-worked, rather than per-hour-paid-for, basis, agreed-upon increases are ad justed if they are on an hours-paid-for basis. 1 Employers’ contractual obligations commonly are to pro 0 vide given levels of pension benefits, rather than to expend specified amounts of money. However, payment per hour worked will rise and this change is taken as the cost effect of the settlement provision. In case of a reduction in the basic workweek, the increase in hourly rates needed to maintain weekly pay is the major item BLS prices. To some extent, a reduced basic workweek may be accompanied by additional overtime work. However, unless this overtime is provided spe cifically in the agreement, it is ignored in the cost estimate. Increases in hourly pay rates are not the only cost effects considered in this instance. Even if there is no change in total employer outlays for particular pay supplements but the contract provides for reduced hours, the out lays for them per hour worked will rise and affect the cost of settlement. Expressing the costs. The total cost of a given settlement is obtained by adding up the centsper-hour-worked costs of each of the individual wage or benefit changes. This sum is then expressed as a percent of pay, as this ratio facilitates inter-company comparisons by elimi nating influences of payroll size and wage level. Furthermore, since economic studies generally emphasize relative rates of change in statistical series, percent-of-pay costs can be integrated into broad economic analyses. Expression of package costs as a percent of pay requires estimation of an appropriate base (the denominator of the ratio) as well as the cost of the settlement terms (the numerator). The base used by the Bureau consists of cur rent outlays per hour worked for all items of employee compensation, as defined, plus em ployer expenditures for legally-required social insurance. The latter is part of employee com pensation, although not subject to change through collective bargaining. Since collective bargaining agreements gen erally are for 2-year periods or longer, BLS expresses the total percent increase over the contract term at an annual rate to permit com parison among agreements for differing time spans as well as to facilitate the use of the data in conjunction with other statistical series. These annual rates of increase take into ac count the compounding of successive changes. MEASURING COLLECTIVE BARGAINING SETTLEMENTS In addition, the Bureau computes first-year changes, i.e., the total change scheduled for the first 12 months of the agreement, expressed as a percent of current hourly compensation. As a general rule, the first-year increase is larger than the average annual increase over the full term of the agreement; contracts com monly are “ front-loaded.” Contracts are considered to run from their effective dates to their termination dates. How ever, where wage reopening clauses are found, the reopening date is taken as the termination date and any agreement under the reopening clause is treated as a new settlement. Pricing of a collective bargaining settlement is illustrated on the sample worksheet. This example assumes that at the time of the settle ment straight-time hourly earnings averaged $3 and that total supplementary benefits were $1 an hour worked, providing total compensa tion of $4 an hour worked (the base). Also assumed is a creep factor of 20 percent, 2,000 annual working hours per employee, and a 3-year agreement effective January 1, 1971, providing the immediate and deferred wage and benefit improvements shown on the work sheet. The settlement provides a 7.1 percent first-year package and a 6.5 percent annual rate of increase over the 3-year term. The worksheet also shows the wage and benefit gains scheduled for each of the 3 calendar years (1971, 72, and 73), from which material the series on changes actually placed into effect is developed,1 and the wage-rate changes apart 1 from benefit improvements. The latter data are computed without reference to creep and relate wage gains to average hourly earnings rather than average hourly compensation. Presentation and Analysis Press releases covering wage changes and wage and benefit changes in major collective bargaining settlements are issued toward the end of the month following the close of each 1 1 Thus, the series on effective package changes essentially is based on the pricing of items at the time settlements are reached. The only exception is the subsequent addition of cost-of-living escalator wage changes. 205 quarter. These releases contain preliminary data for the first 3, 6, and 9 months of a year and for the full year. This material also ap pears in Current Wage Developments (C W D ). Final quarterly material (both for individual quarters and the cumulative quarterly material appearing in the press releases) is included in the CWD article for the full year. An annual summary also appears in the Monthly Labor Review. Summary data for recent individual quarters and 4-quarter periods are shown monthly in CWD together with other statistical series depicting aspects of change in employee com pensation. The presentation facilitates analysis of the interrelations between the series and the divergences in their movements. Uses and Limitations Package cost data are used extensively by union and management officials, for whom data on developments in other firms and industries often provide an important criterion for their own deliberations. In a different vein, the data are examined by government officials and private analysts, concerned with the economic repercussions of collective bargaining on the costs of individual employers and on wageprice-employment relations within the eco nomic system as a whole. The user of the data should remember that the series does not purport to measure all changes in average hourly expenditures for employee compensation. Estimates are derived under the assumption that all factors affecting employer outlays other than contract modifica tions are constant. Nevertheless, changes in the volume of over time and shift work, in the composition of the work force, in the level and stability of employ ment, in factors affecting incentive earnings, etc., are not unusual, and will influence outlays for employee compensation. In some instances, these changes are introduced by management specifically to offset costs of new labor agree ments. In other cases, changes are the result of modified production schedules or of tech nological developments independent of collec- Date Priced Out Previous Expiration Date ----Date Negotiated: -------------Effective Date of Agreement -New Expiration Date ---------- 3/15/71 1 2 /3 1 /7 0 1 /15/71 1 /1 /7 1 Escal. Clau se I n f o r . (if rnsT an a ly sts 206 parkaor worksheet ap p lic ab le) Increases 12/31/73 STATISTICAL INFORMATION MEASURE: First-Year Wages & Benefits First-Year Wages Alone Wages Over Life of Contract Wages and Benefits Effective in Wages and Benefits Effective in Wages and Benefits Effective in --- Exclude •With creep Wages : 5% @ 3 . 0 0 S k i l l a d i u s t - 5d f o r 20% Impr. s h i f t d i f f . add. 2 d h r . on 2 nd s h i f t 30% = . 6C ad d. 3d h r . on 3rd s h i f t 20% = . 6d 2% @ 3 . 1 7 2 5% @ 3 . 2 3 5 3% @ 3 . 3 9 7 2% @ 3 . 4 9 9 1 /1 /7 1 1/1 /7 1 1 /1 /7 1 for creep 15.00 18.00 1 .0 0 1 .2 0 1 .2 0 1.44 7.61 19.42 12.23 8.40 for 10/1/71 7 /1 /7 2 1 /1 /7 3 7 /1 /7 3 6.34 16.18 10.19 7.00 H o l i d a v s : 1 ad d. 1 /1 /7 3 $3,499 @ 8 h r s . = 2 7 .9 9 t 1972 = h r s . = Vacations: Impr. a v e r a g i n g 20 add. h r s . o f v a c . / y r . f o r a l l em pl s. 1 /1 /7 2 $ 3 , 2 3 5 @ 20 h r s . = 6 4 . 7 0 a 1980 hr s= added impact in 1973 $ 3 , 4 9 9 @ 20 h r s . = 6 9 . 9 8 a 19 7 ? h r s . = 3 . 55d ( 3 . 55d - 3 . 2 7 d 5 = P e ns io ns add. 5d / h r . t o fund 1 / 1 / 7 8 5d @ 2 0 8 0 h r s . = $ 1 0 4 . 0 0 A i q 7 ? = 1.42 3.27 .28 5 97 Wages and Benefits Effective in --Wages and Benefits Effective in — - Estimate avg. c o s t o f $1 00 a 1972 = $100/em p l. 5.07 Wages Alone Effective in ---------Wages Alone Effective in Wages Alone Effective in Wages Alone Effective in Wages Alone Effective in Other TOTALS- WaSes and f r i n g e s 83,61 --56.91 Remarks: Note: * * For wages and b e n e f i t s , d i v i d e by the b a s e ; f o r wages a l o n g , d i v i d e by the A . H . E . For th e e f f e c t i v e m e a s u r e s , d i v i d e by the a d j u s t e d base or A . H . E . as o f the s t a r t o f the a p p r o p r i a t e y e a r . Priced Out By: John Doe Over l i f e t or con tract * No data should be supplied in these columns. When verification is made, the data will be supplied. BLS H ANDBOOK OF M ETHODS Wages and Benefits (over life of contract) in c e n t s P r o v is ion N/A MEASURING COLLECTIVE BARGAINING SETTLEMENTS tive bargaining, and may either add to or subtract from the cost of the union-manage ment settlement. In any event, an important influence on the level of employee compensa tion, social insurance taxes, is essentially out side the scope of the package cost estimates. Two other factors must be considered. First, package costs are only estimates of future changes in employer outlays. As already em phasized, completely accurate estimates should not be expected. Secondly, the data apply pri marily to settlements for 5,000 workers or more. Although package cost estimates are ex tremely valuable as comprehensive measures of change resulting from union-management ne gotiations, to use the estimates as precise, unambiguous, and unfailing measures of the economic effects of collective bargaining is adding an assignment which the data are in capable of fulfilling. Technical References Number 1. David, Lily Mary and Sheifer, Victor J. "Estimating the Cost of Collective Bargaining Set tlements,” Monthly Labor Review, May 1969, pp. 16-26. Reprint No. 2617. A more detailed description of the Bureau of Labor Statistics procedures for pricing collective bargaining settlements. 2. Sheifer, Victor J. "Th e Relationship Between Changes in Wage Rates and in Hourly Earn ings,” Monthly Labor Review, August 1970, pp. 10-17. Reprint No. 2688. Compares general wage rate and hourly earnings changes in 87 manufacturing estab lishments over a 7-month period. The issues raised are pertinent in any consideration of the relation between package cost estimates and actual changes in hourly compensation expenditures. 207 -V ic t o r J. S h e i f e r Chapter 24. Wage Chronologies and Salary Trend Reports Background Most Bureau of Labor Statistics series show ing the movement of money wages— such as data on average hourly earnings— apply to large aggregates of workers, e.g., all produc tion workers in a given manufacturing indus try either nationwide or in a particular State or local area. Wage chronologies and salary trend reports, on the other hand, apply to more narrowly defined employee groups. Chronologies report on wage-rate changes made by specific employers. Except for the study on Federal Classification Act employees, they deal with developments under collective bargaining agreements. In all cases they report on supplementary benefit as well as wage changes. Salary trend reports present and analyze changes in salaries of selected categories of government employees.1 In most instances, however, they do not contain separate data for individual employers. Both programs date back to the early postWorld War II period. The wage chronology program was instituted in 1948 and the first salary trend report was issued in 1950. To the extent possible, material for earlier years was included in the initial reports. Description o f Program Chronologies. The following 32 chronologies currently are being maintained :1 2 1. 2. 3. 4. 5. 6. 7. 8. Aluminum Company of America American Viscose Division of FMC Corp. The Anaconda Company Armour and Company A .T. & T.— Long Lines Department Berkshire Hathaway Inc. Bethlehem Atlantic Shipyards Bituminous Coal Mines 1 Although other BLS studies report on salary trends, they do not have this specific title. The time series in these studies are often by-products of repetitive Bureau survey activity. 2 Several chronologies have been discontinued, generally either because of fragmentation of bargaining units, declining importance, or because standardization of collective bargain ing eliminated the value of more than one chronology in a given industry. 9. The Boeing Company 10. Commonwealth Edison Company 11. Dan River Mills 12. Federal Classification Act Employees 13. Firestone Tire and Rubber Company and B.F. Goodrich Company 14. Ford Motor Company 15. International Harvester Company 16. International Paper Company (Southern Kraft Division) 17. International Shoe Company 18. Lockheed-California Company (A division of Lockheed Aircraft Corp.) 19. Martin-Marietta Corp. 20. Massachusetts Shoe Manufacturing 21. New York City Laundries 22. North America Rockwell Corp. 23. North Atlantic Longshoring 24. Pacific Coast Shipbuilding 25. Pacific Gas and Electric Company 26. Pacific Longshore Industry 27. Railroads— Nonoperating Employees 28. Sinclair Oil Companies 29. Swift & Company 30. United States Steel Corp. 31. Western Greyhound Lines 32. Western Union Telegraph Company Each chronology covers either a single wagedetermination unit or a group of closely related units. It may cover an individual company and union (e.g., Ford Motor Company and the United Automobile Workers), a single com pany and two or more unions (e.g., the Alumi num Company of America and the United Steelworkers and the Aluminum Workers), a group of employers and a single union (e.g., New York City Laundries and the Amalga mated Clothing Workers), a group of com panies and a group of unions (e.g., Pacific Coast shipbuilding companies and a number of craft unions), or a governmental body (e.g., the chronology covering Federal Classification Act employees). The program is designed to summarize long term wage-benefit developments in a variety of industrial environments. Accordingly, chro nologies cover groups that (1) have existed over a period of years; (2) are important in their own industry; (3) employ a significant number of workers; (4) are of general public interest; and (5) are willing to cooperate with the Bureau and for which appropriate infor mation is available. 209 210 BLS HANDBOOK OF METHODS Each chronology is divided into a narrative synopsis of the collective bargaining or legisla tive developments resulting in wage and benefit changes and a tabular summary of the changes themselves. Each chronology contains separate tables showing general wage changes and changes in supplementary benefits. As used in the wage chronology program, general wage changes are defined as upward or downward changes that affect an entire unit or a substantial group of workers at one time. Excluded are adjustments in individual rates (such as promotions, or merit or seniority in creases) and minor adjustments in wage struc ture (such as changes in individual job rates or incentive rates) that do not have an im mediate and noticeable effect on the general wage level. Because of the omission of non general wage changes, fluctuation in incentive earnings, and other factors, the sum of the wage changes listed in each chronology will not coincide necessarily with the movement of average hourly earnings over the same period. The tables generally include benefits such as guaranteed minimum earnings, shift premiums, daily and weekly overtime, weekend premiums, pay for holiday work, paid vacations and holi days, other paid leave provisions, reporting time, waiting time, paid lunch periods, pay for travel time, and health, welfare, and pension benefits. When minimum plant rates, common labor rates, occupational wage rates, or rates for labor grades are important in the wage struc ture, they are shown in chronological sequence, parallel to the general wage changes. Salary trend reports. These reports currently are issued for Federal classified employees, firefighters and police patrolmen, and urban public school teachers. They all contain indexes of long-term salary movements. Data on recent and current salary levels also are provided. Reports for firefighters, police patrolmen, and teachers basically apply to cities of 100,000 population or more, and provide separate fig ures for regions and city-size groups, as well as overall national data. Material on individual cities, however, is not shown. Because it deals with a single employee group, the report on Federal classified employees contains consider able detail on developments in wage structure. Data Sources Both wage chronologies and salary trend reports are developed primarily from published data. Collective bargaining agreements, pen sion and welfare documents, and newspaper and periodical articles provide the main sources of chronology information. These are supple mented by direct requests to the parties for information when available written records are inadequate. Thus, most of the information used to compile chronologies is, in one form or another, already a matter of public knowledge. In all instances, to avoid dissemination of erroneous material, pre-publication drafts of reports are submitted to the parties for their review and comments. Salary trend reports are prepared largely from salary data collected by other groups. That for Federal classified employees is based on data published by the U.S. Civil Service Commission in its annual report on Pay Struc ture of the Federal Civil Service. Reports on teachers and firefighters and police patrolmen are based mainly upon com pilations of data for individual cities made by the National Education Association and the International City Management Association, respectively. These are supplemented by annual surveys of salaries and working conditions conducted by the International Association of Fire Fighters and the Fraternal Order of Police and, when necessary, by direct inquiries by the Bureau of Labor Statistics. Statistical Procedures Statistical analysis in these programs is confined largely to the preparation of indexes of salary movements of groups of government employees. Sampling problems do not arise since in each case an effort is made to examine the total universe, i.e., all Federal classified employees, and all teachers, firefighters, and police patrolmen in cities of 100,000 population or more. WAGE CHRONOLOGIES AND SALARY TREND REPORTS Indexes generally are computed by a method that minimizes the effect of year-to-year changes in relative employments in the cities or occupational categories covered. As a rule, chain indexes are employed, i.e., the index for the current year is obtained by adjusting the index for the prior year by the percentage change in average salaries over the intervening period. Normally, to preserve a pure measure of salary change, average salaries for each of the two years are computed using current-year employments as weights. Presentation and Analysis Wage chronologies. Wage chronologies are published individually as BLS Bulletins and revised periodically to incorporate material resulting from new collective bargaining settle ments or legislative developments. Bulletins are updated after every other contract settlement or legislative enactment. Intervening develop ments are reported in supplements to existing bulletins. Thus, when 3-year collective bar gaining agreements are negotiated, a revised bulletin should be issued once in 6 years. Salary trend reports. Articles covering develop ments for Federal classified employees and fire fighters and police patrolmen appear annually in Current Wage Developments. Press releases containing summary data for firefighters and police patrolmen precede publication of the articles. Salary changes for teachers are re ported on in biennial CWD articles, since the basic data are issued at two-year intervals. Reprints are available of all CWD articles. In addition, all articles issued up to the mid-1960’s have been collected and reprinted in the BLS Bulletins listed in the technical references at the end of this chapter. Chronologies are primarily listings of wage and benefit changes, with background material limited to descriptions of the collective bar gaining or legislative processes leading up to the changes. Greater effort is made in salary 211 trend reports to analyze the data. Background factors are presented and the wage movements are compared with wage changes for other employee groups. Uses and Limitations Both wage chronologies and salary trend re ports serve two audiences; they are useful as sources of comparative wage data for union, management, and government officials engaged in wage setting and as research tools for eco nomic analysts. Chronologies are particularly useful for negotiators because the studies pre sent detailed information on developments in units that not infrequently provide wage lead ership for their industries. Moreover, com parisons of wage and benefit changes in such units provide valuable insights into wage set ting in the American economy. In addition, the data help to explain the movements in aggrega tive statistics such as the Bureau series on average hourly earnings.3 Salary trend reports are noteworthy as one of the relatively few sources of data on wage movements and levels in the government sector. Although wage chronologies describe changes in supplementary benefits, they do not measure the effect of these changes on employers’ hourly labor costs. In this connection, decisions on whether or not to adopt a given benefit change may hinge on its cost, which may vary among bargaining units with work forces of differing composition. The temptation may be to use salary trend reports as indicators of salary movements for government employees in general. However, the particular groups covered by these reports are by no means a representative sample for this purpose. 3 For examples of the use of chronologies by economists, see Harold M. Levinson, Postwar M ovem ent of Prices and Wages in Manufacturing Industries, Study Paper No. 21, Study of Employment, Growth, and Price Levels, Joint Economic Committee, 86th Cong., 2d Sess. (Washington: 1960); and John E. Maher, “ An Index of Wage Rates for Selected In dustries, 1946-1957,” Review of Economics and Statistics, August 1961, pp. 277-282. 212 BLS HANDBOOK OF M ETHODS Technical References Num ber 1. Arnow, Philip, Bloch, Joseph W ., and Quant, Willis C. “ The New Wage Chronology Series,” Monthly Labor Review, December 1948, pp. 581-583. Describes the aims of the chronology series at the time it was introduced. 2. U.S. Department of Labor, Bureau of Labor Statistics. A Directory of Wage Chronologies, 1948-June 1969 (1969). Lists chronologies that were maintained in 1969, the time span covered, and the place of publication. 3 . ---------- Salary Trends: City Public School Teachers, 1925-65 (Bulletin 1504, 1966). 4 . ---------- Salary Trends: Federal Classified Employees, 1989-64 (Bulletin 1444, 1965). 5. --------- - Salary Trends: Firemen and Policemen, 1924-64 (Bulletin 1445, 1965). Compilations of previously published salary trend articles. — V ic t o r J. S h e i f e r Productivity and Technology Chapter 25. Output Per Man-Hour: Private Sector Background and Description of Measures To provide information about the relation ship between productivity, prices, wages, em ployment, and economic growth, the Bureau of Labor Statistics publishes indexes of output per man-hour, compensation per man-hour, unit labor costs, and related costs for broad economic sectors. Measures of output per man hour have been developed for the total private sector and the farm and nonfarm sectors an nually from 1909 to the present. Since 1947, these data have been supplemented with com parable measures on hourly compensation and related costs for these sectors as well as manu facturing. Post World War II indexes are avail able quarterly as well as annually. In addition, to the farm, nonfarm, and manufacturing mea sures, annual information on productivity and costs is available for the nonmanufacturing sec tor and its component major industrial sectors. Productivity measures, first published in 1959,1 represent the culmination of a long series of developments in productivity measurement in the Bureau.1 2 The Bureau of Labor Statistics’ output per man-hour measures specifically refer to the ratio between dollar gross product (GNP) originating3 in the private or individual sec tors and the corresponding hours of all persons employed. Indexes of output per man-hour in dicate the relationship between output and la bor input (man-hours). Index changes through time show the effectiveness of man-hours in producing current levels of output. Man-hours are based mainly on BLS estab 1 Trends in Output per Man-Hour in the Private Economy, 1909-58 (BLS Bulletin 1249, 1959). 2 For a detailed description of the historical development of the productivity measurement program of the BLS, see Chapter 26. 3 Gross Domestic Product equals Gross National product less the net return on foreign investments. lishment payroll data on employment and hours. These man-hours refer to hours of work and paid time for vacation holidays and sick leave. The Bureau also develops an output per man-hour series based on labor force man-hour data from the Current Population Survey (CPS). The concept underlying the labor force man-hours is hours worked, rather than hours paid. Theoretically, the difference between the labor force man-hours series and the establish ment payroll data series is equal to paid vaca tion time, sick leave, and other paid leave. However, actual differences in man-hours be tween the two series also reflect statistical variation and differences in method. Indexes of compensation per man-hour meas ure the hourly costs of wages and salaries, in addition to supplemental payments, such as the contributions of employers to social security, unemployment insurance tax, and payment for private health and pension plans. Measures of real compensation per man-hour reflect the adjustment of hourly compensation for changes in the Consumer Price Index. Unit labor costs measure the cost of labor input required to produce one unit of output and is derived by dividing compensation per man-hour by output per man-hour. Unit non labor payments (costs) measure the cost of nonlabor inputs such as depreciation, rents, interest, indirect taxes, in addition to profit income such as corporate profits and income of proprietorships and partnerships. Data Sources and Estimating Procedures Output. The real gross national product orig inating in the private or individual sectors is used to prepare output per man-hour estimates. Gross national product is the market value of final goods and services produced within a certain time period. It includes purchases of goods and services by consumers gross private domestic investment, net foreign investment, and purchases by Government. GNP is equal to income received by labor and property for 213 214 BLS HANDBOOK OF METHODS services rendered in the current production of goods and services, in addition to capital con sumption allowances, indirect business taxes, and several other minor items. Gross national product in current dollars cannot be used directly as the output measure since it reflects price changes in addition to changes in physical volume. The Office of Busi ness Economics (OBE), U.S. Department of Commerce, prepares estimates of constantdollar GNP for the total private sector and major sectors. These estimates indicate only changes in the volume of production and are used to develop output per man-hour measures.4 Compensation. The OBE develops compensa tion of employees data as part of the national income accounts. These data include direct payments to labor, such as wages and salaries inclusive of executive compensation, commis sions, tips, and bonuses and payments in kind which represent income to the recipients and supplements to these direct payments. The last item consists of employer contributions for social insurance, private pension and health and welfare plans, compensation for injuries, doctors’ fees, pay for military leave, etc. Since these compensation measures refer only to wage and salary workers, they do not reflect the cost of labor expended by proprietors and unpaid family workers. Labor costs can be seriously underestimated in sectors like the farm where proprietor man-hours represent a substantial portion of the labor input. The Bu reau, therefore, imputes a payment for labor services of proprietors and family workers. Compensation per man-hour of proprietors is assumed to be the same as that of the average employee in that sector. Unit labor and nonlabor costs. The Bureau also presents data on labor and nonlabor costs per unit of output for the private sector and its 4 A detailed description of the methods and procedures of estimating GNP and GDP in current and constant dollars is given in the 1954 National Income Supplement to the Survey to Current Business, U.S. Department of Commerce. Further information on Gross Product Originating estimates for major industry sectors is presented in the October 1962 issue of Survey of Current Business. major components. Unit labor costs relate hourly compensation of all persons to output per man-hour. Unit nonlabor costs are com puted by subtracting compensation of all per sons from current dollar GNP and dividing by output. Labor input (man-hours). The primary source of man-hours and employment data is the BLS Current Employment Statistics (CES) pro gram, which provides data on employment (all employees and production or nonsupervisory workers) and average weekly hours of produc tion workers in nonagricultural establishments. Jobs rather than persons are counted, so that multiple jobholders are counted more than once. Weekly hours are hours paid rather than plant hours. These statistics are based on pay roll records from a sample of establishments where probability of sample selection is pro portionate to the establishment size; large establishments (relative to the sector) fall into the sample with certainty. Data on em ployment, hours, and earnings are collected monthly; however, the reference period for these data is the payroll period including the 12th of the month. Methods are described in Chapter 2. Establishment data are published monthly in Employment and Earnings and in an annual summary. Since the CES includes only nonfarm wage and salary workers, data from other sources (National income accounts or the CPS) are used for the uncovered sectors (farm, proprie tors, unpaid family workers, and private household workers) to develop employment and man-hour estimates for the total private sector. Separate estimates for employment and man hours paid are developed for each major indus trial sector and these are aggregated to total private and private nonfarm levels. Each man hour is treated as a homogeneous unit; no dis tinction is made between workers who have different skill levels or rates of pay. In the manufacturing sector, separate esti mates for production and nonproduction worker man-hours are derived and then aggre gated to the manufacturing total. Production worker and nonproduction worker employment and production worker average weekly hours OUTPUT PER MAN-HOUR: PRIVATE SECTOR are taken directly from CES data. Average weekly hours for nonproduction workers are developed from BLS studies of wages and sup plements in the manufacturing sector which provide data on the regularly scheduled work week of white-collar employees.5 For nonmanufacturing sectors, employment and weekly hours paid are taken from the pay roll series. Although payroll average weekly hours data refer to nonsupervisory workers for man-hours computation, the assumption is that the length of the workweek in each nonmanu facturing industry is the same for all wage and salary workers. When establishment employment data are not available (proprietors, unpaid family workers, and private household workers), either labor force data (CPS) or national in come employment data are used. On the other hand, average weekly hours are based on labor force data for hours worked rather than hours paid. However, persons who have a job but are not at work are assumed to have been paid for an average workweek in the sector where the job was located. Analysis and Presentation Indexes of output per man-hour show changes in the ratio of output to labor input (man-hours). These indexes relate output to man-hours but should not be interpreted as representing labor’s sole contribution to pro duction. Rather, they reflect the interaction of many forces in addition to labor and skill, such as changes in technology and increased capital investment per worker. BFor manufacturing nonproduction workers, average hours are estimated as follows: Estimates of vacation time, holidays, paid sick leave, and personal time off are subtracted from an estimate of scheduled annual hours paid nonproduction workers: scheduled annual hours are derived by extrapolat ing the 1959 level of scheduled weekly hours with data from BLS Area Wage Surveys, and then multiplying by the num ber of workweeks in each year; the level of scheduled weekly hours for nonproduction workers was calculated from data collected by BLS for the study, Employer Expenditures for Selected Supplementary Remuneration Practices for Production Workers in Manufacturing Industries, 1962, BLS Bulletin 1428 (1965): estimates of vacation time, holidays, sick leave, and personal time off are derived primarily from data from Area Wage Surveys and Social Security A d ministration studies. 215 For economic aggregates like productivity in the private sector, changes over time reflect movements within the various component in dustries as well as shifts in the relative im portance of each of the industries. For example, increases in output per man-hour are influenced not only by the increments or decrements in the component sectors but also by the shift from low to high productivity industries. Within industries, other forms of shifts also take place and are not accounted for ade quately. In output measures, for example, changes in income and taste may be reflected in shifting consumption patterns to higher quality goods or to services rather than goods. Short-term movements in productivity and unit labor costs often result from cyclical varia tion which tends to distort the underlying relationship between the numerator and de nominator of the ratio. For example, because of market imperfections, an employer often has difficulty in adjusting his staff immediately to changes in production. Sharp increases or de creases in demand usually are followed by erratic movements in productivity and unit labor costs. When longer periods are analyzed, however, these fluctuations tend to iron out and a secular trend is more apparent. Long-term productivity trends tend to move more smoothly over time and present a more stable picture of historical patterns than short term movements. The latter tend to reflect temporary changes in level of demand, utiliza tion of productive resources, and other short term phenomena, such as strikes, which affect current conditions but should not be expected to continue into the future. The period selected for analysis may affect these long-term trends, so that some element of judgment is present in any analysis of trend. Indexes of output per man-hour, compensa tion per man-hour, and related cost data are published quarterly in the BLS press release, Productivity and Costs. Historical indexes of these data are available on request. Detailed procedures for developing output per man-hour indexes appear in Trends in Output per ManHour in the Private Economy, 1909-58 (BLS BLS HANDBOOK OF METHODS 216 Bulletin 1249, 1959). Indexes of output per man-hour and related cost data appear in Em ployment and Earnings, the Monthly Labor Review, the Handbook of Labor Statistics, The Economic Report of the President, the Man power Report of the President, and the Sta tistical Abstract of the United States. in hourly compensation tends to increase unit labor costs, and an increase in output per man hour tends to reduce these costs. Therefore, through its relationship to unit labor costs, out put per man-hour is a crucial element in the wage-price relationship. It indicates the extent to which compensation gains can occur without putting pressure on prices or reducing profits. Uses and Limitations Certain characateristics of the output per man-hour and related cost data should be recognized when applying them to specific situations. First, the data reflect not only changes in various component industries but also changes in the relative importance of these industries. Second, these measures represent the culmination of a chain of economic events. Data for a single time period often do not show direct causal relations and should be interpreted in light of previous economic events as well as current conditions. Third, underlying concepts and data available for estimation limit to some extent measures of productivity, out put, compensation, and employment. In fact, in several sectors where output data are difficult to obtain, output changes are equal to employ ment changes. This definition understates pro ductivity growth. Consequently, the meaning of these measures should be interpreted with caution because of conceptual and practical limitations and statistical errors which can arise in any numerical series. Data on output per man-hour, compensation per man-hour, and related costs are designed for use in economic analysis and public and private policy planning. These data apply in areas such as wage determination and analysis of prices and living conditions. An especially relevant use of output per man-hour and related costs is the relationship of productivity, wages, prices, profits, and costs of production. Within the framework of na tional income and product accounting, gross national product for a specific time period represents the market value of all final goods and services produced or the sum of all costs of production— compensation, profits, deprecia tion, interest, etc. Unit labor costs, or compen sation per unit of output, represent a major portion of total unit costs and reflect the com bined effect of changes in output per man-hour and compensation per man-hour. An increase Technical References Number 1. Edward Denison, W hy Growth Rates Differ; Sources of Economic Growth, Washington, D.C., The Brookings Institution (1967). A study of output and productivity growth in 9 Western countries. Includes a discus sion of factors affecting productivity growth and the way varying effects of these factors can be attributed to differential growth rates between countries. 2. Kendrick, John W ., Productivity Trends in the United States (National Bureau of Eco nomic Research, Number 71, General Series), Princeton, N.J., Princeton University Press (1961). A presentation of historical measures of output, input, and productivity for the U.S. economy and industry groups, including descriptions of concepts and methods of measure ment. Also includes discussion of implications of productivity change for economic growth, prices, incomes, and resource allocation. 3. National Bureau of Economic Research, Conference on Research in Income and Wealth, “ Output, Input, and Productivity Measurement, (Studies in Income and Wealth, Vol. 25), Princeton, N. J., Princeton University Press (1961). A collection of papers and comments devoted to an appraisal of the measurement of output, input, and productivity. OUTPUT PER M A N -H O U R : PRIVA TE SECTOR Technical References— Continued 4. 5. 6. 7. 8. 9. National Bureau of Economic Research, Conference on Research in Income and Wealth, “ In dustrial Composition of Income and Product,” (Studies in Income and Wealth, Vol. 32), New York, Columbia University Press (1968). A selection of papers concerning the development of output, input, productivity and cost measures for individual industrial sectors. National Bureau of Economic Research, Conference on Income and Wealth, “ Production and Productivity in the Service Economy,” (Studies in Income and Wealth, Vol. 34) New York, Columbia University Press (1969). A collection of papers concerning the concepts, definitions, procedures, and data limita tions of measuring output, input and productivity in service producing industries. U.S. Department of Labor, Bureau of Labor Statistics, Trends in Output per Man-Hour in the Private Economy, 1909-1958, (Bulletin 1249, 1959). A presentation and analysis of output per man-hour indexes and trends for the total private U.S. economy and major sectors. Includes a description of methods and sources for developing output per man-hour measures. U.S. Department of Labor, Bureau of Labor Statistics, Productivity: A Bibliography (Bul letin 1514,1966). A collection of nearly 600 references concerning productivity and productivity measure ment. Each reference includes a brief annotation giving the gist of the subjects covered. Mark, Jerome A ., Wage-Price Guidepost Statistics: Problems of Measurement, American Sta tistical Association, Proceedings of the Business and Economic Statistics Section (1968). A paper describing some of the problems of developing the measures which were used in the specification of the guideposts. Ziegler, Martin, “ Productivity in Manufacturing,” Monthly Labor Review, October 1967. An analysis of the effect of changes in capacity utilization on productivity. — Shelby A . H erm an 217 Chapter 26. Output Per Man-Hour Measures: Industries Background Studies of output per man-hour for individ ual industries have long been a part of the BLS program. The first Commissioner of Labor, Carroll D. Wright, conducted a study of 60 manufacturing industries in 1898. The findings in his report on “ Hand and Machine Labor,” provided striking evidence of the savings in labor resulting from mechanization in the last half of the 19th century. Commissioner Wright’s study was prompted by concern on the part of Congress that human labor was being displaced by machinery. The impact of productivity advance upon employment re mained an important focus of the BLS program throughout the 1920’s and 1930’s. It was also during this period after World War I that the Bureau began the preparation and publication of industry indexes of output per man-hour, based on available production data from the periodic Census of Manufactures and employ ment statistics collected by the BLS. In 1940, Congress authorized the Bureau of Labor Statistics to undertake continuing stud ies of productivity and technological changes. The Bureau extended earlier indexes of output per man-hour developed by the National Re search Project of the Works Projects Admin istration, and published measures for selected industries. This work was reduced in volume during World War II, owing to the lack of meaningful production and man-hour data for many manufacturing industries. The advent of World War II also caused a change in program emphasis, from problems of unemployment to concern with the most effi cient utilization of scarce manpower. The BLS undertook a number of studies of labor require ments for defense industries, such as synthetic rubber and shipbuilding. After the war, the industry studies program resumed on a regular basis, and was supplemented by a number of industry studies based on the direct collection of data from employers. Budget restrictions after 1952 have prevented the continuation of direct collection of data. Consequently, the preparation of industry measures is limited to those industries where readily available data can be utilized to construct measures. In recent years, public interest in produc tivity has grown, and there has been greater recognition that increases in output per man hour are important indicators of economic progress and the means to higher levels of in come, rather than merely a threat to job oppor tunities. The Industry Studies program covers a variety of manufacturing and nonmanufactur ing industries. For these industries, indexes of output per man-hour, output per employee and the related data on output, employment, and man-hours are prepared and published on an annual basis. The indexes are generally avail able for most years from 1947 to the most recent year for which data are available, and for many industries also for 1939. Concepts Industry indexes of output per man-hour measure changes in the relationship between the physical volume of output of an industry and the man-hours expended in that output. Although, traditionally, output per man-hour has been the measure most frequently used, discussion of physical output per man-hour often is simplified if conducted in terms of its reciprocal: man-hour requirements per unit of output (unit man-hours). Therefore, this form of index is used in the following description. For an industry producing a single uniform output, the unit man-hours index is simply the ratio of the man-hours expended to produce a unit of output over two periods of time. This ratio may be expressed as follows: Where /„ represents the unit man-hour index, 7P represents the output per man-hour index, and li and l0 denote unit man-hours expended in the current and base periods, respectively. For an industry producing a number of prod ucts— the more typical case— the unit man hours index is the ratio for two periods of the 219 220 BLS HANDBOOK OF M ETH ODS total hours required for the output of a given composite of products. Indexes of such indus tries vary with the composite of products spe cified and can take many forms. Letting q0 and Q represent base period and current period i quantities of a given product, respectively, two of these forms are: a. Using a current period composite Methods and Sources / ZqJ„ Output Per Man-Hour b. Using a base period composite SqJ* Zqn l„ An index constructed according to (a) com pares the man-hours expended in the produc tion of the current composite with man-hours which would have been required to produce the current composite in the base period. An index constructed according to (b) compares the man-hours required in both periods to pro duce the base period composite. These indexes thus eliminate the effects of variations over time in the relative importance of products on unit man-hours. In either form, an index of unit man-hours also can be viewed as the quotient of an index of man-hours and an index of output, i.e., Man-hours index j _ ZLqi Zhq,, Output index _ Unit man-hours ’ (Laspeyres) index (Paasche) SLffi _ 2 l<qi SJotfi Man-hours Output index Unit man-hours index index -s- (Paasche) = (Laspeyres) u~ unit man-hours index employs a base period weighted output index divided into the man hours index. Conversely, a base period weighted unit man-hours index is consistent with an output index which utilizes current period weights. ' 2.l„q0 ~ ~ tL q 7 ~ tL q 0 ~ ~ tlq 7 The man-hours index measures the change in aggregate man-hours between the base and current periods. The man-hours data are the total hours expended by employees in establish ments classified in the industry, in producing the base period and current period composites. As can be seen in the formulas, the appro priate output index is one which compares the quantities of the various products in the cur rent and the base periods, each weighted by the man-hours expended per unit produced in a given period. A current period weighted The Bureau of Labor Statistics computes an index of output per man-hour by dividing an output index by an index of aggregate man hours. Measures are prepared separately re lating output to (a) all employee man-hours, (b) production worker man-hours, and (c) nonproduction worker man-hours. (The stan dard definitions of production workers and nonproduction workers are used.) Three cor responding measures also are computed relat ing output to the number of employees. Output BLS industry output indexes are based primarily on the physical output of the prod ucts of the industry combined with fixed period weights. However, the availability of quantity data on physical output varies among indus tries, and, for manufacturing and mining in dustries, may vary depending on whether the data are for a year when a Census was con ducted or for a noncensus year. For manu facturing and mining industries, quantity data on physical output are usually most compre hensive for years covered by a Census. To make maximum use of the comprehensive census data, output indexes are derived from data for two consecutive censuses; these in dexes are referred to as benchmark indexes. For intercensal years, annual indexes are based on either physical output data (generally, in less detail than for Census years) or if such data are not available, value of output adjusted for price change (i.e., the value of output in constant dollars). The annual series subse quently are adjusted to the benchmark levels for the census years. OUTPUT PER M A N -H O U R Weights. The mathematical form of the output index implies use of unit man-hour weights, and such weights are used whenever possible. These weights are derived from special surveys or from data published for specialized estab lishments in the Census of Manufactures. In some industries, however, unit man-hour infor mation is not available for individual products. Consequently, the BLS uses substitute weights which are assumed to be proportional to unit man-hour weights. Usually these are unit value weights. Unit value weights are computed from Census or survey data on the quantity and value of shipments of the primary products of the industry. The introduction of these substi tute weights results in an industry output per man-hour index which reflects shifts in value per man-hour of the various products in the industry. Thus, a change can occur in the index without any change in the output per man-hour for any product of the industry. The extent to which error or bias may be introduced by the use of unit value weights is not known. The index is equivalent to one weighted with unit man-hours if the unit man hours and unit values among the products are proportional or if there is no correlation be tween the relative change in quantity and value per man-hour1 There is evidence that unit values are fairly reliable approximations for individual products where wages constitute a large proportion of total value of output. The error generated in the output index by an error in the weights is generally considerably smaller than the error in the weights themselves. In some industries, unit value weights for specific products and unit man-hour weights for product groups are used at different stages in constructing the industry output indexes. When this procedure is used, the individual products are first aggregated into primary product group indexes with unit value weights. 1 See Irving H. Siegel, “Further Notes on the Difference Between Index Number Formulas,” Journal of the American Statistical Association, December 1941, pp. 519-524. 2 The “specialization ratio” is the value of shipments of primary products of plants in the industry as a percent of total shipments of all products (primary plus secondary) made by these same establishments. The “ coverage ratio” is the value of shipments of the primary products made by plants classified in the industry as a percent of the total ship ments of the industry’s primary products made by all pro ducers, both in and out of the specified industry. M EASURES: IN D U S T R IE S 221 These indexes in turn are combined into an industry output index with primary product group man-hours. The primary product group man-hours relate to a base period, as do value weights. To obtain primary product group man-hour weights, total man-hours for plants specializing in each primary product class are derived from published census data on production worker man-hours and nonproduction worker employ ment. These are supplemented by unpublished BLS estimates of nonproduction worker man hours. (See page 223 for the procedures used to estimate nonproduction worker man-hours.) Ratios of man-hours to value of shipments are multiplied by the corresponding value of pri mary products shipped by the entire industry to yield the estimated primary product group man-hour weights. This procedure assumes that the man-hours per dollar for each product class shipped by the whole industry are the same as those for plants specializing in the product group. This procedure is used only when the “ specialization” and “ coverage” ratios of the industry are high and specializa tion data for all or most of the product groups are available.1 2 Most published industry indexes have used 1947 weights for 1947-58, 1958 weights for 1959-63, 1963 weights for 1964-67, and 1967 weights for years after 1967. The Bureau policy is to revise the weights as more current data become available from the periodic censuses. Benchmark indexes. For most manufacturing and all mining industries, indexes reflecting changes in output between census years are constructed. These are called benchmark in dexes. For manufacturing industries, benchmark indexes are developed using the following pro cedure : Price indexes for each primary product class are developed from data on the value of each individual product within the class whether made in the industry or elsewhere. Wholesale price indexes are used wherever possible to convert the product values to con stant dollar estimates. If a wholesale price index is not available, a price index is de 222 BLS H ANDBOOK OF M ETH ODS veloped using both the quantity and value data reported for the product in the Census of Manufactures. The primary product class price indexes are derived from the sum of the current dollar values and the sum of the con stant dollar values. These “ wherever made” primary product class price indexes are used to deflate the value of primary products produced only by the in dustry. This procedure assumes that the price movements of the primary products within the industry are the same as the price movements for all primary products wherever made. These constant dollar values are related to cor responding base year values in order to derive separate primary product indexes within the industry. These separate primary product indexes in turn are combined with man-hour weights to derive the total industry primary product out put index. The index of primary products of the industry is multiplied by a “ coverage” ad justment to represent the total output of the industry. This “ coverage” adjustment is the ratio of the index of value of industry ship ments (after inclusion of net additions to in ventories) to the index of value of shipments of primary products. The final industry output index thus reflects inventory buildups and changing proportions of secondary products. For the mining industries, benchmark in dexes are computed from unweighted tonnage data as reported in the Census of Mineral Industries. Annual Indexes. Annual output indexes are * constructed by the following described pro cedures. For manufacturing and mining in dustries, the annual indexes are adjusted, if necessary, to the levels of the benchmark in dexes previously described. The adjustment factors for 2 census years are used to de termine the adjustment factors for the inter vening years by linear interpolation. 1. Physical output. Most annual output indexes are based on physical quantities of products combined with fixed-period unit man hour or unit value weights. The basic quantity data are generally primary products of an in dustry classified into product groups; the greatest amount of detail available is used. The quantity data relate to primary products “ wherever made” and in some cases to ship ments of the products. The Bureau’s annual measures of production are constructed from data on physical quanti ties of products comprising a high percentage of the total value of an industry’s output. Coverage varies between 60 and 100 percent. Complete coverage generally is obtained in mining and other well-defined industries with a relatively homogeneous output. 2. Deflated value. When adequate annual physical quantity data are not available, in dexes are derived from data on the value of industry output, adjusted for price change. Since the adjustment for price change is most often downward, the indexes usually are called “ deflated value” indexes. Such indexes are conceptually equivalent to indexes which use data on physical quantities of products com bined with unit value weights. To derive this index, data on the industry’s value of output are divided by an industry price index. An index of these deflated values shows the change in the real value of output between the past and current periods.3 Often data on value of production are not available and data on value of shipments must be used. In this case, data on value of ship ments for each year are divided by an industry price index representing the average annual price for the year. Beginning- and end-of-year finished goods and work in process inventories are also deflated. The estimated value of ship ments in constant dollars is then adjusted by the net change in inventories, also in constant "F o r example: Value Index _i_ Price index (Paasche) = Output index (Laspeyres) SPiQ, . ZP.Qi _ ZPoQt 2P„Q„ ' 2 P „Q< _ S P 0Q0 where p , and p,. represent prices of products in the industry in the current and base periods, respectively. This index requires quantities of all items produced in each year. These data are not available for the particular industries where this measure is used, and quantity data are usually available for the base year only, so that the deflated value indexes em ployed usually take the following form: Value Index SPiQi 2 P„q„ Price index ^ (Laspeyres) 2P(q„ 2P„q0 _ ~ ~ = Output index (Paasche) SP«3i 2PiQ„ OUTPUT PER M A N -H O U R dollars, to yield an estimate of the constant dollar value of production. Sources. Industry output indexes are prepared from basic data published by various public and private agencies, using the greatest amount of detail available. The Bureau of the Census, U.S. Department of Commerce, is the major source of output statistics for the manufacturing industries. The Bureau of Mines, U.S. Department of the Interior, compiles most of the information for the mining, cement, coke, and metals indus tries. Other important Government sources include the U.S. Department of Agriculture, the Fish and Wildlife Service, U.S. Department of the Interior, the Interstate Commerce Com mission, the Internal Revenue Service, and the Civil Aeronautics Board. Important sources of trade association data include the Textile Eco nomics Bureau, Inc., National Association of Hosiery Manufacturers, Inc., National Canners Association, Rubber Manufacturers Associa tion, and the American Iron and Steel Institute. For deflated value series, industry price in dexes are derived from wholesale price indexes published by the Bureau of Labor Statistics. Man-Hours An index of man-hours is computed by divid ing the aggregate man-hours for each year by the base-period aggregate. Man-hours are treated as homogeneous and additive with no distinction made between hours of different groups of employees. Data on changes in qual itative aspects of man-hours, such as skill, efficiency, health, experience, age, and sex of persons comprising the aggregate, are not used and generally not available. Man-hours indexes are developed for all employees, production workers, and nonproduction workers. Sources. Industry employment and man-hours indexes are developed from basic data compiled by the Bureau of Labor Statistics or the Bu reau of the Census. For most private nonagricultural industries (including manufacturing), the Bureau of Labor Statistics publishes em ployment and average weekly hours data for MEASURES: IN D U S T R IE S 223 production or nonsupervisory workers and em ployment data for all employees. For manu facturing industries, the Bureau of the Census publishes employment and aggregate man hours data for production workers and em ployment data for all employees (including nonproduction workers). The Bureau of the Census provides data in greater industry detail within manufacturing than BLS. The two sources differ in their definition of man-hours. The Census data include all hours at the plant, worked or paid for, and exclude paid time for vacations, holidays, or sick leave, when the employee is not at the plant. Over time and other premium pay hours are included on the basis of actual time at the plant. In contrast, the Bureau of Labor Statistics data include time for paid vacations, holidays, and sick leave, as well as plant man-hours. Differ ences in the data from the two sources for the same industry, however, also stem from sampling and reporting differences. Whenever employment and hours data are available from both the Bureau of the Census and the Bureau of Labor Statistics, the labor input data which are used are those consistent with the data on output. Thus, when output data from the Bureau of the Census are used, employment and hours data from the same source usually are preferred. Nonproduction Worker Hours. While both the Bureau of the Census and the BLS provide data on production worker man-hours, neither source provides annual data by industry on nonproduction worker nor all-employee man hours. Therefore, these measures are esti mated. The estimates of aggregate nonproduction worker man-hours for the manufacturing in dustries are derived from published employ ment data, and estimates of average annual hours worked or paid per nonproduction worker. The estimates of average annual hours worked are calculated by multiplying the num ber of work-weeks in the year times the sched uled weekly hours. This produces an estimate of average annual hours paid. Estimated hours for vacations, holidays, disability, and personal 224 BLS HAN DBOOK OF M ETH ODS time off are subtracted from average annual hours paid, to obtain an estimate for average annual hours worked. Vacation and holiday trends are based on data from various BLS surveys. Estimates of disability time are based on studies of the De partment of Health, Education, and Welfare, and data from BLS surveys. Personal time off has been estimated as a constant from refer ences in relevant publications. All employee man-hours estimates for manu facturing industries are derived by summing the aggregate man-hours for production work ers, and the estimated aggregate man-hours for nonproduction workers. Comparability of Output and Man-Hours Data Man-hours data are based on total man-hours of establishments classified in an industry, whether the man-hours are applied to produc tion of primary or secondary products. Annual physical output data, on the other hand, usually include only primary products of an industry. In addition, they are usually reported on a “ wherever made” basis. Thus, there can be some discrepancy in the coverage of output and man-hours measures. This is not a serious problem unless there is considerable variation from year to year in the proportion of primary products to total products of an industry, or if there is change in the proportion of primary products which are made in other industries. The comparability of the man-hours and out put data is indicated by the specialization and coverage ratios which the Bureau of the Census publishes. All industries in the BLS industry measurement program have high specialization and coverage ratios. In selecting industries for the measurement program, attention is also given to changes in the degree of vertical integration. Man-hours relate to all operations performed by establish ments of an industry, while output usually is measured in terms of the final product. If es tablishments undertake additional operations, such as the manufacture of components which had previously been purchased from suppliers, man-hours will increase but there will be no corresponding increase in final output. Thus, output per man-hour indexes would be biased. In developing industry indexes, the BLS ex amines data such as the ratio of cost of materi als to value of shipments for any indication of a change in the degree of vertical integration. Presentation BLS indexes are published annually in the form of a bulletin, Indexes of Out-put Per ManHour, Selected Industries. As new industry indexes are developed, they are presented as articles in the Monthly Labor Review. The articles contain an analysis of productivity, output and employment trends in the industry. Technical notes describing the methodology used to develop the indexes are available on request. Indexes of output per man-hour also are published in the Statistical Abstract of the United States and in the Handbook of Labor Statistics. Some indexes for earlier years are published in Historical Statistics of the United States. Uses and Limitations Industry measures of output per man-hour are particularly useful for studying changes in manpower utilization, projecting future man power requirements, analyzing trends in labor costs, comparing productivity progress among countries, examining the effects of technolog ical improvements on employment and unem ployment, and analyzing related economic and industrial activities. Such analysis usually re quires that indexes of output per man-hour be used in conjunction with other industry data. For example, to study technological effects, related data on production and employment are useful; to study trends in labor costs, data on earnings and other labor expenditures are necessary. Although the measures relate output to one input— labor time— they do not measure the specific contribution of labor, capital, or any other factor of production. Rather, they reflect OUTPUT PER M A N -H O U R the joint effect of a number of interrelated in fluences such as changes in technology, capital investment per worker, utilization of capacity, layout and flow of material, skill and effort of the work force, managerial skill, and labormanagement relations. Also, indexes which re late output to one' group of employees represent the total output of the industry resulting from all employees and do not represent the specific contribution of that group of employees. These industry measures of output per man hour are subject to certain qualifications. First, existing techniques cannot fully take into ac count changes in the quality of goods and services produced. Second, although efforts have been made to maintain consistency of coverage between the output and labor input estimates, some statistical differences may re main. Third, changes in the degree of plant integration and specialization often are not reflected adequately in the production statis M EASURES: IN D U S T R IE S tics. This may result in overstatement of pro ductivity gains in some years, understatement in others. Fourth, indexes involving nonpro duction worker man-hours are subject to a wider margin of error than are the indexes using production worker man-hours because of the technique for estimating average man hours of nonproduction workers. Errors in estimating man-hours for nonproductibn work ers, however, have a relatively insignificant effect on the estimates of man-hours for all employees. Fifth, year-to-year changes in out put per man-hour are irregular, and, therefore, not necessarily indicative of basic changes in long-term trends. Conversely, long-term trends are not necessarily applicable to any one year or to any period in the future. Because of these and other statistical limitations, these indexes cannot be considered precise measures; instead they should be interpreted as general indicators of movements of output per man-hour. Technical References Num ber 1. Dunlop, John T. and Diatchenko, V. L a b o r P r o d u c tiv ity (New York, San Francisco, Toronto, London: McGraw-Hill Book Company, 1964). An international collection of papers presented by participants of the conference on labor productivity held at Cadenabbia, Lake Como, Italy, in 1961. Papers cover concepts and measurements of productivity; international comparisons of productivity; wages and productivity; and technical, managerial, and organizational factors affecting productivity. 2. European Productivity Agency, Organization for European Economic Co-operation, P ro d u c t iv i t y M e a s u r e m e n t, Volume I (Concepts) Prepared by G. Deurinck. Paris, August 1955, Project No. 235. The following listed essays are included: The Concept of Productivity and Its Corol laries ; Alternative Productivity Concepts; Aspects of Productivity Measurement and Mean ing; The Role of Official Statistics in Measuring Productivity; Productivity, Efficiency and Wages; and Indices of Industrial Efficiency. 3. Evans, Duane and Siegel, Irving H. “The Meaning of Productivity Indexes,” Jou rn a l o f the A m e ric a n S ta tistic a l A sso c ia tio n , March 1942, pp. 103-111. The nature, use, and limitations of productivity indexes. 4. Fabricant, Solomon. “Of Productivity Statistics: An Admonition,” R e v ie w o f E c o n o m ic s and S ta tistic s, November 1949, pp. 309-311. The deficiencies and limitations in productivity measurement. 5. Greenberg, Leon. D a ta f o r M e a s u r e m e n t o f In d u stria l P r o d u c tiv ity in the U n ited S ta te s, U.S. Department of Labor, Bureau of Labor Statistics, 1965. A review of the status of available statistical data and determination of what gaps exist in order that improvements can be made. 6. International Labour O ffice. M e a su r in g L a b o r P r o d u c tiv ity , Geneva., 1969. Discusses meth odology and problems in the measurement of productivity, analysis of national series and international comparisons of productivity measures. 7. Kendrick, John W., P o s t w a r P r o d u c tiv ity T r e n d s in the U n ited S ta te s, 1 9 ^ 8 -6 9 , National Bureau of Economic Research, forthcoming. 225 226 BLS HANDBOOK OF M ETH ODS Technical References— Continued 8. 9. 10. 11. 12. 13. 14. 15. Presents trends in productivity by industry groupings from 1948-66 with preliminary estimates through 1969. Long term trends, patterns of productivity growth and inter relationships among variables are analyzed. Also included are descriptions of concepts, methods and sources. Klotz, Benjamin P., P r o d u c tiv ity A n a ly s is in M a n u fa c tu r in g P la n ts, U.S. Department of Labor, Bureau of Labor Statistics, 1970. An econometric analysis of industry production functions derived from data for over 1700 plants. Klotz, Benjamin P. and Herman, Shelby A., P r o d u c tiv ity in the R a ilroa d h id u s tr y , U .S . De partment of Labor, Bureau of Labor Statistics, March 1970. An analysis of productivity in the railroad industry using econometric techniques. Mark, Jerome A. “Industry Indexes of Output Per Man-Hour,” M o n th ly L a b o r R e v ie w , No vember 1962, pp.1269-1273. A description of the methodology used in the construction of Bureau of Labor Statis tics indexes of output per man-hour. Covers methods and sources, construction of produc tion and man-hour indexes, and limitations. National Bureau of Economic Research, Conference on Research in Income and Wealth. O u t p u t, In p u t, and P ro d u c tiv ity M e a s u r e m e n t, Volume 25 of S tu d ies in In co m e and W e a lth , Princeton, N.J., 1961. A collection of papers and comments devoted to an appraisal of the measurement of output, input, and productivity. Siegel, Irving H. C o n cep ts and M e a s u r e m e n t o f P ro d u ction and P r o d u c tiv ity , Working Paper of the National Conference on Productivity, 1952. The rationale and techniques of measurement of changes in the physical volume of pro duction and the level of productivity. Includes an extensive bibliography on production and productivity measurement. U.S. Department of Labor, Bureau of Labor Statistics. In d e x e s o f O u tp u t p e r M a n -H o u r , M o t o r V eh ic les and E q u ip m e n t in d u str y, 1957-66, December 1968. A study of productivity trends in the motor vehicles and equipment industry. Contains a detailed technical note describing the methodology used to derive the measures. United Nations. In d e x N u m b e r s o f In d u s tria l P ro d u ction , Series F, No. 1 of Studies in Methods, New York: 1961. Deals with the more technical aspects of compiling production indexes. P r o d u c t i v it y : A B ib lio g ra p h y (BLS Bulletin 1226, 1958) and (BLS Bulletin 1514, 1966). Two collections of annotated references concerning productivity and productivity mea surement. — Charles W. A rdolini, A rthur S. H erman , and John L. Carey Chapter 27. Technological Change Background Studies of technological changes and their labor implications have been undertaken by the BLS over the years for a variety of purposes. During the 1930’s, public interest focused on the unemployed and reports were prepared on technological changes and displacement of workers in various industries. During World War II, emerging technologies were studied for purposes of improving manpower utilization. Beginning in the mid-1950’s, nationwide at tention was focused on the implications of new developments classified under the general term “ automation.” The BLS made a series of stud ies on a plant basis, in the insurance, petroleum refining, bakery, air lines, and electronics in dustries, to explore the manpower implications of various changes. Later, broader studies were undertaken, including a survey of manpower impacts of changeover to electronic computers in 20 large companies and intensive studies of technological change in the coal and paper in dustries. These studies formed the basis, beginning in the early 1960’s, for a more systematic investi gation of future changes. Research now under way pinpoints technologies which will become increasingly important over the next decade in key industries and attempts to provide advance information about their manpower implications. Description o f Studies The Bureau’s research program on tech nological change involves a variety of reports and studies of different degrees of detail and approach. The current program thus provides detailed case studies of changes within a single plant or office; summary reports surveying trends in major industries; detailed industry studies; and studies of major technological innovations, such as computers, that affect workers in different industries.1 1 For discussion of various research methods used in study ing automation, see technical references 1 and 2. Case Studies BLS case studies provide detailed informa tion on various aspects of adjustments made to technological change. The framework of such studies is a single plant or office. A typical case study covers such topics a s: management’s objectives and results of introducing technolog ical change at the plant or office; exent of displacement and reassignment of employees; practices regarding transferring, retraining, and selecting employees for new occupations; characteristics of employees whose jobs were eliminated and who were assigned to new posi tions ; and implications of automation for older employees. Both qualitative and quantitative data are presented. The case study approach has also been used to investigate in detail special aspects of ad justments to technological change. Among the topics covered by case studies are: the per formance of older workers in industry retrain ing programs, as shown in four plants in different industries; manpower planning pro cedures followed in connection with a series of technological changes at a public utility and in telephone offices; the adjustments to elec tronic data processing in a government agency; experience under an adjustment program in the railroad industry; post-layoff work ex perience of displaced workers in plants in dif ferent industries and regions; and job redesign for older workers at different plants. In these studies, information about single plants is presented, as illustrative, rather than as rep resentative cases. Reviews of Technological Industries Trends in Major To provide a broad overview of significant trends in the economy, the Bureau prepares a summary report, applying to key industries on new types of machinery, processes and prod ucts which are believed likely to have an im portant effect over the next 5 to 10 years. The industries covered comprise a cross section of the economy and include those where the pace may be slow as well as those where change is 227 228 BLS H ANDBOOK OF M ETHODS rapid. The first version of this report, entitled Technological Trends in 36 Major American Industries, was issued by the President’s Advisory Committee on Labor-Management Policy in 1964. A revised edition covering 40 industries was published in 1966.2 The emphasis of the report is on technolog ical developments within each industry in an early stage of the innovation’s commercial use— i.e., the period after introduction on the market but before widespread adoption. In ventions and discoveries still in the “ drawing board” stage are considered unlikely to have as much impact over the next decade as those already tested and are generally not discussed. The report briefly describes recent technolog ical developments, indicating insofar as prac ticable some economic advantages of various types of new equipment, processes or products; their importance in terms of the man-hours engaged in the operations affected; estimated extent of use currently and in 5 to 10 years, and some factors affecting adoption such as the volume of investment and expenditures for research and development. The advantages de scribed include not only labor savings per unit, but also quality improvements, fuel and material economies, greater accuracy, new markets, etc. In assessing the employment implications of technological changes, account is taken of the possible rate of growth in output per man-hour and in the industry’s total output. Appraisal also is made of the changes in occupational structure and of some issues and examples of adjustment that are taking place. Outlook Reports Trends on Industry Technological Intensive studies are made of selected major industries where far reaching changes, on a large scale, are taking place, such as coal, rail roads, and textiles. These studies involve de tailed analysis of the economic implications of major technological developments within in dividual industries. Factors analyzed include investment trends and factors affecting the * See technical reference 12. 3 See technical references 9, 10, and 12. prospects for the diffusion of recent technolog ical advances, such as the structure of the industry. Estimates are developed of the dis placement of present by new methods over the next 10 years. Unit labor requirements under new and old technologies are compared, wherever possible. Since the focus of the study is on the industry as a whole, data on recent industry trends in output per man-hour, pro duction, and employment are examined in re lation to long-term trends and projections of future trends are developed. Technological Innovation Studies Some technical innovations have applicability in many industries. Among these are such de velopments as computers, numerical control of machine tools, materials handling equipment, and control instruments.3 Because of their far reaching impact, special studies have been made of the nature, status, prospects for adop tion, and implications for unit labor require ments, occupational change training needs, and problems of industrial relations. In analyzing their impact in different industries, differences as well as similarities are revealed. Data Sources and Collection Methods A variety of data sources and collection methods are utilized in making studies of tech nological change and its impact. Personal Interviews In making case studies, analysts personally conduct intensive interviews with plant man agers, personnel directors, and other officials who have direct knowledge of changes at their plant. Union officials at the plant, and in some cases, individual workers are interviewed. The analyst uses a checklist of questions in con ducting informal interviews in order to elicit the maximum amount of data. Plants and offices included in these studies are selected on the basis of having recently made a major change in their equipment, products, or meth ods of production. T E C H N O L O G IC A L CHANGE 229 Personal interviews also are utilized to help determine industry trends. Informal inter views are conducted with engineers, scientists, economists, and other experts in companies which produce and use new technology, unions, trade associations, government agencies, uni versities, etc., who have specialized knowledge of particular technological development of in dustry trends. One objective in these cases is to obtain their expert judgment about the nature, pace of introduction, and possible im pact of developments with which few plants have had any experience. The emphasis in these interviews is on the technological change rather than on experiences in adjusting. inventory of metalworking machinery; and Bureau of Mines: annual reports on mining equipment. Statistical information on industrywide trends are useful in analyzing the economic implications of technological change. Among the important sources used in preparing stud ies are the Bureau’s indexes of output per man-hour and related series on production, employment and hours; the Bureau of Census data on expenditures on plant and equipment; and the National Science Foundation’s esti mates of research and development. Trade and Technical Publications Because of the complexity of the subject, relatively little use has been made of mail surveys in studying the impact of technological change. This technique, however, is useful in obtaining information on a broad scale not otherwise available, to supplement detailed in formation collected through case studies. Thus, to determine the extent and manpower impact of computer use in the insurance industry, the Bureau conducted a two-part survey of 400 companies employing most of the employees in the industry.4 In the first stage, a brief, 1-page questionnaire covering data on employ ment and extent of computer use was sent to a group consisting of all large offices, and a sam ple of small companies. From information re ceived from this screening questionnaire a number of companies were selected to whom a more detailed questionnaire was sent, asking for information about computer uses, number of employees engaged in electronic data proc essing, planned applications, etc. The mail survey technique also was used in followup surveys of workers who have been laid off as a result of technological and eco nomic change.5 The names and addresses of such workers were obtained from plant person nel records, selecting only a sample in cases where a large number of workers had been displaced. The questionnaire was mailed at least 6 months after the layoff occurred, in order to allow a period of time for some adjust- Important sources of information concerning technical trends are trade journals, technical magazines and books, conference proceedings, government hearings, and company reports. Annual reports of leading corporations and company house organs often contain useful information on current technical development in some industries. In making studies of indus tries, these publications are reviewed to obtain information about the status and prospects of important developments and to ascertain which companies and plants merit more intensive field visiting. Reports and publications of firms that produce particular types of equipment often are found useful in studies of industries that use such equipment. Statistical Data Sources Quantitative information about the status of specific technological developments is fragmen tary and scarce. The Bureau makes use of available data from many public and private sources. These sources include, for example: Office of Management and Budget; annual inventory of computers in the Federal Gov ernment; Business and Defense Services Ad ministration : annual survey of numerically controlled machine tools; McGraw-Hill Com pany : survey of computers in industry; Ameri can Bankers’ Association: survey of banking automation; American Machinist Magazine: Mail Surveys 4 See technical reference 5 r See technical reference 4. < 230 BLS HAN DBOOK ment to take place. Information collected from the mail survey dealt with the personal char acteristics of displaced workers such as age, sex, occupation, level of education and skill, post-layoff work experience, such as labor force status at time of survey, type of job held after layoff, source of jobs, etc. Plant Records In making detailed studies of the impact of technological change on individual workers within a plant, analysts sometimes can obtain from employers’ files, data on such aspects as the age, sex, and related personal charac teristics of employees whose jobs are elimi nated and the jobs in the plant held by each individual affected before and after the change; similar data are collected on individuals who are selected for the positions created in con nection with automated equipment.6 Expert Review In preparing forecasts of future technolog ical trends, a critical step is the review of preliminary reports with outstanding experts in each industry.7 Drafts of industry reports are mailed to company executives, union re search directors, trade association officials, technical journal editors, and university and government specialists for their assessment of the validity and adequacy of projected trends. Over 450 persons were contacted in this way in the preparation of a report on technological trends in major industries. Some experts are visited personally to review draft statements in detail. Through this means, reports on tech nological prospects are designed to reflect, as much as possible, the authoritative views of a number of persons who have expert, first hand knowledge of each industry. Analysis and Interpretation For a better understanding of research re sults in this field, it is important to keep in mind the meaning of certain key ideas and OF M ETH ODS concepts. Some of the key problems of interpre tation and analysis in this type of research are therefore set forth, briefly. Definition of Technological Change Technological change is defined broadly in the BLS studies, as encompassing significant changes in processes and equipment, products and services produced, and materials, fuels, and energy used. The term “ automation,” which is sometimes popularly used as a synonym for “ technological change,” designates, strictly speaking, a particular type of current de velopment. It has been variously defined, for example, as “ automatic operation,” “ the mech anization of sensory, control and thought processes,” and “ a concern with production processes as a system.” While BLS studies have been concerned with developments in “ automation,” particularly in anticipating long-term trends in the future, they are not the only technological changes taking place that affect labor requirements and industrial relations. For example, new ways of generating power, piggybacking in transporta tion, use of synthetic materials in manufactur ing, mechanized methods of materials handling, and faster steelmaking processes are impor tant technological developments, not usually covered by technical definitions of “ automa tion,” but having significant manpower im plications. Impact on Productivity Since one of the principal consequences of technological change, so far as manpower utilization is concerned, is an increase in pro ductivity— i.e., output per man-hour, special attention is given in BLS studies to analyzing changes in industrial productivity. Such trend analysis is a useful method of measuring the pace of technological change. Changes in pro ductivity, however, also reflect changes in capacity utilization and many other nontech nical factors; it is important to recognize that BSee technical references 1 and 6 for use of plant records. 7 See technical reference 12 for further detail. T E C H N O L O G IC A L the productivity trend is only a partial meas ure of the rate of technological change. In determining the impact of a specific technology, BLS studies try to indicate the reduction in unit labor requirements that the new processes are designed to achieve. In some cases, estimates of labor savings are derived on the basis of comparisons with the estimated average technology of the industry under study; in others, with the best equipment that is available; or in actual plant studies, with the technology that is actually displaced. It is also important to distinguish between the impact on productivity of the operation directly affected and on productivity of the plant as a whole. An advanced machine tool, for example, may result in a relatively large reduction in unit labor requirements in the machining operation, but would have little im pact on finishing, and assembling, and may even require additional labor in engineering and maintenance work.8 The impact on plant productivity, therefore, would be considerably less than the effect on productivity of any de partment or operation directly affected. Impact on Employment In assessing the impact of technological change on employment, it is necessary to con sider the implications of plant manpower policies and the effects of economic changes, with which technical changes interact. Analysis of the impact of technological change purely in terms of machinery is incomplete. At the plant level, for example, the substitu tion of machinery for labor may substantially reduce job opportunities in operations directly affected. If efforts are made, however, to elimi nate these jobs by not filling vacancies created by quits, deaths, and retirement of employees, or by transfer of affected workers to other positions in the plant or office, labor savings could be achieved without displacing the work ers affected from the plant. Moreover, the employment impact of tech nical change is also interrelated with the effects of the business cycle. Thus, workers whose jobs are eliminated by technical changes may not be displaced from a plant until a decline in CHANGE 231 demand results in layoffs— a long time after the change has been made in some cases. In the subsequent recovery, however, they may not be hired back because their jobs no longer exist. Since many changes exert their effects on employment through the competitive market, the employment trend for the industry as a whole must also be examined. The plant which reduces its unit costs through technological improvement may be able to gain a larger share of the market and increase its employment, but at the expense of the less technically advanced competing plants, which may be forced to shut down, displacing workers far from the location of the change. Because of the whole complex of economic factors that operate through the market, in cluding changes in demand, location, foreign competition, merger, and consumer taste, it is very difficult’ to isolate the expanding and dis placing effects of technological change. Impact on Occupations Two aspects of occupational change result ing from technological changes are examined. Changes in job structure— the distribution of the plant or office work force by function or broad skill grouping— are studied to determine the extent of upgrading or downgrading. Since the content of jobs may be altered as a result of changes in equipment or processes, attention also is directed to intensive before-and-after analysis of job duties and the knowledge and abilities required to perform these duties as indicated by job descriptions and observation. The content of newly created jobs, such as programmer, also is studied and the qualifica tions required and personal characteristics of individuals selected for these new positions are described, so far as possible.9 Adjustments to Technological Change Technological change has important implica tions for personnel management and collective bargaining within plants. The introduction of 8 See technical reference 10 for further discussion. " See technical references 1, 6, and 10. 232 BLS HAN DBOOK new machinery, products, or processes often requires movement of workers among jobs within the plant or office by transfer or promo tion, the setting of wage rates, and selection of persons for new jobs. Often the adjustment proceeds according to rules established in ad vance through collective bargaining. Provisions to assist workers whose Jobs are eliminated include severance pay, retraining, and early retirement. Besides analyzing the operation of formal provisions under collective bargaining, Bureau studies describe informal efforts to pro vide training, to utilize attrition, and to obtain jobs for displaced workers elsewhere.1 The 0 limitations of these measures as well as their advantages are important matters studied. Uses and Limitations BLS studies of technological change are pre pared as part of the U.S. Department of La bor’s program for carrying out the objectives and responsibilities of the Manpower Develop ment and Training Act. Under this act, the Secretary of Labor is required “ to establish techniques and methods for detecting in ad vance the potential manpower impact of auto mation, technological progress, and other changes in the structure of production.” As part of such an early warning system, BLS studies and reports of technological change are useful to managers, union leaders, educators, economists, government officials, and others in planning policies to cushion the impact of 1 See technical references 1 and 6 for further discussion. 0 OF M ETH ODS change. The study of emerging technological trends and possible implications, moreover, provides a basis for more valid projections of productivity and economic growth. They also are useful in pinpointing manpower problems and determining the most productive direction of future research to obtain possible solutions. Some limitations of the Bureau’s studies of technological change must be kept in mind in assessing their appropriateness for particular uses. In general, it is important to recognize that judgments about the future direction and pace of technological change and its implica tions are necessarily complex and difficult. The rate of introduction of new technology depends not only on technical advantages but also on many economic factors, such as the volume of investment, market prospects, and the avail ability of trained workers, all of which are sub ject to significant variations. Moreover, since the period of introduction generally spans a number of years, new developments are con stantly appearing so that assessments of the outlook must be reappraised from time to time in the light of new information. Finally, studies of the impact of technolog ical change deal primarily with changes within individual industries. But these changes often involve changes in the type and amount of goods and services purchased from other in dustries and could therefore have important implications for production and employment in industries supplying inputs. The accumulation of information on interindustry relationships, through the Bureau’s economic growth studies, will provide a quantitative basis for analyzing this aspect of technological change. T E C H N O L O G IC A L 233 CHANGE Technical References Number 1. A d ju s tm e n t to the In trod u ction o f Office A u to m a tio n , 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. (BLS Bulletin 1276, 1960). A study of manpower implications of the installation of electronic data processing in 20 offices in private industry, with special reference to older workers. Discusses problems of measuring impact of electronic data processing on employment and occupational re quirements. A u t o m a t io n : A D isc u ssion o f R esea rch M e th o d s, Labour and Automation Bulletin No. 1, Geneva, International Labour Office, 1964. Fourteen papers on problems of research methods in studying manpower impact of automation at the plant and industry level. A u to m a tio n and T ech n ologica l C h a n ge (edited by John T. Dunlop). The American Assem bly, Columbia University, Englewood, N.J., Prentice-Hall, 1962. Case studies of psychological impact are summarized in Chapter 3, by Floyd C. Mann. Chapter 7, “Employment,” by Ewan Clague and Leon Greenberg, discusses problems of measuring employment impact. C a se S tu d ies o f D isp la ced W o r k e r s , BLS Bulletin 1408 (1964). Studies of post-layoff experiences of nearly 3,000 workers formerly employed at five different plants. Discusses use of mail questionnaire in followup of displaced workers. Im p a c t o f Office A u to m a tio n in the In su ra n ce I n d u s tr y , (BLS Bulletin 1468, 1966). Mail survey of over 400 insurance companies. Covers extent, pace, and employment im plications of electronic data processing. Im p a c t o f Office A u to m a tio n in the In te r n a l R e v e n u e S e rv ic e, (BLS Bulletin 1364, 1963). Detailed case study of large-scale changeover to electronic data processing. Illustrates use of internal personnel records in analyzing effects of office automation. In d u stria l R etra in in g P ro g r a m s f o r T ech n ologica l C h a n g e, BLS Bulletin 1368 (1963). Four case studies of experience of older and younger workers in industrial retraining programs, based on plant records. Discusses some problems of measuring comparative per formance. M a n p o w e r P la n n in g f o r T ech n ologica l C h a n g e — C a se S tu d ies o f T elep h on e O p era to rs, (BLS Bulletin 1574, 1968). Case studies of the manpower policies and experiences of several telephone companies in cushioning the impact of technological change on their employees. O u tlook f o r C o m p u te r P ro c e ss C on trol, (BLS Bulletin 1658, 1970). A study of manpower implications of the introduction of computers to control produc tion processes in six process industries. O u tlook f o r N u m erica l C o n tro l o f M a ch in e T o o ls , (BLS Bulletin 1437, 1965). A study of manpower implications of a key technical innovation affecting metalworking industries. Discusses problem of generalizing about change in productivity as result of specific change. T ech n o lo g y and M a n p o w e r in the T e x tile In d u s tr y o f the 1 9 7 0 ’s, (BLS Bulletin 1578, 1968). A study of the changes in technology in this major industry, and the impact of pro ductivity, employment, and occupational requirements. T ech n ologica l T ren d s in M a jo r A m e ric a n In d u stries, (BLS Bulletin 1474, 1966). Description of outlook for major technological developments based on a variety of data sources and expert review. — E dgar Weinberg Chapter 28. Construction Labor Requirements Background New construction is a major component of the Nation’s output of goods and services and a major source of employment. The jobs it creates occur not only at the site of employ ment, but also in the many manufacturing, trade, transportation, and service industries which furnish the materials and services re quired in the construction process. Because of this far-reaching employment impact, the crea tion of new construction projects often is regarded as a means of counteracting cyclical unemployment. To assist in assessing the extent of the impact of construction expenditures on employment, a series of labor and material requirements stud ies for different types of construction was started in 1959. The program was established as a result of Congressional legislation, requir ing the Bureau of Labor Statistics to provide estimates of the amount of total employment generated directly and indirectly by various kinds of construction per dollar of expenditure. Earlier Bureau efforts to relate employment and volume of construction included the “ Labor Required for New Construction” series, with reference mainly to on-site man-hours. This series, started in the early thirties, appeared intermittently through the years, but was not based on actual up-to-date surveys and was finally halted in the mid-fifties because the factors employed were found to be obsolete. There was also a series of Public Works Ad ministration studies published in 1940, cover ing Federal public works projects constructed in the mid-thirties; and a few individual stud ies of specific types of construction. The present studies include the major types of building construction (schools, hospitals, public and private housing, etc.) and also heavy construction (highways, dams, etc.). However, only one or two selected types of construction are surveyed in any given year. Selected types of construction are resurveyed periodically. These resurveys may, in addition to providing current information on labor requirements, contribute information useful in preparing con struction cost indexes and estimates of changes in productivity of on-site construction labor. Description o f Survey The surveys are designed primarily to deter mine the number of man-hours represented by a fixed dollar volume ($1,000) of contract con struction. Man-hours, as defined by the surveys, include both on-site construction employment and the off-site employment required to pro duce and deliver materials used in the con struction. Data for on-site labor include total man-hours for the supervisory, engineering, clerical, and custodial employees, as well as those for workers in each construction trade at the site of construction. Data for off-site labor include employment in the off-site (e.g., office and warehousing) activities of construc tion contractors; in building materials and equipment manufacture and distribution; and finally, employment in all the other industries which are affected directly or indirectly by the production and distribution of building ma terials from raw materials to the final manu factured product. Man-hours are also estimated for the employment created by overhead ex penditures of contractors. Certain types of employment are not covered by the survey. For example, no estimate is made of the employment used in the planning and designing of the projects studied. Also excluded are estimates of employment required in government and public utility agencies which might be affected by the construction being studied. Employment created by the re spending of wages and profits of the workers and their employers— the multiplier effect— are not considered within the scope of the studies. Data Sources and Collection Methods The surveys cover construction which is nonfederally assisted and/or that which may be totally or partially financed through Federal 235 236 BLS H AN D BO O K OF M ETH ODS funds. Although the type of construction labor and material requirements data sought are similar for both federally and nonfederally aided projects, the sources for the data are different, particularly for the on-site man-hour information. For the construction of nonfederally aided projects— those financed entirely by various levels of local government, private individuals or voluntary groups— on-site man-hour data, by occupation, are obtained by field representa tives from local authorities, architects, con tractors, and other direct participants in the projects. These individuals provide the desired man-hour information from payroll records, daily work force reports, or summary time reports. For those projects financed wholly or in part by Federal funds, on-site employment informa tion generally is obtained from contractors’ payrolls submitted to the government under regulations of the Davis-Bacon Act or other Federal legislation covering federally aided construction. These payrolls furnish the data for estimat ing on-site man-hour requirements, as well as data on wages for all hourly rated workers on the projects. Data for on-site salaried employ ees, not accounted for on the payrolls, are ob tained by the field agents from the contractors. Access to these payrolls also makes possible the collection and presentation of additional detail for the projects studied. This includes information on wage relationships, timing of construction operations, and requirements by type of contractor. Information on material and equipment costs for the projects studied— i.e., the items which are used for estimating off-site employment— is collected by the field representatives from the individual contractors and subcontractors engaged in the construction of the sample projects. Sampling and Estimating Procedures Sampling procedures vary with the type of construction being studied. The “ universe” of projects for a specified study generally repre sents all of the projects known to have been completed during a selected 1-year period. The projects are then stratified into cells having similar characateristics which may affect man-hour requirements. Factors con sidered important in cell stratification include (a) regional location, (b) metropolitan or non metropolitan locality, (c) types of structure (when known), (d) purpose of structure (if variable)1 and (e) amount of total construction contract (i.e., size of project). One or more projects are selected at random from each cell and assigned weights to give total representation of the cell. Substitutions, when necessary, are made from the same cells. Data from the sample projects are combined in aggregate form to represent the total dollar volume of each cell. As indicated previously, the off-site employ ment estimates are derived from the materials and equipment cost information obtained from the contractors and subcontractors cooperating in the studies. The contractors provide a list of the value of each type of material used in the construction of sample projects. These material listings are classified into categories consistent with a 4-digit Standard Industrial Classifica tion code as used in Census of Manufactures product groups. For each of these product groups, average amounts of material (in dol lars) required for each $1,000 of contract construction cost are determined. The value of materials is reduced by a ratio representing the difference between valuation by the pur chaser and valuation by the producer. (This ratio is based on valuation data provided by the U.S. Department of Commerce, Office of Business Economics.) This latter step is re quired because all data reported by contractors are in purchasers’ value, and reduction to pro ducers’ value is necessary to obtain figures consistent with Census data which are used to calculate manufacturing employment. Primary employment in manufacturing is considered to be that required to produce the construction bill of materials in their final stage of fabrication. In this stage, man-hours are developed by multiplying average pro 1 For example, elementary or secondary schools. C O N ST R U C T IO N LABOR ducers’ value of each construction material by the ratio of manufacturing man-hours per $1,000 of production. (This ratio is established by using the Census of Manufactures.) Primary man-hours in the trade, transporta tion, and service industries are estimated from the difference between producers’ and pur chasers’ value for each construction material. The value differences are alloted to trade and transportation, and primary man-hours for each component of transportation and trade are then estimated from labor factors provided by BLS. Secondary employment is defined as the em ployment in all industries involved in produc tion and transportation of building materials and equipment from basic extraction to, but not including, the final manufacturing stage. The Interagency Input-output Study is used to obtain these estimates.2 For each group of ma terials, the interindustry study provides infor mation on the amount of secondary products required from each of its 78 industry sectors. The product data are converted to man-hours by use of output per man-hour ratios for each of the sectors. Adjustments for price and pro ductivity are made to provide estimates con sistent with the year of construction and bill of materials. For each off-site stage (primary and secon dary), a man-hour figure per $1,000 for the construction being studied is obtained. When these man-hours, plus the builders’ off-site em ployment, are combined with the direct or on site man-hours, the total employment effect, within the definition used by the studies, is determined. An exact study of the off-site employment of each construction contractor is not attempted, since it is almost impossible to relate accurately such employment to the projects studied. Build ers’ off-site employment is occupied not only with the sample projects studied, but also with other current or future projects. The estimate of contractors’ off-site man-hours for each $1,000 of construction contract is based on the difference between construction worker em ployment and total employment in the COn ^ e e "Input-output Structure of the U.S. Economy: 1963,” Survey of Current Business, November 1969, pp. 16-47. R E Q U IR E M E N T S 237 struction industry, with adjustments for on-site supervisory and administrative employ ment. Analysis and Presentation The construction labor and materials re quirements studies are published in Bureau bulletins. Summary articles, based on the sur vey findings, appear in the Monthly Labor Review. The bulletins and articles highlight the total man-hours generated per dollar volume of construction expenditures, with subtotals for on-site and off-site man-hours. In presenting the labor and materials data, the statistical tabulations are supplemented with an analysis of the various factors which apparently affected the man-hour requirements for the specific types of construction studied. The bul letins contain information on the characteris tics of the sample projects and the man-hours per square foot and per $1,000 of contract cost, with an analysis of the variations in re quirements and costs arising from differences in design, type of structure, and regional or metropolitan versus nonmetropolitan in fluences. The reports include information and analysis o f: on-site man-hour requirements by occupa tion ; the employment share of the general and special trades contractors engaged in the con struction work; direct on-site wage cost; the distribution of employment by periods of con struction tim e; and the cost of major materials. When feasible, comparisons of unit man-hour and material requirements for earlier periods are made and analyzed. Uses and Limitations The results of the labor requirements sur veys are used by other offices of the Bureau, other Bureaus of the Department of Labor, other governmental agencies, congressional committees, and industry research and trade organizations, to assist in gauging the impact of planned expenditures for construction on employment, and the economy, generally. Of special interest to market research analysts 238 BLS H ANDBOOK OF M ETHODS and companies manufacturing materials for use in construction, are the materials listings per $1,000 of construction contract. While the overall estimates of employment are believed to be reasonably accurate, the de tailed data would have a wider margin of sampling error and may be subject to other limitations. Man-hour and material require ments are affected by a number of factors such as location, size of project, type of structure, architectural design, availability of certain ma terials or equipment, labor skills, and local building codes and customs. The effects of these separate factors cannot be isolated. Technical References Labor and Material Requirements Studies Number 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. (BLS Bulletin 1390, 1964). A study of on-site and off-site man-hour and wage requirements for dredging and landtype projects in the U.S. Corps of Engineers’ civil works program from 1959 to 1960. C o lleg e H o u sin g C o n stru ctio n (BLS Bulletin 1441, May 1965). A report based on findings in a survey of 43 college housing projects which were ad ministered by the Community Facilities Administration. The survey is designed primarily to determine the man-hours required for $1,000 of college housing construction. C o n stru c tio n o f F e d er a lly A id e d H ig h w a y s , 1958, 1961, and 1964 (BLS Bulletin No. 299). This study provides measures for 1958, 1961, and 1964 of the labor and material re quirements for federally aided highways, and separate measures of the requirements for on site and off-site construction. For on-site construction, the study also provides a comparison of annual man-hour requirements for 1947-64. F e d e r a l Office B u ild in g C o n stru c tio n (BLS Bulletin 1331, 1962). A statistical study of on-site and off-site labor requirements for constructing 22 Fed eral office building projects in various localities of the United States over a 3-year period from the fall of 1957 to 1960. H o s p ita l C o n stru c tio n (BLS Bulletin 1340, 1962). A statistical study of on-site and off-site labor requirements for construction of selected public and private, profit and non-profit, general hospitals in various localities of the United States between mid-1958 and mid-1959. H o s p ita l and N u r sin g H o m e C o n stru c tio n (BLS Bulletin 1691, 1971). A study similar to the 1962 study but with data shown per square foot as well as per $1,000 of construction contract. P r iv a te O n e -F a m ily H o u sin g C o n stru ctio n (BLS Bulletin 1404, 1964). A statistical study of on-site and off-site labor requirements for constructing a sample of one-family houses built in 1962 in various localities of the United States. P u blic H o u sin g C o n stru ctio n (BLS Bulletin 1402, May 1964). A report based on findings of a survey of 31 public housing projects which were administered by the Public Housing Administration. Projects were selected in various States to be representative of four broad geographical regions of the conterminous United States. S ch ool C o n stru c tio n (BLS Bulletin 1299, 1961). A study of primary and secondary man-hours required per $1,000 of new school con struction based on contracts awarded throughout the United States for 85 elementary and 43 junior and senior high schools. S ch ool C o n stru c tio n (BLS Bulletin 1586, 1968). A survey of selected elementary and secondary public schools constructed primarily dur ing the period of 1964-65. In addition to providing information on man-hours, the study also includes data on the types and value of materials used, wages paid, occupations dis tributed and use of apprentices. S e w e r W o r k s C o n stru ctio n (BLS Bulletin 1490, 1966). This study was designed to measure the total man-hours of labor required for each $1,000 of new sewer facilities construction contract. The basis for this study was 138 con tracts for new sewer works in the years 1962-63. C ivil W o r k s C o n stru c tio n b y th e C o rp s o f E n g in e e r s — Martin Ziegler Occupational Safety and Health Chapter 29. Occupational Safety and Health Statistics Background Passage of the Occupational Safety and Health Act of 1970 (Public Law 91-596) marked a major departure in the collection of work-injury statistics. The act is comprehen sive; it provides that every place of employ ment subject to the act shall be free from recognized hazards which are likely to cause death or serious physical harm. To assist in attaining this goal, the act provides for record keeping and reporting procedures which will identify the seriousness of on-the-job accident and job related illnesses. The Bureau of Labor Statistics has been assigned responsibility for developing and im plementing the statistical program which is required to achieve the objectives of the act. The Bureau has been concerned for many years with standardizing the methods for com piling work-injury statistics. As early as 1911 the Bureau called a formal conference to dis cuss the matter. The work of the conference was continued by the International Association of Industrial Accident Boards and Commissions, resulting in publication of the first standardized procedures in 1920.1 In 1937 the first workinjury standard was published by the American Standards Association, now the American Na tional Standards Institute. The most recent revision is the Standard Method of Recording and Measuring Work Injury Experience (Z16.1), 1967. ~ In December 1969, while Congress was con sidering comprehensive safety and health leg islation, Secretary of Labor J. D. Hodgson (then Under Secretary of Labor) noted in a letter to the American National Standards In stitute that the proposed legislation included a national system for the collection of safety 1 Standardization of Industrial Accident Statistics (BLS Bulletin 276, 1920). 2 Proposed National System for Uniform Recording and Reporting of Occupational Injuries and Illnesses, American National Standards Institute, New York, New York, 1970. and health statistics. He requested that the Institute evaluate whether the standard method contained in Z16.1-1967 was appropriate for the broad universe of employers who would be subject to the proposed legislation. An informal conference called by the Institute concluded that there was cause to believe that the Z16.1 standard was inappropriate for the proposed use and recommended that a study group be formed to consider the matter further and, if necessary, develop a simple method of report ing injuries. This recommendation was based upon a belief that the Z16.1 standard: (1) had grown too complex, through efforts to make it equitable, to form a basis for a mandatory national reporting system, (2) did not ade quately reflect trends in injury experience for employers with good safety records, (3) was not adequate for recording health experience. A study group, formed by the Institute, concluded that a new simple method for record ing and reporting occupational injuries and illnesses was needed. The study group’s pro posal was published by the Institute in Decem ber 1970.1 The recordkeeping regulations which 2 have been issued by the Secretary draw heavily on the proposals of the study group. Records Required to be Kept The Occupational Safety and Health Act of 1970 directs the Secretary of Labor to issue regulations which require all employers sub ject to the act to maintain accurate records of work related deaths, injuries, and illnesses. However, records do not have to be maintained for minor injuries requiring only first aid treatment and which do not involve medical treatment, loss of consciousness, restriction of work or motion, or transfer to another job. The regulations concerned with recording of occupational injuries and illnesses (29 CFR Part 1904) became effective on July 1, 1971. They require employers to maintain in each establishment a log of occupational injuries 239 240 BLS H ANDBOOK OF M ETHODS and illness. Within 6 working days of receiv ing information that a recordable case has oc curred, the employer must enter the facts called for in the log. At the end of each year the employer is required to complete an an nual summary of occupational injuries and ill nesses. Information for the summary is drawn from the log. The summary is to be prepared no later than one month after the close of the year and is to be posted prominently in each establishment in a place accessible to the em ployees. The log and summary must be kept available in the establishment for a period of 5 years following the end of the year to which they refer. In addition, a supplementary record must be maintained for recordable injury or illness. A form is provided for that purpose (OSHA Form 101) but in most cases a workmen’s com pensation, insurance, or internal form which must be completed for other purposes will be an acceptable record in place of form 101. All logs, summaries, and supplementary rec ords must be available in the establishment for inspection and copying by compliance officers or statistical agents of the Federal or State Governments. Collection Methods Collection methods are still being formu lated. Tentative plans call for an initial collec tion early in 1972 for the last half of 1971. After that initial period, collection will be on a calendar year basis. Collection will be pri marily by mail questionnaire. The respondent will merely transcribe the injury and illness data from the summary form to the question naire. The questionnaire will also seek informa tion about the numbers of employee-hours worked and a statement of the product of the establishment. Coverage Data will be collected from a representative sample of establishments. The sample for the first report period (July through December 1971) is designed to produce injury data at most of the 2-digit SIC industry levels. The sample for the first full calendar vear (1972) of oneration under the act will be approxi mately 200,000 establishments, providing data at the 4-digit SIC industry level in manu facturing and at the 3-digit level in nonmanu facturing. Stale Participation The act encourages the States to assume responsibility for the administration and en forcement of occupational safety and health laws, and for carrying out an occupational safety and health educational program. The Secretary of Labor is authorized to make grants to the States to assist them in developing and maintaining an effective pro gram of collection, compilation, and analysis of work-injury and illness statistics. A con siderable number of States are expected to participate in these grant programs. Existing Work-Injury Data Work-injury data have been collected by the Bureau for many years. These data, which have been collected through 1970, will not be com parable to data compiled under the Occupa tional Safety and Health Act. Methods used in the collection of these data are described in the previous edition of this Handbook. (See Chap ter 26 of the Handbook of Methods for Surveys and Studies, U.S. Department of Labor, Bureau of Labor Statistics, Bulletin 1458, 1966). Description of Surveys and Methods Programs for the collection, tabulation, analysis, and publication of occupational safety and health statistics are being formulated. De tails of these programs will be published as supplements to the BLS Handbook of Methods probably about mid-year 1973. These supple ments can be obtained after mid-1973 from the Bureau of Labor Statistics, U.S. Depart ment of Labor, Washington, D.C., 20212, or from any of its regional offices listed on the inside back cover of the Handbook. —Lyle R. S chauer Due to budget cuts and personnel ceilings, Labor Developments Abroad, basic country studies, labor digests, and country or area bibliographies will be discontinued by June 30,1972. Economic Trends and Labor Conditions Chapter 30. Foreign Labor Conditions, International Comparisons, and Trade Research Background Almost from its inception, the Bureau of Labor Statistics (BLS) conducted research on labor conditions and developments abroad. The Bureau carries on two research activities of particular interest to users of research reports: analysis of research data on labor in foreign countries and the preparation of international comparisons.1 The Bureau undertook these re search programs because (1) summary and detailed information on labor conditions pub lished by a majority of foreign countries is not available in English (or in any form readily usable by U.S. employers, labor unions, Gov ernment officials, and others); (2) users cannot keep pace with the variety of source materials from nearly 200 independent countries and de pendencies; and (3) often, none but an expert can judge the quality of source materials. Description o f Reports The foreign labor research reports published by the Bureau are in part general and in part statistical. The principal types may be listed as follows: 1. Labor Developments Abroad, a monthly publication which covers important develop ments in labor in foreign countries and pro vides data on living costs abroad, furnished by the Allowances Program of the U.S. Depart ment of State. 1 Other major BLS activities in the international field in clude providing orientation and factual advice to U.S. policy and program officers on labor in foreign countries, and pro viding training or orientation to foreign statisticians and other foreign visitors, especially participants in exchange programs and technical cooperation programs, on U.S. labor statistics methods and on the economic conditions of U.S. workers. 2. Basic country studies, primarily mono graphs on labor law and practice in selected countries. These reports are factual, but not primarily statistical. 3. Brief labor digests on single countries, published either separately or with directories of labor organizations in major geographic areas. 4. Bibliographies, chiefly for a country or an area, often with annotations or brief summaries. 5. International comparisons presented in feature articles and bulletins. These are mainly statistical in nature and contain explanations of the source data and, where possible, adjust ments for differences in definitions and meth ods. So far, principal comparisons published have concerned hourly compensation, output per man-hour, unit labor costs, unemployment, and price trends. 6. Reports on trade and labor developments in selected industries, including the relation ship of imports to domestic employment. The Bureau also provides data in response to individual inquiries from Government of ficials, U.S. businessmen, labor unions, and students. Data Sources and Collection Methods The Bureau receives much material— de scriptive, statistical, and analytical— on labor developments abroad, which it maintains in files classified by country and by subject. The material includes (1) current reports from labor attaches and other officers of the U.S. Foreign Service throughout the world; and (2) foreign data from various sources. The latter include periodicals and other publica tions issued by agencies of foreign countries (labor ministries, bureaus of statistics, and 24 1 242 BLS HAN DBOOK others); international organizations— govern mental, business, and labor— as well as the In ternational Labour Organization; and private agencies like banks and economic research organizations in the United States and abroad. When specific supplementary information is needed, the Bureau often requests it through Foreign Service reporting channels. Staff members keep informed by contact with research institutions, with individual scholars specializing in international labor, and with U.S. firms employing labor abroad. In addi tion, they are briefed by labor attaches and other Foreign Service officers when they visit Washington. Also, from time to time, members of the research staff study foreign labor conditions at first hand by travel abroad. Analysis and Uses With the exception of international com parisons, the major focus in the presentation of reports is geographic, although an occasional summary by topic is used. Within an individual country study, the organization is by subject. For the labor law and practice series, subjects include: (1) social, economic, and political background; (2) manpower (including labor force, employment, unemployment, manpower planning and utilization); (3) labor law and administration; (4) wages, supplementary benefits, hours and working conditions; and (5) labor organizations, management organiza tions, and labor-management relations; (6) industrial safety and workmen’s compensation; and (7) social insurance. Most reports are concerned either with the basic situation at the time the report was written or with a recent development or trend. With a few exceptions, of which real wages is one, less attention is given to historical de velopment of long-term trends. The extent of material covered is so great that often only the most summary information for a given country can be written more often than every 10 years; many countries have never been covered in individual monographs. Reports often contain an annotated bibliography for readers in terested in more detail. OF M ETH ODS International comparison studies were begun in 1960 on a regular basis. They deal with in dividual subjects and are primarily statistical requiring specific information about details of definition, and information is seldom covered fully by regularly published sources. Therefore, great reliance is placed by the Bureau on re quests to the Foreign Service and statistical agencies abroad for information specifically required. Among subjects covered are international comparisons of labor costs and productivity. For instance, articles containing basic infor mation on “ The Role of Labor Cost in Foreign Trade” and on “ International Comparisons of Unit Labor Cost: Concepts and Methods” have been published. Absolute comparisons have been issued also for the iron and steel industry in four countries, as well as trend data in index form for all manufacturing in ten principal trading natio*ns and in the United States. Rela tive labor cost trends are useful for analysis of trade competitiveness, the balance of pay ments, and inflationary developments. Other topics studied have included unemployment, work stoppages, job vacancies, and price trends. The greatest problem involved in comparing unit labor costs among countries arises from the fact that labor cost data are available by industry, but output is reported by product. Since nearly all industries produce a number of products, the question of labor cost alloca tion to specific products (or product synthesis) arises. The trend data published for all manufactur ing are taken, in most cases, from official statis tical reports issued by the individual countries. Because the data are not entirely comparable from country to country, and because more than one set of data exists for most free indus trial countries, considerable effort is spent in analyzing the individual series and selecting those most consistent and valid. In most coun tries where it was available, the Bureau de cided (1) to use as the measure of labor cost the compensation of employees in manufactur ing, from the national accounts, and (2) to use as the measure of total output the real gross F O R E IG N L A B O R C O N D IT IO N S, IN T E R N A T IO N A L C O M P A R IS O N S , A N D in approach. The methods are often intricate, national product originating in manufacturing. The chief reason for this decision has been the greater comparability and comprehensiveness of these measures. As regards unemployment, the interest in international comparison centers on the per centage of the labor force unemployed. Data published by the individual countries on a cur rent basis are of different types. However, most industrial countries of the free world have made one or more sample surveys of their labor forces, using definitions and methods fairly comparable to those in the United States. The Bureau uses these latter data (adjusted as carefully as feasible to uniform definitions and brought up to date by the best data cur TRADE RESEARCH 243 rently available) as the basis for international comparisons of unemployment rates. A program is under development for con ducting studies of the relationship of foreign trade to U.S. labor. Analyses will be prepared for selected industries in order to understand the factors contributing to employment changes and the effects of rising imports on employ ment. Since the relationship of trade to em ployment is usually indirect and difficult to measure, a multiple approach will be taken, including review of plant closings, mass layoffs of employees, and the effects of changes in demand, imports, and productivity on employ ment. — Juliet F. K idney and John Chandler Chapter 31. Economic Growth Studies Background The Bureau of Labor Statistics has de veloped a program of economic growth studies aimed at providing a more comprehensive and integrated framework for analyzing the prob lems of long-run economic growth in relation to employment opportunities. Because this program has important implications for many agencies of the Government, an interagency planning and coordinating committee provides the guidelines for the program. The chairman of the committee is from the Council of Eco nomic Advisers; other members come from the Bureau of Labor Statistics, U.S. Department of Labor; Office of Business Economics, U.S. Department of Commerce; and the Office of Management and Budget. A primary objective of the program is to develop projections, under alternative assump tions, of the rate and pattern of economic growth in the United States. These provide a framework for assessing a number of impor tant economic problems, including problems of manpower utilization over the next decade. Methods An economic growth model has been de veloped to serve as a tool in making economic projections. This model begins by developing a potential economic growth rate for the United States. To do this, it must project the labor force to the target year, assuming a specified unemployment rate, and project the rates of change in productivity and average hours paid. This potential growth in gross national prod uct (GNP) is distributed among the four major components (or demand categories) of GNP: consumer expenditures; domestic investment; government expenditures— Federal, State, and local; and net foreign demand. The next stage is to develop a projection of the industry composition of demand for each of the four major demand categories. For ex ample, the consumer expenditures category includes the amounts spent on rents, automo biles, medical expenses, and other goods and services purchased by consumers. For each of the four major demand categories, a different procedure is followed in allocating demand to the producing industry. The industry detail to which the categories of demand are allocated matches the input-output classification. Demand as used in the national income ac counts refers only to final demand, i.e., that of the ultimate consumer. To place a value on the output of an industry whose products are not sold to ultimate consumers, but are used in stead by other industries in the course of their own production, an addition set of calculations is necessary using input-output. The inputoutput system translates final demand for a given product into the output that is required from all other industries to produce the ma terials needed to manufacture that product or service. The third stage is to project the input-output coefficients. The input-output tables used as a base in the economic growth model are pub lished by the Office of Business Economics, U.S. Department of Commerce. However, these input-output tables incorporate the technology and product mix for a base year, and may not reflect adequately the technology and product mix which may prevail during the period for which the projection is being made. To account for this difference, two methods are used: De tailed analyses are made of the changes taking place in the technology of various industries as well as the changes expected in product mix due to differing growth rates of product groups within industries. For industries for which detailed studies are not made, analyses are developed to determine the direction and mag nitude of change in the use of its products by other industries. Next, employment estimates by industry are developed. This is accomplished by use of a set of industry productivity projections. The final stage is to balance the model. Projections con tain many complex relationships among eco nomic variables that were developed through a lengthy sequence of operations. It is neces sary to have a set of checks and balances to insure that the various states of the projections 245 246 BLS H AN DBOOK OF M ETH ODS make up an internally consistent model. The economic growth model is designed to provide a feed-back and balancing procedure with re spect to three of its elements: imports, invest ment, and employment. In practice, all three of these elements must be brought into balance simultaneously. Uses The projections developed in the economic growth program serve a number of uses. The employment projections by industry are used in developing occupational outlook projections. The projections developed by the Bureau form an important part of the U.S. Government’s report to OECD on long-term economic outlook for the United States. In addition, other Gov ernment agencies use various facets of the eco nomic growth projection to develop projections in their specific areas of responsibility. Outside of government, the projections of GNP and in dustry growth patterns are important sources of information for industry analysts. Their use of these projections are primarily in diversifi cation studies, market analysis, and long-term capital planning. The structure of the economic growth model is developed so that other analytical use is made of the program. Specifically, the program is used to analyze the industry by industry effect of changing levels and patterns of various types of demand such as defense expenditures. Simi lar types of analysis also is performed for ex ports and imports to provide information on the employment requirements for foreign trade. Technical References Number 1. U.S. Department of Labor, Bureau of Labor Statistics, P a tte r n s o f U .S . E c o n o m ic G r o w t h : 19 8 0 P r o je c tio n s o f F in a l D em a n d , I n te r -in d u s tr y R ela tion sh ip s, O u tp u t, P r o d u c tiv ity , and E m p lo y m e n t (Bulletin 1672, 1970). 2 . _____ P r o je c tio n s 1 9 7 0 : In te r in d u str y R ela tio n sh ip s, P o te n tia l D em a n d , E m p lo y m e n t (Bul letin 1536,1966). 3. Oliver, Richard P., “Increase in Defense-Related Employment During Vietnam Buildup,” M o n th ly L a b o r R e v ie w , February 1970, pp. 3-10. 4. Bowman, Charles T., “Report on Employment Related to Exports,” M o n th ly L a b o r R e v ie w , June 1969, pp. 16-20. Ronald E. K utscher Appendix A. The BLS Seasonal Factor Method Background An economic time series may be affected by regular intra-yearly ( “ seasonal” ) movements which result from climatic conditions, model changeovers, vacation practices, holidays, and similar factors. Often such effects are large enough to mask the short-term underlying movements of the series. By isolating and re moving the effect of such intra-yearly repeti tive movements, the current evaluation of a series may be made more perceptive. Seasonal movements are found in almost all economic time series. They may be regular, yet they do show variation from year to year and are also subject to changes in pattern over time. Because these intra-yearly patterns are combined with the underlying growth or de cline and cyclical movements of the series ( “ trend-cycle” ) and also random irregularities, it is difficult to determine the pattern with exactness. More than a half-century ago, attempts were made to isolate seasonal factors from time series. Some early methods depended upon smoothing curves by using personal judgment. Other formal approaches were periodogram analysis, regression analysis, and correlation analysis. Because these methods involved a large amount of work, relatively little applica tion of seasonal factor adjustment procedures was carried out. In the mid-1950’s, new electronic equipment made more elaborate approaches feasible in seasonal factor methods as well as in other areas. Using a computer, the Bureau of the Census developed seasonal factors based on a ratio-to-moving-average approach. This was a major forward step, as it made possible the uniform application of a method to a large number of series at a relatively low cost.1 Sub sequent improvements in methods and in com puter technology have led to more refined procedures which are both faster and cheaper than the original technique. The Bureau of Labor Statistics began its work in seasonal factor methods in 1959, pri marily to correct a deficiency in the method then used. Prior to this time, when additional data became available and seasonal factors were generated from the lengthened series, the new factors sometimes differed markedly from the corresponding factors based on the shorter series. This difference could affect any portion of the series. It was difficult to accept a process by which the addition of recent information could affect significantly the seasonal factors for periods as much as 15 years earlier, es pecially since this meant that factors could never become final. The first method de veloped by BLS and introduced in 1960 had two goals: first, to stabilize the seasonal fac tors for the earlier part of the series; second, to minimize the revisions in the factors for the recent period. Since 1960, the Bureau, through continued research, has made numerous changes and im provements in its techniques and in methods of applying them. These changes were described 1 2 as they were introduced. The method intro duced in May 1966 is described in the sections which follow. Characteristics o f BLS Seasonal Factor Method The BLS method is a ratio-to-movingaverage method. It assumes that the three component parts— trend-cycle, seasonal, and irregular— are multiplied together (multiplica tive assumption) to give the original observa tions.3 (See illustration in charts 1 and 2.) The BLS method differs from other similar methods in the following respects: 1. The initial trend cycle is improved by restoring to it any residual trend-cycle which may have found its way into the irregular com ponent. This adjustment for the deficiency in trend-cycle is developed explicitly and is avail 1 Julius Shiskin, Electronic Computers and Business Indica tors. New York, National Bureau of Economic Research, (1957). Occasional Paper No. 57. 2 See U.S. Department of Labor items in the list of Tech nical References at the end of this appendix. 3 A parallel method assuming an additive relationship among the components also has been prepared for experi mental use. 247 248 BLS H ANDBOOK OF M ETHODS able for review in the form of a table of trendcycle correction values. 2. The BLS method provides for changes over time in the producing mechanism (changes in samples, method of collection, method of estimation) or in economic factors (introduc tion of guaranteed annual wage plans, etc.).4 3. The BLS method calibrates each observa tion in order to assign a supplementary weight which is used in the various averaging proc esses. These “ credence factors” reduce the effect of observations having large irregulari ties. They increase the smoothness with which the seasonal factors change over time, and they also keep large irregularities out of the final trend-cycle. 4. A second way in which the BLS method attempts to protect the final trend-cycle from large irregular fluctuations is by using modified original values for computing the centered 12-month moving average in the later stages of the procedure. The credence factors are used in obtaining the modified original values. The BLS method is very complex and re quires an enormous amount of arithmetical computations. Therefore, the method has been adopted for use on electronic computers. At present, two program decks are available: one written in FORTRAN II and another in FORTRAN IV. Special emphasis has been placed on keeping the handling and clerical requirements to an absolute minimum, on pro viding as many aids as possible for analysts or others using the results, and on making the application of the computer program as simple* 6 * A 5-year moving period is used to reflect such changes. The use of the 5-year moving period allows the full impact of a change to be reflected in a relatively short period of time. A review of the basic U.S. labor force series, for ex ample, indicates that in the early 1960’s the standard devia tion of the irregular component is only 0.6 as large as it was in the early 1950’s. This reflects the improvements made in the survey, such as the expansion in the number of sample areas, the increase in the number of households covered, the changes in the estimating methods, and the improvements in the design of selecting households so that there would be a three-quarters overlap in adjacent months and a one-half overlap over the year. 6 Upon request, the Bureau of Labor Statistics will make its seasonal factor computer program available in the following form : (1) A source deck of punched cards containing the program instructions: (2) A deck of punched cards contain ing a text problem which will enable the user to verify the output: (3) Two copies of a document containing operating and reference material. and efficient as possible. Only the data and one specification card are needed to produce a set of completely labelled tables for each series. Any number of series with varying character istics (length of series, starting and ending dates, magnitude of original observations, final table patterns desired) can be processed in a single running without manual intervention. The computer program for applying the BLS procedure is available on request.5 Basic Approach The BLS method attempts to separate an economic time series into three constituent parts: the underlying movement or trend-cycle which is a combination of the long-term, cyclical, and subcyclical movements ( T ) ; the annual repetitive movement within the year, or “ seasonal” ( S ) ; the irregular or residual unexplained movement (I). These three com ponents, when multiplied together, completely and exactly exhaust the original observations ( 0) . (See charts 1 and 2.) The exact allocation among the components is somewhat arbitrary, because there are no simple criteria or gen erally accepted techniques for separating them. The BLS method attempts to strike a good balance between the conflicting objectives of smoothness in the trend-cycle, stability in the seasonal, and randomness in the irregular. The process is an iterative one; each successive iteration provides an improved estimate for each of the components of the original series. Three iterations are used, each having two phases. The first phase of each iteration starts with a centered 12-month moving average as the estimate of trend-cycle. Seasonal-irregulars are then secured and partitioned into seasonal and irregular components. The second phase in each iteration starts with an improved trendcycle in which the centered 12-month moving average has been improved by recovering the residual trend-cycle from the irregulars of the first phase. The seasonal and irregular com ponents then are developed as in the first phase. The Bureau’s method uses “ credence factors” based on the irregulars from the second phase of each iteration to reduce the effect of large Chart 1. United States Unemployment, Males, Age 20 and Over, 1948-70 Actual Data and Trend-Cycle Millions Millions THE BLS SEASONAL FACTOR METHOD 0 .5 1948 49 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 1970 249 50 250 Chart 2. U n ite d S tates U n e m p lo y m e n t, M ales, A g e 2 0 and O v e r, 1 9 4 8 - 7 0 Seasonal and Irregular Com ponents Percent 140 Percent 140 £Seasonal 130 i on lO U 120 1 120 S H | H i a 5 i 5 ! S i i 5 i S i 5 i S i S i 5 no 100 1AA lU U 90 90 80 on oU 70^ V jjjw.liiijmiipww \0 70 \ smmuumiiiimi.... 130 130 jrregular i | i^ U l lCi\J I A no | H lyrnf^n^^ 100 1 94 8 4 9 50 51 52 53 54 55 HQ 56 57 58 59 60 61 62 63 64 1 0 0 65 66 67 68 v 90 ^0 6 9 1970 BLS HANDBOOK OF METHODS 110 251 THE BLS SEASONAL FACTOR METHOD irregularities in the original observations. These are supplementary weights in which observa tions with small6 irregulars are given more weight than observations with large irregulars in calculating the trend-cycle and the seasonal factors. Detailed Procedure o f BLS Seasonal Factor Method First iteration First Phase. The initial estimate of trend-cycle (T) is the centered 12-month moving average of the original observations. The trend-cycle value for each month is divided into its cor responding original value (0 ) to produce a series of seasonal-irregular (SI) ratios (0 / T = S I) . Treating each month separately (i.e., all January’s, all February’s), the SI ratios are arrayed by years and moving 7-term weighted averages 7 (S') are secured (in per centage form) as estimates of the unforced seasonal factors. The unforced seasonals (S') for each year are then adjusted to yield a monthly average of 100.0 for the calendar year.8 These forced seasonal factors (S) are the seasonal factors for the first phase of the first iteration. Each seasonal-irregular ratio (SI) is then divided by its forced seasonal factor (S) to yield the random or irregular ratio (SI/S=I). The irregular ratios at this stage may contain some residual trend-cycle.9 To separate the residual trend-cycle from the truly random variation contained in the ir regulars, a 9-term weighted moving average 1 0 (trend-cycle correction) of the first phase ir regulars is secured. This completes the first phase of the first iteration. Second Phase. The second phase of the first iteration starts with an improved estimate of the trend-cycle. This is secured by multiplying the trend-cycle used in the first phase (the centered 12-month moving average of original observations) by the trend-cycle correction (weighted moving average of first phase ir regulars). Using this improved trend-cycle, the second phase repeats the computational steps of the first phase to develop new SI ratios, new unforced seasonals (S '), new forced seasonals (S), and new irregulars (I). At this point, the generated trend-cycle and seasonal components represent the components of the original series fairly well except for the effect of highly 0 References to “ large” or “ small” irregulars have to do with the deviation of the irregular component from 1.000 in relation to the standard deviation of such irregulars. A value of 1.423 or .577 would usually be considered large, while .997 or 1.003 would be small. 7 The weight patterns used are: Weight pattern assigned to seven consecutive seasonal-irregular ratios Seasonal 1st year ... 2nd year ________________ 3rd year __ ____________ Middle years _____ __ 3rd from end ___________ 2nd from end ___________ End year _______________ The underlined value weighted average applies. .281 .213 .160 .120 .270 .221 .179 .141 .137 ___ _ _ _ _ indicates _ _ .242 .207 _ .213 .191 .162 _ ___ .185 .179 .160 .137 _ .157 .164 .157 .141 .120 .160 .179 .185 .179 .160 .162 .191 .213 .221 .213 _ _ .207 .242 .270 .281 the year to which the 8 The forcing is performed in two stages: If the unforced seasonals do not start in January, the first 12 unforced sea sonals are summed and the total divided into 1200 to provide a forcing factor. This factor is then multiplied by the un forced seasonals for the partial year only (through the first December value) to provide the forced seasonals for the incomplete at the beginning of the series. A similar pro cedure is followed at the end of the series if the unforced seasonals do not end in December. For each of the full calendar years between, a forcing factor is computed by dividing the sum of the unforced seasonals for the year into 1200. This factor is then multiplied by the unforced seasonals in that year to produce the forced seasonals. 0 This is because the 12-month moving average is not very good at following abrupt or curvilinear changes in the level of the original series, particularly in the vicinity of peaks and troughs. Since the trend-cycle, seasonal, and irregular components completely and exactly exhaust the original series, any deficiency of the first estimate of trend-cycle is transferred to the seasonal and/or the irregular. However, the seasonal factors are secured by averaging seven SI ratios, each one year apart. This averages the deficiency for seven different years, with the result that the deficiency of the 12-month moving average, as an estimator of trend-cycle, appears mainly in the irregular component. The periods where the 12-month moving average is a poor estimator of trend-cycle, usually have a run of consecutive irregular ratio all on the same side of the base reference ratio of 1.000 instead of being scattered above and below this base. 1 The weight patterns used are: 0 Month Weight pattern assigned to nine consecutive irregulars ____ ____ ____ .370 .341 .256 .115— .082 ____ _ 1st month _ ____ ____ .254 .276 .254 .191 .086— .061 ____ 2nd month ____ .160 .214 .231 .214 .160 .072--.051 ____ 3rd month 4th month _ . .067 .150 .199 .216 .199 .150 .067—-.048 Middle months ___ - -.050 .071 .157 .209 .226 .209 .157 .071--.050 4th month -.048 .067 .150 .199 .216 .199 .150 .067 from end 3rd month from end — --.051 .072 .160 .214 .231 .214 .160 -2nd month _ _ _ ___ __ -.061 .086 .191 .254 .276 .254 from end __ — — -.082 .115 .256 .341 .370 End month __ — The underlined weight indicates the month to which the weighted average applies. 252 BLS HANDBOOK OF METHODS deviant original values. In the BLS method, the influence of such values is diminished by using supplementary weights in addition to the regular weights given in footnotes 7 and 10. The use of these supplementary weights (called “ credence factors” ) gives less impact to the extreme observations (those with large irregu lar components) and relatively more to the neighboring values with smaller irregular com ponents.1 1 With the preliminary credence factors, the SI ratios of the second phase are repartitioned into seasonals and irregulars. The adjusted un forced seasonal is a 7-term moving average of the SI ratios, using as relative weights the product of the standard 7-term weights (given in footnote 7) and the preliminary credence factor associated with each value. These un forced seasonals (S') are forced in the usual manner to total 1200 for the calendar year, after which adjusted irregulars are calculated by dividing the adjusted forced seasonals (S) into the SI values. This repartitioning of the SI ratios removes the large irregular variation from the seasonal component and puts it in the irregular com ponent instead. However, the trend-cycle is still contaminated by deviant original values, be cause it was calculated before the credence factors were developed. The second iteration is designed to provide an improved 12-month average for use in de veloping revised components. In preparation for this, the adjusted irregulars of the first iteration are used to develop revised (inter mediate) credence factors. These intermediate credence factors are computed from the ad justed irregulars in exactly the same way (de scribed in footnote 11) as the preliminary credence factors are obtained from the earlier second phase irregulars. The first use of the intermediate credence factors is to create a modified original series having no large irregularities. Each original value having a credence factor less than 1.000 is replaced by a modified value in which the irregularity has been reduced.1 The creation 1 2 of the modified original series marks the end of the first iteration. Second iteration First Phase. The second iteration is similar to the first iteration, except that the intermediate credence factors are used along with the weight patterns of footnotes 7 and 10. The second iteration, like the first, has two phases. The first phase begins with the centered 12-month moving average of the modified original series previously described. SI ratios are secured by dividing the actual original series (not modi fied) by this 12-month average. The SI ratios are arrayed by month and moving 7-term aver ages, using the weights of footnote 7 and the intermediate credence factors, are taken to yield unforced seasonals (S '). (The credence factors prevent extreme observations from affecting the seasonals.) The forcing process is then applied to yield seasons (S) which aver age 100.0 for the calendar year. Irregulars (I) are secured by dividing the latest SI ratios by their corresponding forced seasonals (S). These irregulars may include some residual trend-cycle because of the failure of the 12month moving average to fully penetrate the 1 The credence factors are computed as follows. First, a 1 test is made to locate extreme irregulars falling outside of the 3 sigma limit. These irregulars are replaced by a value of 1.000 for a new sigma calculation. A moving 61-term standard deviation of the irregulars is computed for calibrat ing the irregular associated with the middle (31st) term. For the 30 terms at the beginning (end) of the series, the first (last) centered value is used for the calibration. Each Ir regular is then standardized by getting its absolute difference from the mean of the 61 terms used to secure its standard deviation, and dividing this difference by the standard devia tion. A preliminary “ credence factor” is assigned to each value, based on its standardized irregular, as follows: For a stan dardized irregular of 1.000 or less, the credence factor is 1.000. For a standardized irregullar of 2.800 or more, the credence factor is 0.000. For a standardized irregular between 1.000. For a standardized irregular of 2.800 or more, the standardized irregular. The 2.8 sigma limit makes it e x tremely unlikely for a "good” value, not affected by a strike or other such aberration, to be assigned zero credence. Only one-half of one percent of the values in the normal distribu tion lie beyond this limit. On the other hand, “bad” values which deserve to be disregarded have a much higher prob ability of falling outside the limit. 1 The amount of reduction for each observation is such 2 that the deviation of its new irregular from 1.000 is the product of the credence factor and the deviation of its old irregular component. Thus, each modified original value is the product of the trend-cycle and seasonal components de veloped in the first iteration, times an irregular which has been scaled down if it exceeded one standard deviation. The modified original values are used only to secure an improved 12-month moving average with which to start the second iteration: seasonal-irregulars are always calculated from the actual original values. THE BLS SEASONAL FACTOR METHOD peaks and troughs of the modified original series. A trend-cycle correction is computed by arraying the irregulars in normal time se quence and taking moving 9-term averages, using the weights of footnote 10 and the inter mediate credence factors. This completes the first phase of the second iteration. Second phase. The second phase of the second iteration starts with an improved trend-cycle, which is the product of the centered 12-month moving average of the modified originals and the trend-cycle correction previously described. The second phase repeats the steps and pro cedures of the first phase to develop new SI ratios, new unforced seasonals (S') making use of the intermediate credence factors, new forced seasonals (S), and new irregulars (I). These second phase irregulars are used to calculate final credence factors in the same way as before (see footnote 11). Then a new modi fied original series is secured in the same man ner as before (see footnote 12), using the final credence factors. This completes the second iteration. Third iteration The third iteration carries the refinement process still further. It follows the same steps as in the second iteration, from the centered 12-month moving average of the newly modi fied original series up to the development of the irregular component near the end of the second phase. This completes the partitioning of the series into the final trend-cycle, seasonal, and 1 Additional iterations yield little further modification. The 3 decision to stop with 3 iterations was based on the very small changes occurring after the third iteration, the rea sonable fit of the trend-cycle to the original data, and the cost of additional processing. 253 irregular components.1 As supplementary in 3 formation to aid in analysis and evaluation, the final irregular is standardized using 61-term moving sigmas, which are printed also. The seasonally adjusted series is calculated by di viding the original series (0 ) by the final sea sonal factors (S). Cautionary notes In applying the method, the user should be aware that the result of combining series which have been adjusted separately will usually be a little different from the direct adjustment of the combined series. For example, the quotient of seasonally adjusted unemployment divided by seasonally adjusted labor force will not be quite the same as when the unemployment rate is adjusted directly. Similarly, the sum of sea sonally adjusted unemployment and seasonally adjusted employment will not quite match the directly adjusted labor force. Separate adjust ment of components is usually preferable if their seasonal patterns are different, provided the increased measurement errors in the com ponents are not excessive and that the amount of work does not proliferate unduly. Finally, it is worth noting that the avail ability of a fast, efficient procedure for making seasonal computations can easily lead to the processing of large numbers of series without allotting enough time to review the results. No standard procedure can take the place of care ful review and evaluation by a skilled analyst. A subjective review of all results is strongly recommended. The computer program for ap plying the BLS method facilitates such review by providing the needed materials in a logical and easily used format. 254 BLS HANDBOOK OF METHODS Technical References Number 1. Barton, H. C., Jr., “Adjustment for Seasonal Variation,” F ed er a l R e s e r v e B u lletin , June 1941, pp. 518-528. The classic account of the FRB ratio-to-moving-average method, in which the analyst uses his skilled judgment to draw freehand curves at key stages of the procedure. 2. Macaulay, Frederick R., T h e S m oo th in g o f T im e S e ries . New York, National Bureau of Eco nomic Research (1931). Publication of NBER No. 19. An early discussion of moving averages and of the criteria for choosing one average rather than another. 3. Organization for Economic Co-operation and Development, S ea son a l A d ju s t m e n t on E le c tron ic C o m p u te r s, Paris, 1961. The report and proceedings of an international conference held in November 1960. De scribes experience in the United States, Canada, and several European countries. Includes theoretical sections relating to calendar (trading day) variation and general properites of moving averages. 4. Shiskin, Julius, E lec tro n ic C o m p u te r s and B u sin e ss In d ica to rs, New York, National Bureau of Economic Research (1957). Occasional Paper No. 57. Also published in Jou rn a l o f B u si n e ss, Vol 30, October 1957, pp. 219-267. Describes applications of the first widely used computer program for making seasonal adjustments. 5. U.S. Department of Labor, Bureau of Labor Statistics, T h e B L S M e th o d o f D e r iv in g S e a son al F a c to r s . Paper presented at the Interstate Conference on Labor Statistics, Newport, R.I., June 16, 1960. 6. U.S. Department of Labor, T h e B L S S ea so n a l F a c to r M e t h o d ; paper presented at the 1960 Annual Meeting of the American Statistical Association, Palo Alto, Calif., Aug. 23, 1960. 7. U.S. Department of Labor, N e w S ea son a l A d ju s t m e n t F a c to r s f o r L a b o r F o r c e C o m p o n e n ts ; Special Labor Force Report No. 8 (1960). Includes article by Morton S. Ralf and Robert L. Stein published in the M o n th ly L a b o r R e v ie w , August 1960, pp. 822-827. 8. U.S. Department of Labor, T h e B u r ea u o f L a b o r S ta tistic s S ea son a l F a c to r M e t h o d ; Paper presented at the Interstate Conference on Labor Statistics, Atlantic City, N.J., June 21, 1962. 9. U.S. Department of Labor, The Bureau of Labor Statistics Seasonal FactorMethod (1963 R e v i s i o n ) ; paper presented at the Interstate Conference on Labor Statistics, San Fran cisco, Calif., June 27, 1963. 10. U.S. Department of Labor, T h e B L S S ea so n a l F a c to r M e th o d , I t s A p p lica tio n b y an E le c tron ic C o m p u te r (1963). 11. U.S. Department of Labor, The BLS Seasonal Factor Method (1961,.). (1964). 12. U.S. Department of Labor, The BLS Seasonal Factor Method (1966). (1966). Appendix B. Industrial Classification Much of the usefulness of BLS statistics pre sented by industries is due to the fact that they can be compared with other types of data for the same industries. This is possible because BLS and other Federal and State agencies fol low as closely as possible a single system to define and classify industries in the U.S. eco nomy. The Government publishes a Standard Industrial Classification Manual (SIC) of in dustries based on principles set forth by a technical group made up of government and industry experts.1 The Bureau of Labor Sta tistics took part in the development of the SIC over a long period of years and continues to work actively with the Office of Management and Budget and other agencies in seeking to improve the system. Four basic principles were followed in de veloping the classification :1 2 (1) The Classification should conform to the existing structure of American industry. (2) The reporting units to be classified are es tablishments, rather than legal entities or companies. (3) Each establishment is to be classified according to its major activity. (4) To be recognized as an industry, each group of establishments must have significance from the standpoint of the number of persons em ployed, volume of business, and other impor tant economic features, such as the number of establishments. As there are thousands of products and ac tivities, the system provides for grouping these into categories, both narrow and broad, to en hance the value of industrial statistics for users interested in different levels of detail.3 Using the SIC as a guide, the Bureau classi fies the reports received from each factory, shop, or store according to major product or activity. The SIC is used in the same way by the agencies supplying the Bureau with its universe lists and benchmark data. Hence, a high degree of orderliness and consistency is attained, which benefits not only the users of all BLS establishment statistics, but also the users of all Government figures. Certain operational problems make it im practicable, however, to secure complete uni formity by this process. Also, specific modes of applying the SIC differ from one statistical program to another. For example, there may be differences in the way in which “ major ac tivity” is determined; or changes in the major activity of individual establishments which occur over time may be handled in statistical time series in different ways. Consequently, the use of the same manual and following a com mon set of principles of application do not always result in identical industry classifica tions of a given establishment by all agencies, or even by all programs within BLS. There fore, any major deviations from the normal method of handling industrial classification will be described in the chapters on BLS estab lishment statistics, such as those on employ ment, work injuries, and the like. The standard definition of establishment is stated as follows: An ‘establishment’ is an economic unit which pro duces goods or services—for example, a farm, a mine, a factory, a store. In most instances, the establishment is at a single physical location; and it is engaged in only one, or predominantly one, type of economic ac tivity for which an industry code is applicable. Where a single physical location encompasses two or more distinct and separate economic activities for which 1 Office of Management and Budget, Standard Industrial Classification Manual, 1967. 2 Ibid. pp. IX, X . 3 The SIC provides for different levels of aggregation. The broadest level divides the economy into 10 Divisions: A. Agriculture, forestry, and fisheries; B. Mining; C. Contract construction; D. Manufacturing; E. Transportation, com munication, electric, gas, and sanitary services: F. Wholesale and retail trade; G. Finance, insurance, and real estate; H. Services; I. Government; and J. Nonclassifiable establish ments. At the 2-digit level all products and services are combined into 99 “major groups.” Thus, in the Manufacturing Division, establishments engaged in manufacturing machinery, apparatus, and supplies for the generation, storage, trans mission, transformation, and utilization of electrical energy are combined into Major Group 36, “ Electrical machinery, equipment, and supplies.” The 3-digit level provides several hundred categories. In the electrical machinery major group the SIC provides 8 groups of industries: 361. Electric transmission and distribu tion equipment; 362. Electrical industrial apparatus; 363. Household appliances; 364. Electric lighting and wiring equip ment; 365. Radio and television receiving sets, except communication types; 366. Communication equipment; 367. Electronic components and accessories; 369. Miscellaneous electrical machinery, equipment, and supplies. Thousands of products and activities are distinguished at the 4-digit level. For example, in Group 367, five industries are defined: 3671. Radio and television receiving type electron tubes, except cathode ray; 3672. Cathode ray picture tubes; 3673. Transmitting, industrial, and special purpose electron tubes; 3674. Semiconductors and related devices; 3679. Elec tronic components and accessories, not elsewhere classified. 255 256 BLS HANDBOOK OF METHODS different industrial classification codes seem applicable, such activities should be treated as separate establish ments and classified in separate industries, provided it is determined that: (1) such activities are not ordi narily associated with one another at common physical locations; (2) no one industry description in the Stan dard Industrial Classification includes such combined activities; (3) the employment in each such econom ic activity is significant; and (4) reports can be prepared on the number of employees, their wages and salaries, and other establishment type data. An establishment is not necessarily identical with the business concern or firm, which may consist of one or more establishments. Also, it is to be distinguished from organizational sub units, departments, or divisions within an establish ment. Supplemental interpretations of the definition of an establishment are included in the industry descrip tions of the Standard Industrial Classification.4 4 Ibid., p. X . Appendix C. Geographic Classification United States and States All statistical series of the Bureau were ex panded in 1959 to include Alaska and Hawaii. Due to the relatively small populations of these new States, the effect on national time series was small. Where statistics are published by States, data for these States and other small States are shown where reliable data are available. Generally speaking, BLS assigns an estab lishment to the State in which its employees are reported for payroll tax purposes, since these sources are used both for deriving bench mark levels and for drawing samples. Standard Metropolitan Statistical Areas Some of the Bureau’s data are presented not only on a national and State basis, but also for selected metropolitan areas. Comparability of area statistics among Federal agencies is main tained by means of a set of published standard definitions.1 Defining metropolitan areas for statistical purposes is done on the basis of in formation about population and about economic and social ties among cities. To qualify as a Standard Metropolitan Statistical Area under current rules, an area must have at least one city with 50,000 or more inhabitants, or two contiguous cities meeting certain other cri teria.1 Which counties are to be included in the 2 standard area is determined on the basis of criteria relating to metropolitan character and extent of economic and social communication among counties.3 The Bureau of Labor Statistics is repre sented on the Federal Committee on Standard Metropolitan Statistical Areas, and plays an active role in the Technical Committee on Area Definitions. Both interagency groups are under the chairmanship of the Office of Management and Budget. State Economic Areas State economic areas are relatively homo geneous subdivisions of States developed by the U.S. Department of Commerce’s Bureau of the Census, in cooperation with the X / . Depart J& ment of Agriculture’s Bureau of Agricultural Economics, and several State and private agen cies. They consist of single counties or groups of counties which have similar economic and social characteristics. The boundaries of these areas have been drawn in such a way that each State is subdivided into relatively few parts, with each part having certain significant char acteristics which distinguish it from adjoining areas.4 — Robert B. Steffes 1 U.S. Office of Management and Budget, Standard M etro politan Statistical Areas, 1967. 2 The population criteria are: 1. Each standard metropolitan statistical area must include at least: (a) One city with 50,000 or more inhabitants, or (b) Two cities having contiguous boundaries and consti tuting, for general economic and social purposes, a single community with a combined population of at least 50,000, the smaller of which must have a population of at least 15,000. 2. If two or more adjacent counties each have a city of 50,000 inhabitants or more (or twin cities under 1 ( b ) ) and the cities are within 20 miles of each other (city limits to city limits), they will be included in the same area unless there is definite evidence that the two cities are not eco nomically and socially integrated. 3 Ibid., pp. 1-2. * U.S. Bureau of the Census, Census of Population 1960. Re port PC (1) 1A. Number of Inhabitants. U.S. Summary, p. X X V II. ☆ U.S . GO VER NM EN T PRINTING OFFICE: 1972 0— 430—780 257 BUREAU OF LA B O R STA TISTIC S RE G IO N A L OFFICES Region I 1603-JFK Federal Building Government Center Boston, Mass. 02203 Phone: 223-6762 (Area Code 617 Region V 8th Floor, 300 South Wacker Drive Chicago, III, 60606 Phone: 353-1 880 (Area Code 31 2) Region 1 1 341 Ninth Ave., Rm. 1025 New York, N.Y. 10001 Phone: 971-5405 (Area Code 212) Region VI 1100 Commerce St., Rm. 6B7 Dallas, Tex. 75202 Phone: 749-3516 (Area Code 214) Region III 406 Penn Square Building 1317 Filbert St. Philadelphia, Pa. 19107 Phone: 597-7796 (Area Code 215) Regions V II and V III Federal Office Building 911 Walnut St., 10th Floor Kansas City, Mo. 64106 Phone: 374-2481 (Area Code 816) Region IV Suite 540 1371 Peachtree St. NE. Atlanta, Ga. 30309 Phone: 526-5418 (Area Code 404) Regions IX and X 450 Golden Gate Ave. Box 36017 San Francisco, Calif. 94102 Phone: 556-4678 (Area Code 415) Regions V II and V III will be serviced by Kansas City. Regions IX and X will be serviced by San Francisco.