document resume ed 067 505 bls handbook of methods for … · methodology; national surveys;...

ED 067 505

TITLEINSTITUTIONREPORT NOPUB DATENOTEAVAILABLE FROM

DOCUMENT RESUME

VT 016 766

BLS Handbook of Methods for Surveys and Studies.Manpower Administration (DOL), Washington, D.C.BLS-Bull-171171248p.Superintendent of Documents, U.S. Government PrintingOffice, Washington, D.C. 20402 (2901-0659, $2.00)

EDRS PRICE MF-$0.65 HC-$9.87DESCRIPTORS Costs; *Data Analysis; *Data Collection; Employment

Patterns; Industrial Relations; *Labor Econotics;*Methodology; *National Surveys; OccupationalSurveys; Productivity; Trend Analysis; Wages

IDENTIFIERS BLS; Bureau bf Labor Statistics

ABSTRACTThis handbook outlines the scope of the Bureau of

Labor Statistics programs and the methods it uses to collect,tabulate, evaluate, and publish a wide range of economic andstatistical information from the studies and tfurveys that itconducts. These statistical programs 62:a organized according tosubject matter areas. The broad subject areas covered in thispublication are current employment analysis, manpower structure andtrends, prices and living conditions, wages and industrial relations,productivity and technology, occupational safety and health, andeconomic trends and labor Conditions. Each chapter contains a list oftechnical references, and the appendix includes procedures for theseasonal factor method of analysis. (MF)

it

soS2.90 03 99/,9TOIA

BLSHandbookHandbookof Methodsfor Surveysand StudiesBULLETIN 1711

U.S. DEPARTMENT DF HEALTH.EDUCATIDN 131 WELFAREOFFICE DF EDUCATIDN

THIS DOCUMENT HAS BEEN REPRO-DUCED EXACTLY AS RECEIVED FROMTHE PERSON OR ORGANIZATION ORIGINATING IT POINTS OF VIEW OR OPINIONS STATED DO NOT NECESSARILYREPRESENT OFFICIAL OFFICE OF EDUCATION POSITION OR POLICY

U. S. DEPARTMENT OF LABORJ.D. Hodgson, Secretary

BUREAU OF LABOR STATISTICSGeoffrey H. Moore, Commissioner

1971

For sale by the Superintendent of Documents, U.S. Government Printing OfficeWashington, D.C. 20402 - Price ;2.00

Stock Number 2901-0859

2

Preface

The Bureau of Labor Statistics is essentially a factfinding agency, the functionsof which are the collection, tabulation, evaluation, and publication of a wide rangeof economic and statistical information. A part of the BLS' responsibility isto 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 Americansocietyadministrators of businesses, large and small; workers; union officials;academicians; technicians; government policymakers and administrators at Fed-eral, State, and local levels; and others of the general public.

The methods employed by the Bureau in collecting, analyzing, and presenting itsdata are often highly technical and complex. A description of these methods at alevel of technical detail which would meet the requirements of all users might borethe 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.

3

,`$

Contents

Introduction

Current Employment Analysis

Chapter :1. Labor force, employment, and unemployment2. Employment, hours, and earnings3. Job vacancies and labor turnover

Page

1

71735

Manpower Structure and Trends

4. Employment of scientific and technical personnel 435. Occupational outlook 476. Projections of the labor force 537. Industry-occupational matrix 55

Prices and living conditions

8. Consumer expenditures and income 599. Family budgets 69

10. Consumer prices 7511. Wholesale prices 9712. Industry-sector indexes 11313. Spot market prices 119

Wages and industrial relations

14. Occupational pay and supplementary benefits 12315. Employee earnings and hours frequency distributions 13716. Union wage rates 14717. Current wage developments 15318. Employer expenditures for employee compensation 16119. Work stoppages 17120. Collective bargaining agreements 18121. Union membership 18722. Annual earnings and employment patterns of private nonagricul-

tural workers 19523. Measuring collective bargaining settlements 20124. Wage chronologies and salary trend reports 209

Productivity and technology

25. Output per man -hour: Private sector 21326. Output per man-hour measures: Industries 21927. Technological change 22728. Construction labor requirements 235

Occupational safety and health

29. Occupational safety and health statistics 239

ContentsContinued

Rag*Economic Trends and Labor Conditions

30. Foreign labor conditions, international comparisons, and traderesearch 241

31. Economic growth studies 245Appendix : A. The BLS seasonal factor method 247

B. Industrial classification 255C. Geographic classification 257

1'4

BLS Handbook of Methods for Surveys and Studies

Introduction.

The country is hungry for information; everything ofa statistical character, or even of a statistical appear-ance is taken up with an eagerness that is almostpathetic; the community have not yet learned to be halfskeptical and critical enough in respect to such state-ments.

With these words Gen. Francis A. Walkergreeted Carroll Wright in 1873, as Dr. Wrightassumed charge of the Massachusetts Bureauof Statistics and Labor. And when as U.S.Commissioner of Labor, he issued his first an-nual report in March 1886, Carroll Wrightestablished the policy of explaining his sta-tistical methods to his readers and of seekingto Lvoid misinterpretation of the figures pre-sented. For example, he said:

In stating the facts as they have been found by theagents of the Bureau, many terms are used which arecapable of varied applicationsome even are of doubt-ful meaning when considered metaphysically, but allsuch terms are used in this report in their commonacceptation ; as, for instance, the term "overproduction"is used to indicate that condition of a locality, state, orcountry when more goods have been produced than aresufficient to meet the ordinary demand . . . .1

In the same report there are statements ontesting the validity of figures (p. 141), prob.lems of nonresponse (p. 90), and restrictionson coverage (headnotes to tables). Warningsas to inadequacies of available informationoccur frequently. During the 86 years whichfollowed the initial report, the definitions,methods, and limitations of the data publishedby the Bureau of Labor and its successor, theBureau of Labor Statistics, have been ex-plained again and again. The reason for thisis not merely to make the readers "skeptical,""critical," and aware of the known limitationsof the statistics, but also to instruct them inthe proper use of the information and to assurethem that proper standards have been ob-

served. Furthermore, whereas one might ex-pect to breed a certain amount of doubt abouta statistical survey by revealing its lack ofperfection, frankness about unavoidable de-fectE more often has the opposite effect, andpublic confidence in the work is reinforced inthe process. The most grave doubts arise whenthings crying for explanation are not ex-plained.

The Committee on Government Statistics andInformation Services emphasized 30 years agothat the Central Statistical Board "should urgeon each collecting agency the importance ofpublication by agencies of frank appraisals ofthe extent of noncomparability, incompleteness,and inaccuracy which may be inherent in theirreports at any given time. This candid policyshould enhance and not diminish the scientificprestige of the collecting agency." 2

Full understanding of the statistical seriesand studies of the Bureau of Labor Statisticsis not to be gained solely from detailed descrip-tions of them, but also from appraisal of thephilosophy and approach of the Bureau and ofthe manner in which it functions.

Background

The history of the Federal Bureau of Laborextends back to 1884. Before the creation of thecabinet post of Secretary of Labor, the Bureaufor a time was known as the Department of

I See Industrial Depressions, The First Annual Report of theCommissioner of Labor, March 1886, pp. 15-16.

!Committee on Government Statistics and InformationServices. Government Statistics, April 1937, p. 53. Recommen-dation #16 of the Committee states: "Continued criticism andanalysis 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, componentitems and weighting systems of index numbers; and (f)methods and practices in the presentation of data.... Frankappraisals of comparability, completeness, and accuracyshould be pUblished." (pp. 48 -49).

2 BLS HANDBOOK OF METHODS

Labor. From the Bureau's beginnings in theadministration of President Arthur until itbecame a part of a cabinet department underPresident Wilson, it accumulated nearly 3 dec-ades of experience in collecting, interpreting,and presenting facts crucial to the welfare ofworkers. Details of early Bureau history andof developments of later years are to be foundelsewhere.' However, in describing the variousstatistical programs 4 in this volume, some ofthe events which led to the development ofparticular statistical measures are recounted.Against this historical background emerges thephilosophy and posture of the Bureau of LaborStatistics as the impartial observer and in-terpreter of trends important to the welfareof workers. Voluntary reporting and the pre-serving of the confidential nature of reporteddata are important characteristics of BLSprograms.

Voluntary Reporting andConfidentiality

In the 86-year history of the Bureau's opera-tion it has asked hundreds of thousands offirms and individuals to provide informationclosely related to their daily affairs and theirpersonal lives. To some of them who have sup-plied the desired information, the Bureau hasgone back a second time, a third time, andperhaps dozens of times, for later informationon the same subject or for new types of infor-mation. The response has been remarkable inits generosity, even when it is remembered thata sustained effort has been made to keep therequests reasonable. In no small measure, thecooperation received is due to the great caretaken to avoid identifying the firm or the per-son supplying the information. The fact thatBureau employees pledge themselves to protectthese data is less important than that they havea deep understanding of the adverse longrunconsequences of even a single lapse. They areaware of the greater worth, in terms of purestatistical validity, of ths information providedvoluntarily as compared with that suppliedunder legal sanctions. The only inducementemployed is to tell the respondent that his

contribution is important to the ultimate suc-cess of the survey and that he may, find thesurvey results useful in his own pursuits. Thepolicy of not identifying the respondent is im-plemented by combining the data reported bythe different sources and issuing the findings insummary form.

Another assurance given the respondent isthat his report will be used for statistical pur-poses only. Attempts to "break" this policy, byorganizations or individuals who wanted accessto data in our possession and were willing togo to the courts to secure it, have been success-fully resisted.' Another form this problemtakes is the case in which an administrativeagency of government seeks court action tocompel a company to release its file copy ofinformation provided in confidence to a sta-tistical agency."

While it cannot be proved that these policiesresult in more reliable statistics, Bureau Com-missioners and their staffs over the years havebeen convinced from their experience that it isso. It is notable that some other Federal agen-cies (esiCecially the Bureau of the Census),well-equipped with authority to compel thesubmittal of certain reports, rarely if ever in-voke this power. Rather, they choose to relyupon forms of suasion similar to our own. TheBureau of Labor Statistics, while its functionsas a statistical agency are prescribed by law,'has always relied upon voluntary cooperationof respondents in collecting information.

3 See, for example, the Secretary of Labor's First AnnualReport, 1913, for history 1884-1913; and U.S Department ofLabor, 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).

4For a handy reference to BLS programs, showing theirprincipal characteristics. see U.S. Department of Labor.Bureau of Labor Statistics, Major BLS Programs (issuedannually).

3 For example, see Norweigan Nitrogen Company v. UnitedStates, 288 U.S. 294; United States V. Kohler, 13 Fed. RidesServ. 33,333 (E.D., Pa. 1949); Hawes v. Walsh, 277 Fed. 569, theCourt of Appeals for the District of Columbia. In all of thesecame, the courts sustained the policy of protecting the con-fidentiality of information given voluntarily and in confidenceto an agency of the Federal Government.

8 See Supreme Court of the United States, St. Regis PaperCompany, Petitioner, v. United States, No, 47, October term,1961.

?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, I 3, 37 Stat. 737.

INTRODUCTION 3

BLS Role, Staff, and Organization

Among Federal agencies collecting and issu-ing statistics, the Bureau of Labor Statisticshas been termed a "general-purpose statisticalagency." 8 The Bureau's figures are preparedto 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 andsituations. While the data serve some admin-istrative purposes, they are free from theconstraints which sometimes result from theclose ties normally existing between operationsand operational statistics. BLS statistics areoften quite specialized, yet they meet generaleconomic and social data requirements. As theneeds of users are likely to differ from eachother and also over time, no statistic is idealfor all. This makes it important that thecharacteristics of the measures and their pos-sible limitations be well understood,

The Bureau plays a larger and more signifi-cant role than merely publishing general-purpose statistics. Its activities frequently in-fluence, and sometimes are crucial to, the de-termining and shaping of public policy. TheBureau's experts have the keen understandingof economic and social forces which resultsfrom intensive, and continuous involvement infactfinding and in the painstaking analysis ofdata.

Staff

The Bureau's work extends beyond the initialcollection and processing of data. Over theyears, it has developed a staff of professionalanalysts, trained in the disciplines of eco-nomics and other social sciences, to search outthe implications of survey findings for thewelfare of workers and to present them ascogently and as promptly as possible in writtenand oral form. How successfully this can beaccomplished depends greatly upon the com-petence of the analysts and of their supportingpersonnd.

In BLS, analytical and statistical work isperformed by economists, statisticians, and

Bureau of the Budget. Statistical Services of the UnitedStates Government (revised edition), 1983, p. 7.

mathematical-statisticians with the aid of anexperienced corps of programers, systemsanalysts, and other professionals, as well asstatistical clerks. For analytical work, eco-nomists at even the lowest grade level mustmeet Civil Service Commission requirementsroughly equivalent to a college major in eco-nomics. There are comparable requirementsfor other professionals. The greatest effort ismade 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. TheBureau provides training needed for on-the-jobskills, as background to special assignments,to keep professionals° abreast of changes intheir fields, and to aid higher level and execu-tive professionals in obtaining the best resultsfrom their staffs.

In training staff, a special effort is made toimpart detailed knowledge of the techniquesused in collecting and compiling the statistics,so that maximum application of data results tocurrent problems can be made without a riskof exceeding the limits of their significance.

Organization

The statistical programs of the Bureau weredeveloped, for the most part, independently ofeach other, taking on characteristics suited tothe requirements of the subject under observa-tion. As a result, the Bureau was organizedaccording to subject-matter areas, an arrange-ment which has proved efficient and has beencontinued over the years. Expertise in tech-niques, economic analysis, and other staff ac-tivities across subject-matter lines were addedto provide better utilization of the Bureau'sresources.

As the Bureau's collection activities in-creased, regional offices were established in1943 to provide administration of the field pro-grams and staff. Another function of the officeswas to disseminate data to local users and tofurnish technical advice and assistance to Stateagencies 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 weutilize In compiling our statistics.


Special recognition of the competence of theBureau in the field of statistics was given bythe Secretary of Labor in 1955 when he dele-gated to the Commissioner of Labor Statisticsthe responsibility for continuously reviewingall of the statistical programs of the Depart-ment of Labor and of making recommendationsfor their improvement.

Consultation and Advice on StatisticalPrograms

A statistical program too much detachedfrom the uses of its data may fail in its prin-cipal mission. To avoid sterility and stagnation,the Bureau continuously invites advice andideas from users and experts in business, labor,and academic organizations and individualmembers of the public. Over the years, theadvice the Commissioner of Labor Statisticshas received on policy and technical mattersfrom responsible parties, relating to the col-lection and analysis of our statistics, hasusually been sound and therefore very helpful.Of course, decisions on statistical policy havealways been the final responsibility of theCommissioner.

In order to keep in touch with the currentand anticipated needs of business and laborgroups 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 advisorycapacity with respect to technical problems,consult on Bureau programs, and provide per-spectives on Bureau programs in relation toneeds of their members. The councils accom-plish their work in general sessions and alsothrough committees designated to subject-matter fields on a more specialized basis. Com-mittee memberships are augmented by otherpersons in industry or labor who have specialcompetence although not council members. Thecouncils 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 ofthe councils and the subcommittees serve intheir individual capacities, not as representa-tives of their organizations.

The members of the Labor Research AdvisoryCouncil are designated by the Commissioner ofLabor Statistics under authorization by theSecretary of Labor, from nominations by theDirector of Research, AFLCIO. All researchdirectors of international unions represented inthe AFLCIO, the Railway Labor Executives'Association, and the railroad operating unionsare invited to attend the general meetings ofthe council. The council provides general direc-tion to the advisory activities of trade unionresearch 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 ofLabor, after consultation with the NationalAssociation of Manufacturers, the U.S. Cham-ber of Commerce, and other organizationsbroadly 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 theirindividual capacities or as members of profes-sional organizations. This is most likely to oc-cur when a conceptual or theoretical questionarises which is considered fundamental to thework of the Bureau in a specialized field, andwhere professional acceptance of the Bureau'swork in that field may be reinforced by thefindings of an independent analyst.

It is a fundamental objective of the Bureauthat its statistical practice be built soundlyupon established statistical theory. The objec-tive can be realized only if BLS practitionersare trained in statistics and if they keep theirknowledge up to date. For this reason the Bu-reau encourages their participation in activitiesof the professional societies, their efforts toimprove their education in statistics, !nd con-tinuing contacts with other experts in theirdisciplines. The efforts of other statistical or-ganizations, public and private, are studiedunremittingly so that BLS may reap benefitsfrom the experiences of others.

9

INTRODUCTION

Standard Definitions

Where related statistics cut across programlines or across. Government bureaus, the Bu-reau of Labor Statistics cooperates to themaximum extent possible in the Office of Man-agement and Budget's (formerly, the Bureau ofthe 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 foundto be the most satisfactory source of data formost industrial statistics because it is the firstlevel of business organization for which com-plete recordsi.e., production, employment,purchases, sales, wages, inventories, etc. aregenerally maintained. The establishment is theprimary unit of organization in the businesseconomy and is the first integrated level orcombination of employees devoted to the pro-duction of a related group of products orservices. In BLS programs in which it is ap-

5

plicable, the standardized definition of theestablishment is utilized.°

The Bureau also follows the Office of Man-agement and Budget's definition of "productionand related workers," 1° and uses the "standardpayroll period."' The reader is referred toappendixes B and C for descriptions of thestandards followed with respect to industrialclassification and geographic classification.

'See appendix B.' "Standard Definitions of Types of Workers, Bureau of the

Budget, November 7, 1944. "Production and related workersare defined to include working foremen and all nonsuper-visory workers (including leadmen and trainees) engaged infabricating, 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 closelyassociated with the above production operations. Excludedare supervisory employees (above the working foremenlevel) and their clerical staffs."

I' Standard Definition of. Payroll Periods for EmploymentReports, Bureau of the Budget, March 28, 1952. "In order tomailtain a coordinated system of employment reports andto reduce the reporting burden on respondents, requests madeto employing establishments for statistical information fromtheir payroll records on the number 0 employees, payrolls.hours worked, or related items, should refer to the payrollperiod containing the 12th of the month."

10

Current Employment Analysis

Chapter 1. Labor Force, Employment, and Unemployment

Background

Each month, the Bureau analyzes and pub-lishes statistics on the labor force, featuringinformation on employment, unemployment,and nonparticipation, classified by a variety ofdemographic, social, and economic charac-teristics. These statistics are derived from theCurrent Population Survey, (CPS), which isconducted by the Bureau of the Census. Thesurvey is based on a probability sample ofhouseholds, representative of the civilian non-institutional 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-me of unemployment and employment thanwer previously available. Before the 1930's,aside from attempts in some of the decennialcensuses, no direct measurements were madeof the number of jobless persons. The develop-ment of mass unemployment in the early 1930'sincreased the need for statistics, and widelyconflicting estimates based on a variety of in-direct techniques began to appear. Dissatisfiedwith these methods, many research groups, aswell as State and municipal governmkints, be-gan experimenting with direct surveys of thepopulation o:. samples of the population. Inthese surveys, an attempt was made to classifythe population as employed, unemployed, orout of the labor force, by means of a series ofquestions addressed to each individual. In mostof the surveys, the unemployed were defined asthose who were not working but were "willingand able to work." This concept, however, didnot meet the standards of objectivity that manytechnicians felt were necessary to measureeither the level of unemployment at a point intime or changes over periods of time. Mt cri-

terion "willing and able to work," when appliedin specific situations, appeared to be too intan-gible and too dependent upon the interpretationand attitude of the persons being interviewed.

Out of this experimentation, a set of conceptswas developed in the late 1930's which soughtto meet these various criticisms. According tothe new concepts, the classification of an in-dividual depends principally on his actual ac-tivity within a designated time period, i.e., washe working, looking for work, or engaged inother activities? These concepts were adoptedfor the national sample survey of householdsinitiated by the Works Progress Administra-tion (WPA) in 1940.

Originally termed the Monthly Report onUnemployment when the WPA was responsiblefor the collection of labor force statistics, thehousehold survey was transferred to the Bu-reau of the Census late in 1943. Its name waschanged at that time to the Monthly Report onthe Labor Force (MRLF). The survey titlewas changed once morein 1948to the pres-ent "Current Population Survey" in order toreflect more accurately its expanding role as asource for a wide variety of demographic andeconomic characteristics of the population. In1959, responsibility for analyzing and publish-ing the CPS labor force data was transferredto thLS, although the Census Bureau hascontinued to collect and tabulate the statistics.

Description of Survey

The CPS provides statistics on the civiliannoninstitutional population 16 years of age andover. Figures on the Armed Forces (obtainedmonthly from the Department of Defense) areadded to the CPS estimates to derive estimatesof the total labor force and the total nonin-


stitutional population. Persons under 16 yearsof age are excluded from coverage in the sur-vey because child labor laws, compulsory schoolattendance, and general social custom preventmost of these children in the United Statesfrom working. The institutional population,which is excluded from coverage. consists ofinmates of penal and mental institutions, tuber-culosis sanitariums, and homes tor the aged,infirm, and needy.'

The CPS is collected each month from aprobability sample of approximately 47,000households. Participation in the survey is on apurely voluntary basis. Respondents are as-sured that all information obtained is com-pletely confidential and is used only for thepurpose of estimating national totals.

The time period covered in the monthly sur-vey is a calendar week. A calendar week wasselected as the survey reference period becausethe period used must be short enough so thatthe data obtained is "current" and the timereference not so short that the occurrence ofholidays or other accidental events causes ex-tremely erratic fluctuations in the informationobtained. A calendar week fulfills these condi-tions as well as being a convenient and easilydefined period of time. Since July 1955 theweek containing the 12th day of the month hasbeen used. The actual survey conducted dur-ing the foi,c.wing week, which is the week con-taining the 19th day of the month.

Concepts

The criteria used in classifying persons onthe basis of their labor force activity are asfollows:

Employment. Employed persons comprise (1)all those who, during the survey week, did anywork at all as paid employees, or in their ownbusiness, profession, or farm, or who workedat least 15 hours as unpaid work' rs in a family-operated enterprise and (2) all those who werenot working but who had jobs or businesses

1For a fuller explanation of the Current Population Survey,see Concepts and Methods Used in Manpower Statistics fromthe Current Population Survey, U.S. Department Com-merce and U.S. Department of Labor (1967). Report .13.

from which they were temporarily absentbecause of illness, bad weather, vacation, labor-management dispute, or various personal rea-sons. Each employed person is counted onlyonce. Those who held more than one job arecounted in the job at which they worked thegreatest number of hours during the surveyweek. Excluded from the employed group arepersons whose only activity consisted of workaround their own home (such as housework,painting, repairing, etc.) or volunteer work forreligious, charitable, and similar organizations.

Unemployment. Unemployed persons includethose who did not work at all during the surveyweek, were looking for work, and were avail-able for work in the reference period. Thosewho had made efforts to find work within thepreceding 4-week periodsuch as by register-ing at a public or private employment agency,writing letters of application, canvassing forwork, etc.and .who, during the survey week,were awaiting the results of these efforts, areconsidered to be looking for work. Also in-cluded as unemployed are those who did notwork at all during the survey week and (a)were waiting to be called back to a job fromwhich they had been laid off, (b) were waitingto report to a new wage or salary job scheduledto start within the following 30 days (andwere 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 thelength of time (through the current surveyweek) during which persons classified as un-employed had been continuously looking forwork. For persons on layoff, duration of un-employment represents the number of fullweeks since the termination of their most re-cent employment. A period of 2 weeks or moreduring which a person was employed or ceasedlooking for work is considered to break thecontinuity of the present period of seekingwork. Average duration is an arithmetic meancomputed from a distribution by single weeksof unemployment.

Labor Force. The civilian labor force comprisesthe total of all civilians classified as employed

LABOR FORCE, EMPLOYMENT, AND UNEMPLOYMENT 9

and unemployed. The total labor force includesmembers of the Armed Forces stationed eitherin the United States or abroad.

Not in. Labor Force. All civilians who are rotclassified as employed, unemployed, or in theArmed Forces are defined as "not in the laborforce." These persons are further classified as"engaged in own housework," "in school,""unable to work" because of long-term physicalor mental illness, and 'other." The "other"group includes retired persons, individuals re-ported as too old or temporarily unable to workand the voluntarily idle. It also includes sea-sonal workers for whom the survey week fell inan "off" season and who were not reported aslooking for work and persons who did not lookfor work because they believed either that nojobs were available in the area or that no jobswere available for which they could qtmlify.Persons doing only incidental unpaid familywork (less than 15 hours in the specified week)also are classified as not in the labor force.

The category "not in labor forcein school"includes persons attending school during thesurvey week who had new jobs to which theyWere scheduled to report within 30 days. Allpersonswhether or not attending schoolwho had new jobs not scheduled to begin untilafter 30 days (and who were not working orlooking for work) are classified as not in thelabor force.

For persons not in the labor force, questionsare asked about previous work experience, in-tentions to seek work, desire for a job at thetime of interview, and reasons for not lookingfor work. The questions for persons not in thelabor force are asked only in those householdsthat are new entrants to the sample and inthose that are reentering the sample after 8months' absence. (See Sampling.)

Sampling

The Survey Design

The CPS sample is located in 461 sampleareas comprising 923 counties and independentcities with coverage in every State and theDistrict of Columbia. In all, about 55,000 hous-

ing units and other living quarters are desig-nated for the sample at any time, of whichabout 47,000, containing about 105,000 persons16 years and over, are occupied by householdseligible for interview. The remainder are unitsthat are vacant, converted to nonresidential11F(.3., containing persons who reside elsewhere,or ineligible for other reasons. Of the occupiedunits eligible for enumeration, about 3 to 5percent are not interviewed in a given monthbecause the residents are not found at homeafter repeated calls, are temporarily absent, orare unavailable for other reasons?

Selection of Sample Areas. The entire area ofthe United States consisting of 3,000 countiesand independent cities is divided into 1,931primary sampling units. With some minor ex-ceptions, a primary sampling unit (PSU) con-sists of a county or a number of contiguouscounties. Each of the 237 standard metropoli-tan statistical areas (SMSA's)3 in existence atthe time of the 1970 Census constitutes aseparate PSU. By combining counties to formPSU's, greater heterogeneity is accomplished.Moreover, another important consideration isto have the PSU sufficiently compact in area sothat, with a small sample spread throughout,it can be canvassed efficientli without unduetravel cost. A typical priniary sampling unit,for example, includes both urban and ruralresidents of both high and low economic levelsand encompasses, to the extent feasible, diverseoccupations 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'a) are separatestrata representing themselves. In general,however, a stratum consists of a set of PSU'sas 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.

'For a more detailed description of the sample design andother technical phases of the CPS program, see U.S. Depart-ment of Commerce, Bureau of the Census, Technical PaperNo. 7, "The Current Population SurveyA Report on Meth-odology," (1969).

' See appendix C.

13


Except for the 156 areas mentioned, each ofwhich is a complete stratum, the strata are es-tablished so that their sizes in terms of 1970population are approximately equal. From eachstratum a single PSU is selected to representthe entire stratum. In the 156 strata in whichthere is only a single PSU (the 146 SMSA'sand 10 special cases), the single PSU auto-matically falls in the sample. When the stratumhas more than one PSU, the sample PSU isselected in a random manner in such a waythat its probability of selection is proportionateto its 1970 population. For example, within astratum the chance that a PSU having a popu-lation of 50,000 would be selected is twice thatfor a unit having a population of 25,000.

Selection of Sample Households. The sampledesign calls for a sampiing ratio which de-pends on the predetermined total sample size.For 1971-73, it. is 1 household for every 1,300households in each stratum. The sampling ratiois modified slightly by month, as the size of thesample is held relatively constant despite theoverall growth of the population. The samplingratio used within each sample PSU dependson the proportion that the population of thesample area was of the stratum population atthe time of the 1970 Census. In a sample areawhich was one-tenth of the stratum, the within-PSU sampling ratio that results is 1 in 130.0thereby achieving the desired ratio of 1 in1,300 for the stratum. For each PSU that is astratum representing only itself, the samplingratio is 1 in 1,300 regardless of the size of thePSU.

With each of the 461 sample PSU's, the num-ber of households to be enumerated each monthis determined by the application of the within-PSU sampling ratio rather than through theassignment of a fixed quota. This proceduremakes it possible to reflect, on a current basis,population changes within the sample area.Consequently, the sample as a whole properlyreflects the changing distribution of the popu-lation and avoids the distortion which wouldresult from the application of fixed quotas ofhouseholds or persons based on the populationat an earlier date.

Within each designated PSU, several stages

of sampling may be used in selecting the unitsto be enumerated. The first step is the selectionof a sample of census enumeration districts(ED's), which are administrative units usedin the 1970 Census and contain, on the average,about 250 households. These are selected sys-tematically from a geographically arrangedlisting, so that the sample ED's are spread overthe entire PSU. The probability of selection ofany 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 withineach designated ED. This selection is madewherever possible, from the list of addressesfor the ED compiled during the 1970 Censusor, if the addresses are incomplete or inade-quate, by area sampling methods. The addresslists are used in about two-thirds of the cases,primarily in urban areas, and area sampling isapplied in the remainder. An effort is made tohave all units at one single address includedwithin the same segment. This makes it rela-tively easy for the interviewer to cover all unitsdesignated for the sample.

The list sample is supplemented by a selec-tion of the appropriate proportion of unitsnewly constructed in the PSU since the Censusdate. The addresses of these units are obtainedmainly from records of building permits main-tained by the offices responsible for issuingpermits in that area. A special procedure forupdating parts of the census lists also is fol-lowed to reflect either units missed in theCensus or new construction in areas wherethere is no adequate system of Intilding per-mits. In those enumeration districts where areasampling methods are usedmainly ruralareasthe ED's are subdivided into segments,that is, small land areas having well-definedboundaries 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; theprobability of selection is proportionate to theestimated "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

14


remaining housing units in the segment arethen available for future samples.

Rotation of Sample. Part of the sample ischanged each month. A primary reason for ro-tating the sample is to avoid the problems ofuncooperativeness which arise when a constantpanel is interviewed indefinitely. Another rea-son for replacing households is to reduce thecumulative effect of biases in respOnse thatsometimes occur when the same persons areinterviewed indefinitely. To accomplish thisrotation of the sample on a gradual basis, mapsand other materials for several samples areprepared simultaneously. For each sample,eight systematic subsamples (rotation groups)of segments are identified. A given rotationgroup is interviewed for a total of 8 months,divided into two equal periods. It is in thesample for 4 consecutive months one year,leaves the sample during the following 8months, 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 theirfirst month of enumeration, another eighth arein their second month, and so on ; the lasteighth are in for the eighth time, the fourthmonth of the second period of enumeration.Under this system, 75 percent of the samplesegments are common from month to monthand 50 percent from year to year. This pro-cedure provides a substantial amount of month-to-month and year-to-year overlap in the panel,thus reducing discontinuities in the series ofdata, without burdening any specific group ofhouseholds with an unduly long period of in-quiry.

Collection Methods

Each month, during the calendar week con-taining the 19th day, interviewers contact someresponsible person in each of the sample house-holds in the CPS. At the time of the firstenumeration of a household, the interviewerprepares a roster of the household members,including their personal characteristics (dateof birth, sex, race, marital status, educational

attainment, veteran status, etc.) and their re-lationship to the household head. This rosteris brought up to date at each subsequent inter-view to take account of new or departed resi-dents, changes in marital status, and similaritems. The information on personal charac-teristics thus is available each month for iden-tification purposes and for cross-classificationwith economic characteristics of the samplepopulation.

At each monthly visit, a questionnaire iscompleted for each household member 16 yearsof age and over. The interviewer asks a seriesof standard questions on economic activityduring the preceding week, the calendar weekcontaining the 12th day of the month, calledthe "survey week." The primary purpose ofthese questions is to classify the sample popu-lation into the three basic economic groupsthe employed, the unemployed, and those not inthe labor force.

Additional questions Are asked each monthto help clarify the information on employmentstatus. For the employed, information is ob-tained on hours worked during the surveyweek, together with a description of the cur-rent job. For those temporarily away fromtheir jobs, the enumerator records their reasonfor not working during the survey week,whether or not they were paid for their timeoff, and whether they usually work full or parttime. For the unemployed, he records (1)methods used to find work, (2) the reason theunemployed perSons had started to look forwork, (3) the length of time they had beenlooking 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. Forthose outside the labor force, their principalactivity during the survey weekwhetherkeeping house, going to school, or doing some-thing elseis recorded. In addition, for allhouseholds in the incoming or returning rota-tion group, questions on the work history, rea-sons for nonparticipation, and job seekingintentions of individuals not in the labor forceare asked.

The questionnaires containing the informa-

i5


tion obtained for each person in the sampleare subjected to a field edit by clerks in eachof the 12 regional offices of the Census Bureau.The field edit serves to catch omissions, incon-sistencies, illegible entries, and errors at thepoint where correction is still possible. Manyof the error corrections made in the field editprevent delays in further processing of thequestionnaires in Washington.

After the field edit, the questionnaires areforwarded to the Washington office of theCensus Bureau. All of the questionnaires arereceived in Washington by the end of the weekafter enumeration. The raw data are trans-ferred to computer tape and checked for com-pleteness and consistency.

The interviewers on the CPS, are chieflypart-time workers, although most of the str.ffat any time consists of persons who have hadseveral years experience on the survey. Theyare given intensive training when first re-cruited and have either direct or home studytraining each month, before the survey. More-over, through editing of their completedquestionnaires, repeated observation duringenumeration, and a systematic reinterview ofpart of their assignments by the field super-visory staff, the work of the interviewers iskept under control and errors or deficienciesare brought directly to their attention.

Estimating Methods

To increase the reliability of the labor forcestatistics derived from the sample, the esti-mation procedure uses two stages of ratioestimates and a "composite estimate". Achieve-ment of this rather complicated procedure ismade rapidly and automatically because of theavailability of high-speed electronic digitalcomputers. The principal steps involved are asfollows.

Adjustment for Households Not Interviewed.The weights for all households interviewed areadjusted to the extent needed to account forunits occupied by persons eligible for interviewbut 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 farm). The adjustment ismade separately within each pair of rotationgroups (the incoming pair, the two continuingpairs, 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 asa whole in such basic characteristics as age,color, sex, and farm-nonfarm residence, amongother things. These particular populationcharacteristics are correlated closely with la-bor force participation and other principalmeasurements made from the sample. There-fore, some of the sample estimates can beimproved substantially when, by appropriateweighting of the original returns, the samplepopulation is brought as closely into agreementas possible with the known distribution of thesecharacteristics in the entire population. Suchweighting .is accomplished through two stagesof ratio estimates :

1. First stage. The first stage of ratio esti-mates takes into account differences in thedistribution by color and residence of the popu-lation estimates from the sample PSU's andthat of the total population in each of the fourmajor regions of the country. However, in-dependent distributions of the total populationby residence, cross-classified by color, are notavailable on a current basis. Therefore, using1970 Census data, estimated population totalsby color and residence for a given region arecomputed from population counts for PSU's inthe CPS sample. Ratios then are computedbetween these estimates (based on samplePSU's) and the actual population totals forthe region as shown by the 1970 Census. Inderiving 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, theweights for all sample households from non-self-representing PSU's in a given region are


multiplied by the population ratio for thatregion for the appropriate color-residencegroup.

2. Second stage. The second stage of ratioestimates takes account of current differencesbetween the population distributions of thesample and that of the Nation as a whole byage, color, and sex. Independent estimates ofthe entire population, by these characteristics,are prepared each month. They are calculatedby carrying forward the most recent censusdata (1970) to take account of subsequentaging of the population, mortality, and migra-tion between the United States and other coun-tries. The CPS sample returns (taking intoaccount the weights determined after the firststage of ratio estimates) in effect are used todetermine only the distribution within a givenage-color-sex group by employment status andvarious other characteristics. In developingstatistics, these sample distributions are multi-plied by the ratio of the independent populationestimate 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 compositeestimate. In this procedure, a weighted averageof two estimates is obtained for the currentmonth for any particular item. The first esti-mate is the result of the two stages,of ratioestimates described above. The second estimateconsists of the composite estimate for the pre-ceding month to which has been added anestimate of the change in each item between thepreceding month and the present month, basedupon that part of the sample which is commonto both months (75 percent). Although .theweights for the two components of such a com-posite estimate do, not necessarily have tobe equal, in this instance the weights used forcombining these two estimates are each one-half. Equal weights in this case satisfy thecondition that for virtually all items there willbe some gain in reliability over the estimationprocedure after the first two stages of ratioestimates.

The composite estimate results in a reductionin the sampling error beyond that which isachieved after the two stages of ratio estimates

described; for some items the redo,Ition is sub-stantial. The resultant gains in reliability aregreatest in estimates of month-to-monthchange, although gains also are obtained forestimates of level in a given month, changefrom year to year, or change over other inter-vals of time.

Presentation and Uses

The CPS provides a large amount of detailon the economic and social characteristics ofthe population of the United States. It is thesouree of monthly estimates of total employ-ment, both farm and nonfarm; of nonfarm self-employed persons, domestics, and unpaid help-ers in nonfarm family enterprises; and of totalunemployment, whether or not covered by un-employment insurance. It is a comprehensivesource of information on the personal charac-teristics such as age, sex, race, educationalattainment, 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 thenumbers of hours worked, as distinguishedfrom aggregate or average hours for an in-dustry, permitting separate analyses of part-time workers, workers on overtime, etc. Thesurvey is a comprehensive current source ofinformation on the occupation of workers,whether teachers, stenographers, carpenters,laborers, etc. It also provides limited statisticson the industries in which they work.

Information is available from the survey notonly for persons in the current labor force butalso for those who are outside the labor force,some of whom may be considered to be a "laborreserve." The characteristics of such personswhether married women with or withoutyoung children, disabled persons, students,older retired workers, etc.can be determined.Also, through special inquiries, it is possibleto obtain information on their skills and pastwork experience, if any.

Each month, a significant amount of basicinformation about the labor force is analyzedand published in Employment and Earnings.The tables in this report provide information


on the labor force, employment, and unemploy-ment by a number of characteristics, such asage, sex, color, marital status, industry, andoccupation. Approximately 150 of the most im-portant estimates from the CPS are presentedeach month on a seasonally adjusted basis.These estimates are adjusted by the BLS Sea-sonal Factor Method, which is described in theappendix on seasonal adjustment.

The CPS is used also for a program of spe-cial inquiries to obtain detailed informationfrom particular segments, or for particularcharacteristics of the population and laborforce. Approximately 8 to 10 such special sur-veys are made each year. The inquiries arerepeated annually in the same month for sometopics, including the earnings and total incomesof individuals and families (published by theCensus Bureau), the extent of work experienceof the population during the calendar year, theextent of overtime work at premium pay, usualweekly 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 school-age youth, the employment of recent highschool graduates, the educational attainmentof workers, and the employment situation inpoverty areas of our major cities. Surveys havebeen made periodically on subjects such as jobmobility,, and length of time on current job. Inaddition, surveys are published in very greatdetail on the characteristics of the unemployedand persons not in the labor force.

Generally, the persons who provide informa-tion for the monthly CPS questions also answerthe supplemental questions. Occasionally, thekind of information sought in the special sur-vey requires the respondent to be the personabout whom the questions are asked.

Information obtained through the supple-mental questions is combined with data in theregular schedule to provide tabulations of allthe 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. Reprintsof the articles, together with technical notesand additional tables, are published as SpecialLabor Force Reports.

18

Limitations

Geographic. The CPS is designed to producereliable National estimates. It is not designedto produce estimates for States and areas. Asample which could produce State estimates asreliable as those now published for the Nationwould have to be approximately fifty times aslarge as the present sample.

Sources of Errors in the Survey Estimates.The estimates from the survey are subject tosampling errors, that is, errors arising fromthe fact that the estimates each month arebased on information from a sample ratherthan the whole population. In addition, as inany survey work, the results are subject toerrors made in the field and to errors that occurin the process of compilation.

Classification errors in labor force surveysmay be particularly large in the case of personswith marginal attachments to the labor force.These errors may be caused by interviewers,respondents, or both, or may arise from faultyquestionnaire design. In spite of a continuousquality control program, interviewers may notalways ask the questions in the prescribedfashion. To the extent that varying the word-ing of the question causes differences in re-sponse, errors or lack of uniformity in thestatistics may result. Similarly, the data arelimited by the adequacy of the information pos-sessed by the respondent and the willingness toreport accurately.

The estimates from the survey are subject tovarious 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 typicalmonth because of temporary absence of theoccupants, refusals to cooperate, or variousother reasons. Although an adjustment is madein weights for interviewed households to ac-count for noninterviews, they still represent apossible source of bias. Similarly, for a rela-tively few households, some of the informationis omitted because of lack of knowledge on thepart of the respondent or because the inter-viewer forgot to ask certain questions or recordthe answers. In processing the completed ques-


tionnaires, entries usually are supplied foromitted items on the basis of the distributionin these items for persons of similar charac-teristics.

2. Independent Population Estimates. Theindependent population estimates used in theestimation procedure may also provide a sourceof error, although on balance their use sub-stantially improves the statistical reliabilityof many of the important figures. (See discus-sion under "Ratio Estimates," p. 12.) Errorsmay arise in the independent population esti-mates because of underenumeration of certainpopulation groups or errors in age reportingin the last census (which serves as the base forthe estimates) or similar problems in the com-ponents of population change (mortality, im-migration, etc.) since that date.

3. Processing errors. Although there is aquality control program on coding and a closecontrol on all other phases of processing andtabulation of the returns, some processingerrors are almost inevitable in a substantialstatistical operation of this type. However, thenet error arising from processing is probablyfairly negligible.

Measuring the Accuracy of Results. Modernsampling 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 fordetermining the effect of response variabilityin the CPS. A measure of sampling variabilityindicates the range of difference that may beexpected because only a sample of the popula-tion is surveyed. A measure of response ,varia-bility indicates the range of difference thatmay be expected as a result of compensatingtypes of errors arising from practices of differ-ent interviewers and the replies of respond-

ents; these would tend to cancel out in anenumeration of a large enough population. Inpractice, these two sources of errorsamplingand response variability, as defined aboveare estimated jointly from the results of thesurvey. The computations, however, do not, in-corporate the effect of response bias, that is,any systematic errors of responsefor ex-ample, those that would occur if, by and large,respondents tended to overstate hours worked.Response biases occur in the same way in acomplete census as in a sample, and, in fact,they may be smaller in a well-conducted samplesurvey because for the relatively small sampleit is feasible to pay the price necessary tocollect the information more skillfully.

Estimates of sampling and response-varia-bility combined are provided in Employmentand Earnings and in other reports based on CPSdata, thus permitting the user to take thisfactor into account in interpreting the data. Ingeneral, the smaller figures and small differ-ences between figures are subject to relativelylarge variation and should be interpreted withcaution. The avail ability of the high-speed elec-tronic computer makes possible considerablymore detailed estimates than were possibleearlier.

Estimation of response bias is one of themost - difficult aspects of survey and censuswork. Systematic studies on this subject arenow an integral part of the CPS,' but in manyinstances available techniques are not suffi-ciently precise to provide satisfactory estimatesof response biases. Considerable experimenta-tion is in progress with the aim of developingmore precise measurements and improving theoverall 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 ProgramSome Notes and Discussion" (March1963).

Technical References

Number1. 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 abrief history of their development, an evaluation of current concepts and techniques, and rec-

19


ommendations for further research and improvements, several of which were inaugurated inJanuary 1967.

2. U.S. Department of Commerce, Bureau of the Census, Technical Paper No. 6, "Tice CurrentPopulation Survey Reinterview ProgramSome 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 otherstudies.

3. U.S. Department of Commerce, Bureau of the Census, Technical Paper No. 7, "The CurrentPopulation SurveyA Report on Methodology" (1963).

A brief history of the Current Population Survey (CPS) from its inception (1943) to thepresent. 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 andthe resultant gain in precision.

4. 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 CurrentPopt: lotion Survey, BLS Report 313 and Current Population Reports, Series P-23, No. 22(19671.

A :oncise description of the methodology in obtaining labor force information from samplehouseittolds. Labor force concepts and definitions are set forth. The adequacy of labor forcedata and quality controls are discussed, and major improvements in the Current PopulationSurvey are listed chronologically.

5. 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 forcestatistics from the Current Population Survey.

-JOHN E. BREGGER

Chapter 2. Employment, Hours, and Earnings

Background

The first monthly studies of employment andpayrolls by the Bureau of Labor Statistics(BLS) began in October 1915 and covered fourmanufacturing industries. Before that year, theprincipal sources of employment data in theUnited States were the census surveysthedecennial Census of Population, and beginningin 1899, the quinquennial Census of Manufac-tures. No regular employment data had beencompiled between the Census dates.

By November 1916, the BLS program hadbeen expanded to cover 13 manufacturing in-dustries, and this number remained unchangeduntil 1922. The depression of 1921 directed at-tention to the importance of current employ-ment- statistics, and in 1922 Congress grantedadditional 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 inducedCongress to provide additional appropriationsfor the program. In the next 4 years, 38 manu-facturing and 15 nonmanufacturing industrieswere added to the list of industries for whichthe Bureau published monthly information onemployment and payrolls.

The onset of the Great Depression in 1930and the deepening economic crisis impelledPresident Hoover to appoint an Advisory Com-mittee on Employment Statistics to study theneed for expanded data in this field. The Com-mittee made its report in the spring of 1931with a number of recommendations for exten-sion of the program. The most important ofthese called for the development of series onhours 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 andaverage weekly hours for the first time werepublished 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 continuingaspect of American life., there was much contro-versy among various authorities concerningthe actual number of the unemployed. Thesediscussions pointed up the fact that no reliablemeasures of either unemployment or employ-ment existed. In the early years of the Roose-velt administration, the Secretary of Laborfrequently referred to the value of the Bureau'semployment estimates as an indirect measureof unemployment. This interest stimulatedefforts to develop comprehensive estimates oftotal wage-and-salary employment in nonagri-cultural industries, and in 1936, the Bureaufirst published such a figure.

The preparation of these estimates of overallemployment 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 derivedfrom a sample of establishments might be fairlyaccurate for short periods, but over long inter-vals of time the series would not represent thetrue movement of employment, unless theywere 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 ofManufacturers, through that of 1939. For non-manufacturing industries, benchmarks weredeveloped from various sources, including theCensuses of Business taken at intervals from1929 on.

From 1915 to the beginning of World War II,interest in employment statistics for States andareas was growing constantly. Even before theBureau of Labor Statistics entered the field in

17

18 BLS HANDBOOK

1915, three States (Massachusetts, New York,and New Jersey) were preparing employmentstatistics. As early as 1915, New York andWisconsin had entered into "co-operative"agreements with the Bureau, whereby sampledata collected by the State agency were to beused jointly with the Bureau of Labor Statisticsfor the preparation of State and national series.By 1928, five other States had entered into suchcompacts, and another five were added by 1936.Over the years, the amount of published dataon employment and payrolls for States andareas underwent a constant expansion. In 1940,estimates of total nonagricultural employmentfor all 48 States and the District of Columbiawere published for the first time.

The onset of World War II in 1939, followedby the entry of the United States after the as-sault on Pearl Harbor in December 1941, placedadditional 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 relatedsubjects which were being prepared by theBLS, the Bureau of the Census, and the agen-cies administering the emerging social securityprograms. While most improvements had toawait the end of the war, several importantadvances took place during those years.

The most far reaching decision was to use asemployment benchmarks the data on employ-ment collected primarily for administrativepurposes by the newly organized social insur-ance programs. Tabulations of such materialsbecame 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 notcovered by the Census of Manufacturers andBusiness, respectively, and they were availableannually. After 1939, these were taken onlyat 5-year intervals.

As the unemployment insurance program de-veloped, the feeling grew that the proper placeto estimate State and area employment was inthe State agencies rather than in Washington.By 1949, all States had joined the system, andsince that year the industry employment sta-tistics program has been a fully integrated

OF METHODS

Federal-State project which provides employ-ment, hours, and earnings information on anational, State, and area basis in considerableindustrial detail. This cooperative program hasas its formal base of authority a Congressionalact of July 7, 1930 (ch. 873, 46 Stat. 1019; 29U.S.C. 2). In 1971 cooperative arrangementswere in effect within 44 States and the Districtof Columbia and with 6 State labor depart-ments.

, Description of the Survey

The Bureau of Labor Statistics cooperates incollecting data each month on employment,hours, and earnings from a sample of establish-ments in all nonagricultural activities includinggovernment. In 1970, this sample included over165,000 reporting units. From these data alarge number of series on employment, hours,and earnings in considerable industry detailare prepared and published monthly for theUnited States as a whole, for each of the 50States and the District of Columbia, and formost of the metropolitan areas. The data in-clude series on total employment, production ornonsupervisory worker employment, womenemployed, average hourly earnings, averageweekly hours, and average weekly overtimehours (in manufacturing). For many series,seasonally adjusted data also are published.

Concepts

An establishment is defined as an economicunit ve.tich produces goods or services, such as afactory, mine, or store. It is generally at a singlephyziical location and it is engaged predomi-nantly in one type of economic activity. Where asingle physical location 'encompasses two ormore distinct and separate activities these aretreated as separate establishments, providedthat separate payroll records are available andcertain other criteria are met.1 In the collectionof data on employment, payrolls, and man-hours, the BLS usually requests separate re-

Standard Industrial Classification Manual (Bureau of theBudget, Executive Office of the President. 1987), p. x.

Y.

is

EMPLOYMENT, HOURS, AND EARNINGS 19

ports by establishment. However, when a com-pany has more than one establishment engagedin the same activity in a geographic area, theseestablishments may be covered by a combinedreport.

Industry employment statistics published byBLS and the cooperating State agencies repre-sent the total number of persons employedeither full-time or part-time in nonagriculturalestablishments during a specified payroll period.In general, data refer to persons who workedduring, or received pay for, any part of the payperiod that includes the 12th of the month.However, at the national level, data for FederalGovernment establishments generally refer tocivilian personnel who worked on, or receivedpay for, the last day of the month, plus inter-mittent employees who worked any time duringthe month (e.g., Christmas temporary em-ployees of the postal service).

Employed persons include both permandntand temporary employees and those who areworking either full- or part-time. Workers onan establishment payroll who are on paid sickleave (when pay is received directly from theemployer), on paid holiday or paid vacation, orwho work during only a part of the specifiedpay period are counted as employed. Persons onthe 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 receivepay for any part of the specified pay period,but are not 'considered employed if they receiveno pay at all for the pay period. Since pro-prietors, the self-employed, and unpaid familyworkers do not have the status of "paid em-ployees," they are not included. Domestic work-ers in households are excluded from the data fornonagricultural establishments. Governmentemployment statistics refer to civilian em-ployees only.

The figure which includes all persons whomeet these specifications is designated "all em-ployees." Major categories of employment aredifferentiated from this overall total, primarilyto ensure the expeditious collection of currentstatistics on hours and earnings ; these groups

of employees are designated production work-ers, construction workers, r,r nonsupervisoryworkers, depending upon the industry.

In manufacturing industries, data on em-ployment, man-hours, and payrolls are collectedfor production workers. This group, in general,covers those employees, up through the level ofworking foremen, who are engaged directly inthe manufacture of the product of the establish-ment. Among the exclusions from this categoryare persons in executive and managerial posi-tions, and persons engaged in activities such asaccounting, sales, advertising, routine officework, professional and technical functions, andforce account construction." Production 'work-ers in mining are defined in a similar manner.A more detailed description of the classes ofemployees included in the production and non-production workers categories in mining andmanufacturing is shown on the facsimile ofthe BLS 790 A schedule on page 31 of thisbulletin.

In contract construction, the term construc-tion workers covers workers, up through thelevel of working foremen, who are engageddirectly on the construction project either at thesite or working in shops or yards at jobs ordi-narily performed by members of constructiontrades. Exclusions from this category includeexecutive and managerial personnel, profes-sional and technical employees, and routineoffice workers.

Data on the employment, man-hours, andpayrolls 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 theservice industries. Nonsupervisory workersinclude most employees except those in topexecutive and managerial positions. (See fac-simile of BLS 790 E, the reporting form forwholesale and retail trade, p. 32.)

The series on hours and earnings is based onreports of gross payroll and corresponding paid

,Force-accnunt construction is construction work performedby an establishment, primarily engaged in some businessother than construction. for its own account and use and byits own employees.


man-hours for production workers.8 To derivethese 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 receivedpay for, any part of the pay period includingthe 12thnf the month.

2. Total gross payrolls for production work-ers before deductions for old-age and unemploy-ment insurance, group insurance, withholdingtax, bonds, and union dues. The payroll figuresalso include pay for overtime, shift premiums,holidays, vacations, and sick leave paid directlyby the firm to employees for the pay periodreported. They exclude bonuses (unless earnedand paid regularly each pay period) or otherpay not earned in the pay period concerned(e.g., retrocative pay). Tips and the value offree rent, fuel, meals, or other payment in kindare not included.

3. Total man-hours worked (including over-time hours) during the pay period, hours paidfor standby or reporting time, and equivalentman-hours for which employees received paydirectly from the employer for sick leave, forholidays, vacations, and other leave paid tothese employees. Overtime or other premiumpaid hours are not converted to straight-timeequivalent hours.

4. Overtime man-hours for which premiumswere paid because the hours worked were inexcess of the number of hours of either thestraight-time workday or workweek. Saturdayand Sunday hours (or 6th and 7th day hours)are included as overtime only if overtimepremiums were paid. Holiday hours worked asovertime are not included unless they are paidfor at more than the straight-time rate. Hoursfor which only shift differential, hazard, incen-tive, or similar types of premiums were paidare excluded from overtime hours.

Overtime hours data are collected only fromestablishments engaged in manufacturing in-dustries. For government organizations andprivate educational institutions, payrolls col-lected relate to all employees. Manhour dataare not collected.

3Unless otherwise indicated, the references in this chapterto production workers apply likewise to the constructionworker and nonsupervisory categories.

Industrial Classification

All national, State, and area employment,hours, and earnings series data are classifiedin accordance with the Standard IndustrialClassification Manual, Office of Managementand Budget, 1967. (See appendix D of thisbulletin for a detailed description of this sys-tem.)

Reporting establishments are classified intosignificant economic groups on the basis ofmajor product or activity as determined bythe establishments' percent of total sales orreceipts for the previous calendar year. Thisinformation is collected once each year on an"Industry Class Supplement" to the monthlyreport form. (See p. 33 for a facsimile of thisform.) All data for an establishment makingmore than one product or engaging in morethan 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 ofthe month, which is standard for all Federalagencies collecting employment data on anestablishment basis.

Data Sources

Sample Data

Each month the State agencies cooperatingwith the Bureau collect data on employment,payrolls, and man-hours from a sample of es-tablishments. The respondent extracts thesefigures from his payroll records. These dataare readily available as the employers mustmaintain such records for a variety of tax andaccounting purposes. A response analysis sur-vey of the reporting practices of a scientificallyselected sample of reporting establishments inmanufacturing industries showed that the re-ports were made out almost exclusively fromtheir payroll records. The survey also showed


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 allin the same direction. On balance, they tendedto offset each other, and the net effects of in-correct reporting were quite insignificant'

Participation in the industry employmentstatistics program is entirely voluntary on thepart of the reporters. However, in many in-dustries, particularly in manufacturing, em-ployers who have a high percentage of totalemployment in the industry supply reportsregularly, and many have done so over a longperiod of years.

Benchmark Data

An employment benchmark is defined as areagonably complete count of employment. Theestimates are adjusted periodically, annually ifpossible, to new benchmark levels. Since 1939,the basic sources of benchmark information for"all employees" have been periodic tabulationsof employment data by industry and, beginningwith 1959, by size of establishment. These arecompiled by State employment security agen-cies, according to uniform procedures specifiedby the Manpower Administration of the U.S.Department of Labor, from reports of estab-lishments covered under State unemploymentinsurance laws. The State employment securityagencies receive quarterly reports, from eachemployer subject to the laws, showing totalemployment in each month of the quarter, andtotal quarterly wages for all employees.' If theemployer has more than 50 employees and op-erates more than one establishment in a State,he is required to make separate reports foreach area (e.g., county) in which he operatesand for each establishment in different indus-

Young, Dudley E. and Goldstein, Sidney, "The BLS Em-ployment Series and Manufacturing Reporting Practices."Monthly Labor Review, November 1957, pp. 1387-1371.

n The State employment security agencies are required tosubmit tabulations of these reports to the Manpower Ad-ministration each quarter. These tabulations are due in theWashington office of MA by the middle of the sixth monthafter the end of the quarter of reference. For example, thefirst quarter tabulation, which provides the basis of the BLSbenchmarks, is due on September 15. Review and editing ofthese tabulations and preparation of national summariesfrom them requires several months additional work on thepart of both BLS and MA before the benchmark is completed.

tries. Employment is reported for the payperiod of the month including the 12th, andreports are classified industrially according tothe Standard Industrial Classification. TheState employment security agencies cooperateclosely with the Bureau of Labor Statistics inthe assignments of industry classifications, sothere is a high degree of uniformity in thisrespect between the benchmark' and sampledata.

In 1970, unemployment insurance data ac-counted for three-fourths of the total bench-mark. For the group of establishments exemptfrom State unemployment insurance laws be-cause of their small size ° and for certainclasses of nonprofit institutions, the data usedare those provided by the national old-age in-surance program administered by the SocialSecurity Administration of the U.S. Depart-ment cf Health, Education, and Welfare.

For industries not covered by either of thetwo programs, benchmarks are compiled froma number of special sources. The most im-portant of these are the Interstate CommerceCommission (interstate railroads), the Ameri-can Hospital Association (private nonprofithospitals), 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 GovernmentsDivision of the Bureal of the Census (Stateand local government):

Special efforts are made to classify establish-ments into the same industrial groupings forbenchmark purposes as they are for monthlyreporting. Wherever possible, employment forthe standard midmonth pay period for March isused as the benchmark.

In 1970 the unemp'oyment insurance laws of 21 States andthe District of Columbia covered ail employers of 1 or moreworkers. 3 States covered employers of 3 workers or more.and the remaining 28 States covered employers of 4 workersor more. Beginning in 1972, all State unemployment insurancelaws should cover employ.:rs of 1 or more workers.

'For a more detailed description of the benchmark, seeArmknecht, Paul A., "BLS Establishment Estimates Revised toMarch 1970 Benchmark Levels," Employment and Earnings,September 1971, pp. 18-30.

25


Collection Methods

The primary collection of the current sampledata is conducted by State agencies which havecooperative agreements with the BLS. In mostStates, this is the employment security agency,affiliated with the Manpower Administration,the organization which administers the Stateunemployment insurance program. In a fewcases the State department of labor acts as theagency. The agencies mail schedules to a sam-ple of establishments in the States each month.A "shuttle" schedule is used (BLS form 790series) ; that is, one which is submitted eachmonth in the calendar year by the respondent,edited by the State agency, and returned to therespondent for use again the following month.The State agency uses the information pro-vided on the forms to develop State and areaestimates of employment, hours, and earnings,and then forwards the data, either on theschedules themselves or in machine readableform, to the Washington, D.C., office of theBureau of Labor Statistics, where they areused to prepare estimates at the national level.

The shuttle schedule has been used in thisprogram since 1930, but there have been sub-stantial changes in its design and in the datacollected over the period. All aspects of theschedule, its format, the wording of the re-quested items and definitions, and the conceptsembodied therein have been subjected to a con-tinuing and intensive review over the entireperiod, not only by the staff of BLS and of theState agencies, but also by other governmentagencies and by numerous persons in privatebusiness and labor organizations. The reportforms are not exactly alike for every industry,but most of them request data on total employ-ment, number of women employees, number ofproduction workers (in manufacturing andmining), construction workers (in contractconstruction industries), or nonsupervisoryworkers (in other nonmanufacturing industries), and, for these workers, data on payroll,paid man-hours, and for manufacturing, over-time man-hours. The schedule contains detailedinstructions and definitions for the reports.There are several variants designed to meetthe specific problems of different industries.

(See facsimiles of BLS 790 A and, BLS 790 E,pp. 31 and 32.)

The technical characateristics of the shuttleschedule are particularly important in main-taining continuity and consistency in reportingfrom month to month. The design exhibitsautomatically the trend of the reported dataduring the year covered by the schedule, andtherefore, the relationship of the current figureto the data for the previous month. The sched-ule also has operational advantages; for ex-ample, accuracy and economy are obtained byentering identifying codes and the address ofthe respondent only once a year.

All schedules are carefully edited by theState agencies each month to make sure thatthe data are correctly reported and that theyare consistent within themselves, with the datareported by the establishment in earliermonths, and with those reported by other es-tablishments in their industry. This editingprocess is carried out in accordance with a de-tailed manual of instructions prepared by theBureau of Labor Statistics. When the reportsare sent to Washington, they are screened byuse of an electronic computer to detect process-ing errors and reporting errors which mayhave escaped the first editing. Questionablecases discovered at any stage of the editingprocess are returned, if necessary, to the re-spondent for review and correction.8

Sampling

Sampling is used by BLS in its industryemployment statistics program for collectingdata in most industries, since full coveragewould be prohibitively costly and time consum-ing. The sampling plan for the program must:(a) provide for the preparation of reliablemonthly estimates of employment, hours ofwork, and weekly and hourly earnings whichcan be published promptly and regularly ; (b)through a single general system, yield con-siderable industry detail for metropolitanareas, States, and the Nation ; and (c) be ap-

Mendelssohn, Rudolph C. "Machine Methods in Employ-ment Statistics," Monthly Labor Review, May 1955, pp. 507-569.

EMPLOYMENT, HOURS, AND EARNINGS

propriate for the existing framework of op-erating procedures, administrative practices,resource availability, and other institutionalcharacateristics of the program.

In developing the sample design, the universeof establishments was stratified first by in-dustry and within each industry by size ofestablishment in terms of employment, usingsix standard size classes. Within each industry,an optimum allocation design was obtain& bysampling with probabilit" proportionate toaverage size of establishment within each ofthe strata. A total size of sample necessary toproduce satisfactory estimates of employmenthad to be distributed among the size-class cellson the basis of average employment per estab-lishment in each cell. In practice, this is equiva-lent to distributing the predetermined totalnumber of establishments required in the sam-ple among the cells on the basis of the ratioof employment in each cell to total employmentin the industry. Within each stratum, the sam-ple members are selected at random.

Under this type of design, large eetablisi:ments fall into the sample with ipitainty. Innearly all industries, establiAnw,4,,,s with 250or more employees are inclw3e:i in the samplewith certainty and in many industries the cut-off is lower. The sizes of the samples for thevarious industries were determined empiricallyon the basis of experience and of cost consid-erations. In a manufacturing industry in whicha high proportion of total employment wasconcentrated in a relatively few large establish-ments, a high percentage of total employmenthad to be included in the sample. Consequently,the sample design for such industries providesfor a complete census of the larger establish-ments with only a few chosen from among thesmaller establishments, or none at all if theconcentration of employment in the larger es-tablishments is great enough. On the otherhand, in an industry where a large proportionof total employment is in small establishments,the sample design calls for inclusion of all largeestablishments, and also for a substantial num-ber of the smaller establishments. Many in-dustries in the trade and service divisions fallinto this category. In order to keep the sampleto a size which can be handled with available

23

resources, it is necessary to accept samples inthese divisions with a smaller proportion ofuniverse employment than is the case for mostmanufacturing industries. Since individual es-tablishments in these nonmanufacturing indus-tries generally show less fluctuation fromregular 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 nationalprogram, provides a technical frameworkwithin which State and area sample designscan be determined. Since the estimates forStates and areas generally are not preparedat the same degree of industry detail as thenational estimates, the national design usuallyprovides sufficient reports for the preparationof State and area estimates.°

Estimating Procedures

Employment

The "all-employee!' estimates by industry arebased on reasonably complete employmentcounts or "benchmarks." To obtain employmentestimates 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 thecell.

2. For each cell, the ratio of employment inone 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, theestimate for each month is obtained by multi-plying the estimate for the previous month bythe link relative for the current month.

Application of the estimating procedure inpreparing a series is illustrated by the follow-ing example : Assume that total employment

For the national sample, additional reports needed forState and area samples are added to those required by thenational design.

27


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 theAugust estimate, the change for identical es-tablishments reported in the July-August sam-ple is applied to the July estimate:

50,000x 26,000 (or 1.04) =52,00025,000

This procedure for estimating current em-ployment is known as the benchmark andlink-relative technique. It is an efficient tech-nique, taking advantage of a reliable completecount of employment and of the high correla-tion between levelo of employment in successivemonths in identical establishments.

In addition to estimates of total employmentby industry, the Bureau publishes data onproduction, construction, or nonsupervisoryworker employment. For this purpose, thesample ratio for the current month of produc-tion workers to total employment is used. Forexample, the 60 sample firms which had 26,000employees in August, reported an August pro-duction-worker figure of 19,500 resulting in a

500ratio of 19,or .750. Using this ratio, pro-26,000

action-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."

The estimates for each type of series (all em-ployees, production workers, and women em-ployees) for individual estimating cells aresummed to obtain the corresponding totals forbroader industry groupings and divisions.

Appropriate revisions, based on new bench-marks, are introduced into the employmentseries as required to correct for classificationchanges and for deviations resulting from theuse of sample trends, In general, the bench-mark month is March. The employment esti-mates which had been published previously forthat month are compared with the new bench-

1^ If permanent changes in the composition of the sampletake place, the "production-workers, all-employee" ratios andthe "women" ratios calculated from the sample are modifiedby a wedging technique described in this chapter under"Hours and Earnings." (See p. 25.)

mark data. The amount of adjustment in thepublished employment information is indicatedby this comparison. The all-employee series,for months between the current and the lastpreceding benchmark, are adjusted by wedgingor tapering out the difference between the cur-rent benchmark and the estimate for the bench-mark month back from the current benchmarkto the last previous benchmark. This differenceis assumed to have accumulated at a regularrate. The series for months subsequent to thebenchmark month are revised by projectingthe level of the new benchmark by the trend ofthe unadjusted series.

A comparison of the amounts of the re-visions made since 1966, is presented in table 1.

Table 1. Nonagricultural employment estimates, byindustry division, as a percentage of the benchmarkfor recent years

Industry division 1966 1967 1068 1969 1970

Total .. ._ ..... _____ 99.9 100.0 100.4 99.8 100.0Mining 100.5 99.5 101.7 101.5 100.0

Contract construction ._...._.... 99.7 101.6 995 99.0 100.1Manufacturing 99.4 99.5 99.8 09.8 100.1Transportation and public

utilities . ............... _____ ...... _. 99.7 99.8 100.7 100.4 100.0Wholesale and retail trade 100.1 100.7 100.3 100.0 100.1Finance. insurance, and

real estate 99.5 100.2 99.2 100.0 100.3Services .. ...... _ .... ______ ...... _. 100.3 99.8 992 99.1 99.6Government 100.0 100.0 102.8 100.1 100.2

Hours and earnings

Independent benchmarks are not availablefor the hours and earnings series. Conse-quently, the levels shown are derived from theBLS reporting sample.

Since 1959, when benchmark data stratifiedby employment size became available throughthe employment security system, estimates ofemployment, hours, and earnings have beenprepared by a cell structure which makes useof 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 becombined into a maximum of three size classesfor the purposes of preparing current estimatesof hours and earnings, when stratification by

'i


size is needed." Whenever a new nationalbenchmark becomes available, national esti-mates of average weekly hours and averagehourly earnings using eight size strata andfour regional strata (Northeast, North Cen-tral, South, and West) are prepared. Theseestimates are used as a standard against whichthe published averages are compared. In somecases, this comparison indicates that somemodification of the pattern of stratification isneeded. If this is the case, a change is intro-duced into the estimating structure at the timeof the next benchmark revision.

a. Average weekly hours and gross hourlyearnings. To obtain average weekly hours foran individual estimating cell, the sum of theman-hour totals reported by the plants classi-fied in that cell is divided by the total numberof production workers reported for the sameestablishments. Similarly, in computing aver-age hourly earnings, the reported payroll totalis divided by the reported man-hour total.

The first ratio estimates of average hourlyearnings and average weekly hours are modi-fied at the estimating cell level by a wedgingtechnique designed to compensate for changesin the sample arising mainly from the volun-tary character of the reporting.

For example, a first estimate of averagehourly earnings for the current month, (.11, isobtained from aggregates from a matchedsample of establishments reporting in the cur-rent and previous month. Similarly an estimateof average hourly earnings, U., for the previ-ous month is calculated from the same matchedsample. Hence, Ur-U is a measure of absolutechange between the 2 months.

11 These combinations of size classes have been made be-cause of operational economies. In particular, the preliminaryestimates are based on less than fuli samples, and if theestimation of preliminary estimates was attempted using thefull stratification pattern, there would be a number of cellsfor which there were no samples. Experiments and tests overseveral years indicate that estimates of hours and earningsprepared from the BLS sample using a maximum of threesize strata generally do not differ significantly from thosecomputed with four size strata or more.

12 if the difference between the estimate and the averagecomputed from the sample (17.--U,,), is too great, the sampleaverage is accepted once it has been established that thedifference is due to a permanent change in the composition ofthe sample, and the series is regarded as discontinuous atthat point. In general, a difference greater than 3 percent isconsidered as defining a discontinuity or "break."

Note is then taken of the published estimateof average hourly earnings for the previousmonth, say V.. Because the panel of establish-ments reporting in the sample is not absolutelyfixed from month to month, there may be differ-ences between V. and U0.12 A final figure forthe current month is obtained by making useof both pieces of information ; the estimate is

V1 =(0.9 V. + 0.1 U0)+ (U1U.)The procedure, reflected in this last equation,accepts the advantage of continuity from theuse of the matched sample, and at the sametime tapers or wedges the published estimatetoward the level of the latest sample average.The same procedure is used to adjust theproduction-worker all-employee-ratio and the

Table 2. Number of industries for which "Primary"series are published under the BLS Industry Employ-ment Statistics Programemployment, hours, andearnings, January 1971

AllIndustry division I em-

ployees

Pro-duction

workers%Women

Hoursand

cam-tugs 2

Averageovertime

hours

Total _Mining _____Con tract.

construction ....Manufacturing ..Transportation

and publicutilities ........

TradeFinance,

insurance, andreal estate ....

Services .... ....Government ..Total private .....

....

Total non-agrieulture

41712

27811

2536

152315

1

1

38211

11277

1632

95

1

1

4029

11278

2135

1522

... .1

1

36411

11277

1932

95

1....-

150._.....148

......

*4_

_Production workers in manufacturing and mining: con-

struction workers in contract construction; nonsupervisoryworkers, all other divisions.

Average hourly earnings,average weekly earnings.

average weekly hours, and

ratios for women with regard to changes in thecomposition of the sample between successivemonths.

Weekly hours and hourly earnings for indus-tries and groups above the basic estimatingcell level are weighted averages of the figuresfor component cells and industries. The averageweekly hours for each estimating cell aremultiplied by the corresponding estimate ofproduction-worker employment, to derive ag-gregate man-hours. Payroll aggregates are theproduct of aggregate man-hours and averagehourly earnings. Payroll and man-hour aggre-

29'


gates for industry groups and divisions repre-sent the sum of aggregates for componentindustries.

Average weekly hours for industry groupsare obtained by dividing the man-hour aggre-gates by the corresponding production-workeremployment estimates. Average hourly earn-ings for groups are computed by dividing thepayroll aggregates by the man-hour aggre-gates. This method is equivalent to weightingweekly hours by estimated universe production-worker employment and hourly earnings byestimated universe man-hours.

For all levels, from individual estimatingcells to major industry divisions, averageweekly earnings are computed by multiplyingaverage hourly earnings by average weeklyhours.

b. Overtime man-hours. To obtain averageweekly overtime hours in manufacturing in-dustries, the sum of the overtime man-hoursreported 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 incomebrackets were subject to Federal income andsocial security taxes, gross average weeklyearnings were a satisfactory measure of trendsin weekly earnings available for spending.After Federal income taxes began to affect thespendable earnings of an appreciable numberof workers, a method was developed for ap-proximating spendable earnings by deductingthese taxes from gross earnings.'s

The amount of individual income tax liabilitydepends on the number of dependents sup-ported by a worker as well as on the level ofhis gross income. Spendable earnings for work-ers by major industry division are computedand published for a worker with no dependentsand a worker with three dependents.

Gross and spendable weekly earnings alsoare computed and published in terms of 1967dollars, to give an approximate measure ofchanges in "real" gross and spendable weekly

ln Utter, Carol, "The Sp lendable Earnings Series: A TechnicalNote on its Calculation," Employment and Earnings andMonthly Report on the Labor Force, February 1969, pp. 6-13.

30.

earnings, that is, in purchasing power sincethat base period. This series is computed bydividing the weekly earnings average (in cur-rent dollars) by the BLS Consumer Price Indexfor the same month.

d. Average hourly earnings, excluding over-time, in manufacturing are computed by divid-ing the total production-worker payroll for theindustry group by the sum of total production-worker man-hours and one-half of total over-time man-hours, which is equivalent to pay-rolls divided by straight-time man-hours. Thismethod excludes overtime earnings at 11/2

times the straight-time rates; no further ad-justment is made for other premium paymentprovisions.

e. Indexes of aggregate weekly man-hoursand payrolls are prepared by dividing the cur-rent month's aggregates by the average for1967.

Reliability of estimates

Although the relatively large size of the BLSestablishment sample assures a high degree ofaccuracy, the estimates derived from it maydiffer from the figures that would be obtainedif it were possible to take a complete censususing the same schedules and procedures. Asdiscussed previously a link relative techniqueis used to estimate employment. This requiresthe use of the previous month's estimate asthe base in computing the current month's esti-mate. Thus, small sampling and responseerrors may cumulate over several months. Toremove this accumulated error, the estimatesare adjusted annually to new benchmarks. Inaddition to taking account of sampling andresponse. errors, the benchmark revision ad-justs the estimates for changes in the indus-trial classification of individual establishments(resulting from changes in their product whichare not reflected in the levels of estimates untilthe data are adjusted to new benchmarks). Infact; at the more detailed industry levels, par-ticularly within manufaCturing, changes inclassification are the major cause of benchmarkadjustments. Another cause of differences, gen-erally minor, arises from improvements in thequality of the benchmark data.


One measure of the reliability of the employ-ment estimates for individual industries is theroot-mean-square error (RMSE). The measureis the standard deviation adjusted for the biasin estimates.

RMSE=,/ (Standard Deviation) 2+ (Bias)2If the bias is small, the chances are about 2 outof 3 that an estimate from the sample woulddiffer from its benchmark by less than the root-mean-square error. The chances are about 19out of 20 that the difference would be less thantwice the root-mean-square error.

The hours and earnings estimates for cellsare not subject to benchmark revisions, al-though the broader groupings may be affectedslightly by changes in employment weights.The hours and earnings estimated, however,are subject to sampling errors which may beexpressed as relative errors of the estimates.(A relative error is a standard error expressedas a percent of the estimate.) Measures of root-mean-square errors for employment estimatesand relative errors for hours and earnings esti-mates are provided in the "Technical Note" ofEmployment and Earnings.

Seasonally adjusted series

Many economic statistics, including employ-ment and average weekly hours, reflect aregularly recurring seasonal movement whichcan be measured on the basis of past experi-ence. By eliminating that part of the changewhich can be ascribed to usual seasonal varia-tion, it is possible to observe the cyclical andother nonseasonal movements in these series.Seasonally adjusted series are published reg-ularly for selected employment hours, andearnings series.

The seasonal adjustment method used forthese series is an adaptation of the standardratio-to-moving average method, with a pro-vision for "moving" adjustment factors to takeaccount of changing seasonal patterns. A de-tailed description of the method is given inappendix A of this bulletin.

The seasonally adjusted series on gross aver-age weekly hours, average overtime hours andaverage hourly earnings are computed by ap-plying factors directly to the corresponding

unadjusted series, but seasonally adjusted em-ployment totals for all employees and produc-tion workers by industry divisions are obtainedby summing the seasonally adjusted data forcomponent industries. Selected seasonally ad-justed series also are prepared for aggregateweekly man-hours.

Presentation and Uses

At the national level, the program produceseach 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 computeddirectly from the sample and benchmark data.

Table 3 indicates the "special" series whichare obtained from the primary series by ap-plication of special adjustments, while table 4lists the seasonally adjusted series by type andindustry division.

In addition to the series published on a cur-rent monthly basis, a single annual figure foremployment in March of each year (based onbenchmarks) is published for a number of in-dustries for which monthly estimates do notcurrently meet established standards for pub-lication. In 1971, following revision to the 1970benchmark, data for 239 such industries werepublished.

In April 1971, employment and hours andearnings statistics were available for 50 States,the District of Columbia, and 211 areas. Ap-proximately 8,200 employment series and hoursand earnings series for about 3,400 industrieswere published for these States and areas bythe State agencies. The employment seriesusually covered total nonagricultural employ-ment, major industry divisions (e.g., contractconstruction, manufacturing), and major in-dustry groups (e.g., textile mill products,transportation equipment., retail trade) foreach State and area. Additional industry detailfrequently is provided for the larger States andareas, particularly for industries which arelocally important in the various jurisdictions.

The series on employment, and hours andearnings appear in several BLS publications.

31


Table 3. Number of industries for which special seriesare published under the BLS Industry EmploymentStatistics Program--employnient, hours, and earnings,January 1971

Industry division

Index ofaggre-gate

weeklyman-hours

Index ofuggre-gate

weeklypay-rolls

Spend-able

averageweeklyearn-ings 1

Crossweeklyearn-ings

(1967dollars)

Averagehourlyearn-ings

(exelud-ing

over-time)

Totalprivate

Mining, con-struction andmanufacturing I

1 1

_Mining .... ........ 1 1 1 1 -----Contract con-

struction .. 1 1 1 1Manufacturing 24 1 1 1 23Transportation

and publicutilities 1 1

Trade ..... ____ 1 1 --Finance, insur-

ance and realestate 1 1 _

Services 1 1 --In current and 1967 dollars.

The summary data are first publishedmonth in a press release which contains pre-liminary national estimates of nonagriculturalemployment, weekly hours, and gross averageweekly and hourly earnings in the precedingmonth, for major industry categories. The re-lease also includes seasonally adjusted data onemployment, average weekly hours, and aver-age overtime hours. The preliminary estimatesare based on tabulations of data for less thanthe full sample to permit early release of fig-ures. This release normally is issued 3 weeksafter the week of reference for the data. Thepress release also includes a brief analysis ofcurrent trends in employment, hours, and earn-ings, pointing up current developments as com-pared with those for the previous month andthe same month in the preceding years.

National estimates in the detail described intables 2, 3, and 4 are published in the monthlyreport Employment and Earnings. This pub-lication is issued about 5 weeks after the weekof reference. Employment data for total non-agricultural employment and for the majorindustry divisions, as well as hours and earn-ings for all manufacturing, are published forStates and areas in Employment and Earnings1 month later than those for the Nation. Specialarticles analyze long-term economic movementsor describe technical developments in the pro-

each

3

gram. Many of the national series are repub-lished in the Monthly Labor Review with datashown for each series for the most recent 13months.

Following each benchmark revision, an his-torical volume called Employment and Earn-ings, United States is published. This provideshistorical data, monthly and annual averages,from the beginning date of each series, in afew instances as far back as 1909. A companionvolume, Employment and Earnings, States andAreas, provides historical data (annual aver-ages) on all employees and on production-worker hours and earnings series published byState agencies for States and areas back tothe beginning of these series, in some instancesto 1939. This volume is published annually. De-tailed industry rates are available monthly inreleases published by the cooperating Stateagencies.

The data are disseminated also through thepublications of many other Federal agencies;e.g., the Department of Commerce, the Boardof Governors of the Federal Reserve Systemand the Council of Economic Advisors repub-lish all or part of the data. They are also reg-ularly republished in summary form or forspecific industries in many trade associationjournals, 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 laborunions and industry, as well as others responsi-ble for analyzing business conditions, use thetrends reflected in these particular statistics aseconomic indicators. The average weekly hoursseries are utilized as lead indicators of swingsin the business cycle. Labor economists andother social scientists find these series to bean important indicator of the Nation's eco-nomic activity, as well as a measure of thewell-being of the millions of Americans whodepend on salaries and wages. Industrialgrowth and progress may be assessed by usingthe employment and hours series in conjunctionwith other economic data to yield measures ofproductivity.

Analysts study employment trends to detect


Table 4. Number of seasonally adjusted series published under the BLS Industry Employment Statistics Pro-gramemployment, hours, and earnings, January 1971

Industry divisionSeasonally adjusted series

Allem-

ployeesPro-

ductionworkers

Indexof

employ-ment

I Averageweeklyearn-ings

Averagehourlyearn-ings

Averageweeklyhours

Manhourindex

Averageover-timehours

Total nonagriculture 1 _ I ._ _Total private _ _ 1 I I _

Mining, construction, and manufacturing .Mining 1

1

1_

1 1 1

1

1

Contract construction 1 1 1 1 1 1Manufacturing 24 24 1 1 24 24 3Transportation and public utilities ______. 1 _ 1 1 1 _Trade 3 _ 3 1 3Finance, insurance, and real estate 1 _ 1 I 1Services 1 _ 1 1 1Government 1 _ 1 _

changes in industrial structure, and to observegrowth and decline proclivities of individualindustries. They also are used in the Bureau'sOccupational Outlook program as a basis forprojection of future trends.

Executives use the employment, earnings,and hours data for guidance in plant location,sales, and purchases. Also, firms negotiatinglong-term supply or construction contractsoften utilize series on average hourly earningsas an aid in reaching an equitable agreement ;"escalation clauses" may be included in thecontracts, which permit an increase or a lower-ing of the settlement price depending on themovement of average hourly earnings in aselected industry." Wide need has been demon-strated by both labor and business for industryseries on hourly earnings and weekly hours, toprovide a basis for labor-management negotia-tions. They not only furnish current and his-torical information on a given industry butprovide comparative data on related industries.

Limitations

Employment

Total employment in nonagricultural estab-lishments "rom the "payroll" survey is notdirectly comparable with the Bureau's esti-mates of the number of persons employed innonagricultural industries, obtained from themonthly "household" survey." The "payroll"

"Nelson, Darrell, "BLS Earnings Statistics for Use inEscalation Agreements," Employment and Earnings andMonthly Report on the Labor Force, March 1968, pp. 18-21.

" See chapter 1 for a description of this survey.

series excludes unpaid family workers, domes-tic servants in private homes, proprietors andother self-employed persons, all of whom arecovered by the household survey. Moreover,the "payroll" series counts a person employedby two or more establishments at each place ofemployment, while the "household" surveycounts him only once, and classifies him accord-ing to his single major activity. Certain per-sons on unpaid leave are counted as employedunder the "household" survey, but are not in-cluded in the employment count derived fromthe "payroll" series. In addition to these differ-ences in concept and scope, the surveys employdifferent collection and estimating techniques.Therefore, although each survey measureschanges in employment, direct comparabilityshould not be expected. However, over time, thetrends are similar. The household survey placesits primary emphasis on the employment statusof individuals and also provides a great dealof 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 thegeographic distribution of employment. Theestablishment survey, while providing limitedinformation on personal characteristics ofworkers, is an excellent vehicle for obtainingthese detailed industrial and geographic data,and in addition, it provides hours and earningsinformation which is directly related to the em-ployment figures. The payroll and householdsurveys therefore may be regarded as comple-mentary.

Employment estimates derived by the Bu-

30 BLS HANDBOOK

reau of the Census from its quinquennialcensuses and from the annual sample surveysof manufacturing establishments may differfrom BLS employment statistics. The most im-portant reason for difference stems from thedegree to which multiproduct establishmentsfile separate or combined reports in one surveybut not the other, which m...y result in differentindustrial classification of employment. Thereis also a significant difference at the more de-tailed industry levels, since Census classifiesauxiliary units and central and district ad-ministrative units on the basis of the mostappropriate 2-digit major group, while BLScodes these units to the most appropriate4-digit industry. For broad categories, how-ever, the two surveys do show similar levelsand trends.

Hours and earnings

The workweek information relates to aver-age hours paid for, which differ from scheduledhours or hours worked. Average weekly hoursreflect the effects of such factors as absentee-ism, labor turnover, part-time work, andstrikes.

The gross average hourly earnings seriesreflect actual earnings of workers, includingpremium pay. They differ from wage, rates,which are the amounts stipulated for a givenunit of work or time. Gross average hourlyearnings do not represent total labor costs perman-hour for the employer, for they excluderetroactive payments and irregular bonuses,various welfare benefits, and the employer'sshare of payroll taxes. Earnings for thoseemployees not covered under the productionworker and nonsupervisory-employee cate-gories are, of course, not reflected in theestimates.

The series on spendable weekly earningsmeasure the net earnings of workers who earnthe average gross weekly earnings, have thespecified number of dependents, and take thestandard deductions for Federal income taxpurposes. Spendable earnings reflect deductionsonly for Federal income and social security

34

OF METHODS

taxes (calculated on the basis of total annualliabilities), and thus represent only a roughapproximation of disposable earnings." Theydo not take into account payroll deductions forsuch purposes as State income taxes, uniondues, or group insurance, and they do not re-flect such factors as total family income or taxdeductions above the standard amount.

The "real" earnings data (those expressed in1967 dollars), resulting from the adjustmentof gross and spendable average weekly earn-ings by means of the Bureau's Consumer PriceIndex, indicate the changes in the purchasingpower of money earnings as a result of changesin prices for consumer goods and services.These data cannot be used to measure changesin living standards as a whole, which areaffected by other factors such as total familyincome, the extension and incidence of varioussocial services and benefits, and the durationand extent of employment and unemployment.

To approximate straight-time average hourlyearnings, gross average hourly earnings areadjusted by eliminating only premium pay forovertime at the rate of time and one-half. Thus,no adjustment is made for other premium pay-ment provisions such as holiday work, late-shift work, and premium overtime rates otherthan 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 industriesin the Nation as a whole, and also for all sig-nificant industries in all States and all Stan-dard Metropolitan Statistical Areas, as definedby the Office of Management and Budget. Whilevery substantial progress toward this objectivehas been made over the years, and particularlysince the end of World War II, there remainsome important areas where the goal is yet tobe realized. Efforts constantly are being di-rected toward strengthening the sample so thatseries for employment, hours, and earnings foradditional industries may be published, andalso toward developing series for additionalstandard metropolitan areas.

16For a complete analysis of the difference between spend-able md disposable earnings, see Paul Schwab, "Two Mea-sures of Purchasing Power Contrasted," Monthly Labor Re-view, April 1971.

1

4

f

ti

BL

S C

odes

......

. --

--S

umo

Rep

ort N

o.le

d.

Form

BL

S 79

0 A

LOC

AT

ION

OF

ES

TA

IILLS

HIH

EN

TS

CO

VE

RE

D IN

TH

IS R

EP

OR

T(N

umbe

r of

est

ablis

hmen

ts)

(City

)(C

ount

y)(S

tate

)

Bef

ore

ente

ring

dat

a se

e ex

plan

atio

ns o

n ot

her

side

The

Bur

eau

of L

abor

Sta

tistic

s an

d th

e St

ate

agen

cies

coo

pera

ting

in it

s st

atis

tical

pro

gram

s w

ill h

old

all i

nfor

mat

ion

furn

ishe

d by

the

resp

onde

nt in

str

ict c

onfid

ence

.

Y S

A1

AN

DM

orns

(1)

PA

Y P

ER

IOD

Do

NO

TU

SE

LIP

OD

ALL

EM

PLO

YE

ES

PR

OD

UC

TIO

N A

ND

RE

LAT

ED

WO

RK

ER

S

Ent

erch

ange

sea

rnin

gs

YO

UR

CO

MM

EN

TS

in c

olum

n 13

the

mai

n fa

ctor

s re

spon

sibl

e fo

r si

gnifi

cant

toon

thto

nont

hin

em

ploy

men

t, av

erag

e ho

urs

wor

ked

(col

. Mea

d. 9

). a

vera

ge h

ourly

(col

. Mvc

ol. M

. etc

.. as

Indi

cate

d by

this

rep

ort.

Exa

mpi

es a

re:

wag

e re

fe J

arro

wS

trik

eO

vert

ime

Mot

t bot

tom

.,i.

Woo

Der

If an

y ge

nera

l wag

e-ra

te c

hang

es (

not I

ndiv

idua

l cha

nges

for

leng

thof

serv

ice.

mer

it, o

r pr

omot

ion)

hav

e oc

curr

ed s

ince

last

mon

th's

rep

ort n

ote

amou

nt o

f Inc

reas

e or

dee

rem

e (a

s +

2%, 5

g), t

he e

ffect

ive

date

of lh

och

ange

, and

the

appr

oxim

ate

num

ber

of p

rodu

ctio

n w

orke

rs a

ffect

ed.

(13)

Ent

erbe

ginn

ing

Aw

l end

I n

gda

tes

ofpa

ype

riod

whi

chin

clud

est h

e13

thnf

t h e

mon

th

Ent

erth

enu

mbe

rof

days

wor

ked

plu.

pai

dho

liday

. and

pai

d vu

-ca

tion

days

tor

intr

.i t

yof

prod

uct o

ilw

orke

r...

(.ve

.,,,,

Il.e

dog)

NU

MB

ER

Incl

ude

all p

erso

nsw

how

orke

ddu

ring

or r

ecei

ved

pay

for

any

pan

of p

erio

d re

gard

less

of

type

of w

ork

perf

orm

ed

Ent

er in

ther

e co

lum

n. th

e nu

mbe

r of

pro

duct

ion

and

re-

liste

d w

orke

rs w

ho w

orke

d du

ring

or r

ecei

ved

pay

for

um. p

art o

f the

per

iod

repo

rted

, the

pay

ear

ned

(bef

ore

dedu

ctio

n.),

and

all

hour

s w

orke

d or

pai

d fo

r.In

clud

epa

y an

d m

an-h

ours

for

over

time.

sic

k le

ave.

hol

iday

s.an

d vt

anat

iono

Fro

mca

ouut

tin

(2)

Thr

ough

bidu

eN)

(3)

Dur

ing

the

entir

e ps

ype

riod

(4)

Dur

ing

the

7co

nsec

utiv

e-da

y pe

riod

whi

ch in

-el

udes

the

12th

(5)

Bot

h se

xes

(7)

Wom

enon

ly

(V

Nur

se, o

rP

RO

DU

CT

ION

WO

RIL

L113

(9)

TO

TA

LP

RO

DU

CT

ION

.W

OR

EIR

PA

YR

OLL

(Osi

ff en

ds)

(II)

Tor

o.P

RO

DU

CT

ION

.W

aim

eaM

AIII

IOU

RZ

I(O

nrit

Indi

o's)

(II)

DO

NO

TU

SE

Ezo

.C

ode

(III

1970

Dec

_19

71Ja

n

Feb

Man

Apr

May

June

July

Aug

Sep

t

Oct

Nov

Dec

$-

-..

..-...

.

-

a

(Per

son

to b

e ad

dres

sed

U ti

c;es

tloct

s ar

ias

rego

rdIn

g th

is r

epor

t)(P

etiti

on)

BL

S C

odes

1i!

Stat

e1

Rep

ort N

o..

Ind.

Form

BL

S 79

0 E

Bef

ore

ente

ring

dat

a se

eex

plan

atio

ns o

n ot

her

side

LOC

AT

ION

OF

E.S

TA

BLI

SH

IHE

NT

ISI C

OV

ER

ED

IN T

HIS

RE

PO

RT

I Num

ber

at e

Nta

hlel

inle

nts1

(City

l...

.iS

tate

,T

he B

urea

u of

Lab

or S

tatis

tics

and

the

Stat

e ag

enci

es c

oope

r-at

ing

in it

s st

atis

tical

pro

gram

s w

ill h

old

all i

nfor

mat

ion

fur-

nish

ed b

y th

e re

spon

dent

in s

tric

t con

fide

nce.

YE

AR

AN

DM

ON

TII

(I)

PA

Y P

ER

IOD

DO

N 0

Tus

E

(5)

ALL

EM

PLO

YE

ES

NO

NS

UP

ER

VIS

OR

Y E

MP

LOY

EE

S

DO

NO

T U

SEE

nter

beg

inni

ng..,

el e

ii d

i n

itda

te.

ofpa

rpe

riod

whi

chin

ClU

ilt,

(lit

12th

ofth

em

onth

Eid

er th

e nu

mbe

r of

dar

eco

rked

plu

g vo

id b

oil.

d'Y

' 2"d

l'il'

,","

. "d.

i.),

fur

...P

rl'i

ii.ri:

n1s1

.....p

rr. %

iwio

ri:.,,

,e, i

n

&iv

)

NU

MB

ER

Incl

ude

all p

erso

ns w

ho w

orke

ddu

ring

or r

ecei

ved

pay

tor

any ,

part

of p

erio

d re

gard

'. ,,.

type

ot w

ork

perf

orm

ed

Ein

e. in

thew

vol

uint

ei th

ele

e of

twon

wip

eoiw

ny to

opl

occe

s w

ho w

orke

d du

ring

or r

ecei

ted

to*

for

any

pait

earn

ed (

befo

re d

efin

e.riot,.

but e

xclu

ding

con

eee

e i,d

olu.

)vat

il al

l hou

rs w

orLe

dor

pai

d to

rhol

ida

Incl

ude

pay

and

inan

.hou

rs fo

r m

ettim

e...i

ck li

ote,

. and

titr

atio

n.

UO

SIM

ISA

ION

A o

r N

ON

SE

rrItA

tA01

0tE

siru

trar

s

Fro

m(D

oll d

ates

(3)

Thr

ough

Meb

ane)

,3)

Dur

ing

the

entir

e pa

ype

riod

II)

1)ur

ine

Or

7 al

umni

..'r' w

hich

incl

ude.

the

12th

(Si

Bot

h se

xes

(71

Won

teno

nly

(A)

Nou

nto

orN

W:R

OP

ER

-T

I3O

RT

EM

MO

TE

LS

(It

NO

NsE

rER

Ilsol

tiE

urLO

tEE

PA

YR

OLL

(KS

CIU

ding

cum

-tu

bwio

ru r

epor

ted

in r

ot. I

nA)

(Om

it re

nt.)

(10)

TO

TA

LN

oNA

TF

ER

1IS

OR

1E

mrL

oVE

RM

AN

liouR

A(O

mit/

ear/

ion.

)

OH

Am

ount

of

corn

mom

orw

Ifo

nill

eras

)

DO

A)

Per

iod

in w

hich

rArn

.A1

P R Oh)

! i

H.

Exp

lco

ne

Mil

112)

Fro

m.

I Thr

ough

(B

ath

date

s In

clus

irel

HO

B)

Ii 1

0C)

1970 D

ecg

1971

Jan

Feb

Mar A

prM

ayJu

ne

July

Aug

Sept

Oct

Nov

Dec

.

.

.

.....

....

.

-.. -

.

.

..

.

.

....

..

. ...

..

.....

..

..

.

....

.

....

.-

.....

....

....

...

.

..

..

..

....

...

. .....

.

...

..

-.

.

:

....

....

YO

UR

CO

MM

EN

TS

ON

CH

AN

GE

S IN

EM

PLO

YM

EN

T. P

AY

RO

LL

. OR

WA

GE

RA

TE

SE

nter

in c

olum

n 13

the

mai

n fa

ctor

s re

spon

sibl

e fo

r si

gnif

ican

t mon

th-t

o-m

onth

cha

nges

in th

e re

port

abo

ve.

Exa

mpl

es a

re:

Wag

e ra

le in

crea

se,

mor

e bu

sine

ss, f

ire,

tem

pora

ry s

umm

er h

elp,

over

time,

str

ike,

wea

ther

.If

any

GE

NE

RA

L W

AG

E-R

AT

E C

HA

NG

ES

(not

indi

vidu

al c

hang

es f

or le

ngth

of

serv

ice,

mer

it, o

r pr

omot

ion)

hav

e oc

curr

ed s

ince

last

mon

th's

rep

ort,

note

the

amou

nt o

fin

crea

se o

rde

crea

se (

u +

2%, 5

e), t

he e

ffec

tive

date

of

the

chan

ge, a

nd th

e ap

prox

imat

e nu

mbe

r of

non

supe

rvis

ory

empl

oyee

s af

fect

ed.

(13)

(13)

(13)

1971

Jan

Feb

Mar

.

Apr

1971

May

June

July

Aug

1971

Sept

Oct

Nov

Dec

_

(Per

son

to h

e ad

dres

sed

it qu

estio

ns a

rise

reta

rdin

g th

is r

epor

t)

a

(Pos

ition

)

BLS HANDBOOK OF METHODS

BLS 700 Industry Class Supplement

MANUFACTURING STATEMENT OF PRODUCTS

U.S. DEPARTMENT OF LABORBUREAU OF LABOR STATISTICS

WASHINGTON, D.C. 20212

r-

L

Return this form u soon as possi-ble in the enclosed envelope whichrequires no postage.

SAMPLE COPY

33

Budget Bureau No. 44-R746Approval expires January 21. 1972

The Bureau of Labor Statistics and the State agencies cooperatingin its statistical programs will hold all information furnished bythe respondent in strict confidence.

LOCATION(fits) (County) (State)

ICODES I Stat. I Report No. I Ind. I Empi. Mo. Yr. Proposed Ind.BLS II alga I Ind. Mo. Yr. I

This report will be used to insure the proper industrial classification of your regu ar 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 characteristicsprovided on the enclosed sheet. This list is not complete but represents the kind of information which should be reported.

PRINCIPAL PRODUCTS OR Arrmitro DURINO

1970(List Items separately)

(a)

PERCENT OrTOTAL SALES

VALUE ORRecstrroDURINO

1970

(b)

PRINCIPAL MATERIALS USED

(For each product listed In column (a) )

(e)

WAS MATERIALUs=

PRODUCEDIN THIS

ESTABLISHMENT?

Id)

'''V"V"2,',011A. Manufacturing (Specify below) , ,.?.,,,,,..r.,.,..,.; '. '144e. t.,.e,,,,,,+e, 4

%

Yes No

96 tt

%

96

%

96 .

3. Is the establishment primarily engaged inperforming services for other units of thecompany? Yes No

If "Yes," indicate nature of activity of this establishment:

Central administrative (Alice

Research, development, or testing

Storage (warehouse)0 Other (Specify: power plant, etc.)

1B. Nonmanufacturing (Specify below) 11;4'7yb

%

%

100%. r4Ail..., ''. Combined Total

2. Is this establishment part of a multiunit company?0 Yes 0 No

If "Yes," enter name and location of controlling company.

4. Space for Your Comments.

(Person to be addressed if questions Arles regarding this report)*Hi ROVERNMDft PRINTING C*7112,11170-.0-31.-104

(Position)

34 EMPLOYMENT, HOURS, AND EARNINGS


NumberI. 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, April1966, 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 ofemployment 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 thelabor force in the United States.

5. Nelson, Darrell R., "BLS Earnings Statistics for Use in Escalator Agreements," Employmentand 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 earningsseries 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, Pain 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 percapita 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 LaborReview, July 1952, pp. 57-59.

A discussion of the use of BLS average hourly earnings series in escalation clauses incontracts.

10. Wymer, John P., "The Revised and Expanded Program of Current Payroll Employment Sta-tistics," Employment and Earnings, November 1961, pp. ivvii.

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 ManufacturingReporting Practices," Monthly Labor Reoiew, 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.

PAUL A. ARMKNECHT,

Chapter 3. Job Vacancies and Labor Turnover

Background

Job vacancies, as used by the Bureau of La-bor Statistics, refers to the number of unfilledwage and salary jobs available to workers atthe end of a month. Labor turnover refers tothe gross movement of wage and salary work-ers into and out of employment status withrespect to individual establishments during themonth.

The current job vacancy-labor turnover pro-gram began in January 1969 with the additionof questions about job vacancies to the laborturnover questionnaire. The labor turnoverprogram, however, has been in existence formany years.

In January 1926, the Metropolitan LifeInsurance Co. began the collection of laborturnover data from a small sample of manu-facturing establishments. By February 1927,the sample included 175 establishments with800,000 employees, which was estimated to beabout 8 to 10 percent of total manufacturingemployment at the time. The original purposeof this series was to provide personnel man-agers with national figures on labor turnoverrates for manufacturing industries against

--which they could measure the experience oftheir own plants. Between November 1927 andJuly 1929, the Metropolitan Life Insurance Co.published labor turnover rates for total manu-facturing. By the latter date, the company feltthe project was sufficiently successful and wellestablished to warrant turning it over to theBureau of Labor Statistics for further develop-ment. A decade later, in December 1939, serieson labor turnover rates were being publishedfor 30 manufacturing industries, and the sam-ple upon which the rates for all manufacturingwere based contained 5,500 establishments andnearly 2,600,000 employees.

For a number of years, State employmentsecurity agencies affiliated with the Bureau ofEmployment Security (now the U.S. Train-

ing and Employment Service) had collectedlabor turnover information for use in jobmarket 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 thejoint collection of labor turnover data beganwith an agreement with Connecticut in 1954.By 1964, the cooperative program had beenextended to cover all 50 States and the Districtof Columbia.

Experimental programs to determine thefeasibility of collecting job vacancy informa-tion were conducted in 1965 and 1966. The1965 study was carried out in 16 metropolitanareas and the 1966 study was conducted in 14metropolitan areas and three States. Followingthe successful completion of these programs,cooperative arrangements for the collection ofjob vacancy data were set up between the BLSand 48 of the State employment services,through the Manpower Administration.

In December 1970, these agencies publishedabout 8,000 labor turnover series in manu-facturing and mining industries for State andareas, and about 130 job vacancy area series inmanufacturing industries. These rates werebased on a sample of approximately 32,000reports in manufacturing and about 1,200 inmining.

Description of the Survey

Labor turnover actions are divided into twobroad groups: accessions or additions to em-ployment, and separations or terminations ofemployment. These two broad groups are fur-ther divided; accessions into new hires andother accessions, and separations into quits,discharges, layoffs, and other separations. La-bor turnover is expressed in the BLS series asa monthly rate per 100 employees. Separaterates are computed for each of the componentitems.

85


The primary difference between types ofseparations is whether action is initiated bythe employee or employer, i.e., whether it isvoluntary on the employee's part or involun-tary. Voluntary actionsquits--are initiatedby the employee for an almost unlimited va-riety of reasons, financial, personal, or social,(e.g., lack of housing and transportation, poorcommunity facilities, etc.). Involuntary actionseither may be initiated by the employer or bebeyond the control of both employer and em-ployee; these actions may arise from economiccauses such as business conditions, physiolog-ieal reasons such as aging, or performancereasons such as incompetence.

Job vacancy data are collected both for totaljob vacancies and for vacancies which havecontinued unfilled for 30 days or more. Data onthe occupations for which the vacancies existand the wage rates offered for them are also col-lected quarterly. Vacancy data are expressedas a monthly rate equal to job vacancies dividedby the sum of vacancies and employment, withthe quotient multiplied by 100.

Concepts

Separations are terminations of employmentof persons who have quit or been taken off therolls for reasons such as layoff, discharge, re-tirement, death, military service expected tolast more than 30 consecutive calendar days,physical disability, etc. Since January 1959,transfers of employees to other establishmentsof the same company also have been clakisifiedas separations.

Quits are terminations of employment initi-ated by employees for any reason except re-tirement, transfer to another establishment ofthe same firm, or service in the Armed Forces.Included as quits are persons who failed toreport after being hired (if previously countedas accessions), and unauthorized absenceswhich, on the last day of the month, have lastedmore 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, forreasons such as lack of orders, model change-

over, termination of seasonal or temporaryemployment, inventory-taking, introduction oflabor saving devices, plant breakdown, orshortage of materials.

Discharges are terminations of employmentinitiated by the employer for such reasons asincompetence, violation of rules, dishonesty,laziness, absenteeism, insubordination, failureto pass probationary period, etc.

Other separations include terminations ofemployment for military duty lasting or ex-pected to last more than 30 days, retirement,death, permanent disability, failure to meet thephysical standards required, and transfers ofemployees to another establishment of thecompany.

Accessions are all permanent and temporaryadditions to the employment roll, whether ofnew or rehired employees. Transfers fromanother establishment of the same companyalso are counted as accessions (beginning withJanuary 1959).

New hires are permanent and temporaryadditions to.i. the employment roll of personswho have never before been employed by theestablishment, and lamer employees rehiredalthough not specifically recalled by the em-ployer. This category excludes transfers fromother establishments of the same company andemployees returning from military service orunpaid leaves of absence.

Other accessions include all additions to theemployment roll other than new hires.

Job vacancies are defined as vacant jobswhich are immediately available for filling, andfor which the firm is actively trying to find orrecruit workers from outside the firm.

"Actively trying to find or recruit" meansthat the establishment is engaged in currentefforts to fill the job vacancies.

Long-term job vacancies are those currentvacancies which have continued unfilled for 30days or more.

The reporting establishment is also asked toindicate the number of openings with futurestarting dates for which the firm is activelytrying to recruit workers from outside thefirm. Job openings with future starting datesmay exist for such reasons as: Job unavailableuntil expected separation of present incumbent

40

JOB VACANCIES AND LABOR TURNOVER 37

occurs; work will not start until some futuredate ; new branch to be opened in the future ;or anticipated increase in business.

Industry Classification

The classification system used for compilingand publishing rates is that described in the1967 Standard Indvstrial Classification Manualissued 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 onthe basis of major product or activity as de-termined by annual sales data for the previouscalendar year. Most establishments in the jobvacancy-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 ofindustry classification in the two programs isgiven under the heading, Industrial Classifica-tion in chapter 2 of this bulletin.

Occupational Classification

Occupational classifications are made inaccordance 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 Stateemployment service operations.

Data Sources

Each month cooperating State employmentsecurity agencies collect data on the number ofjob vacancies and on labor turnover actionsfrom a sample of establishments drawn froma list of those subject to State unemploymentinsurance programs. In nonmanufacturing,supplemental sources are also used to obtainlists of establishments which are not coveredby Unemployment Insurance laws. (See chap-ter 2, p. 17 of this bulletin.) The respondent ex-tracts the figures largely from his personnelrecords, though some smaller establishmentswhich do not maintain special personnel rec-ords utilize their payroll records in making out

the reports. Response analysis surveys, whichanalyzed the reporting practices of a scien-tifically selected sample of the establishmentsin 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 thesedeviations on the published data appeared tobe quite insignificant, particularly for thebroader classes, such as total accessions, totalseparations, and total job vacancies.

Collection Methods

Job vacancy and labor turnover data arecollected primarily at the State level by em-ployment security agencies from cooperatingemployers 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 entryof current data. The respondent reports thetotal number of job vacancies, the number ofactions for each turnover item during the cal-endar month and total employment. These em-ployment figures, which are the bases used tocompute the rates, represent the number of per-sons who worked or received pay for any partof the pay period (usually 1 week) which in-cludes the 12th of the month.

Information on the occupations' for whichjob vacancies exist and the pay rate beingoffered for them is collected quarterly on asupplemental schedule, form DL 1219A.

The State agency uses the information pro-vided on the schedule to develop job vacancyand labor turnover rates for the States and formetropolitan areas, and forwards the data toWashington, where they are used by the Bu-reau of Labor Statistics to prepare rates atthe national level.

Sampling

Sampling is used by BLS for collecting datain its job vacancy-labor turnover statistics pro-

41

38

DL 1219


andTHE MANPOWER ADMINISTRATION



Stale Report No. Ind. Area Local OtT.

MONTHLY REPORT ON JOB OPENINGSAND LABOR TURNOVER

Enter the data requested and return Inthe ,reload envelop. as soon as theIntorasstIon Is svallsbIe each month.

CHANGE NAME AND MAILING ADDRESS IF INCORRECTINCLUDE ZIP CODEr".

LDelors entering amo is. explanations on other Ode

Budget Bureau No. 44-111064.Approval expires January 31. 1972.

The Bureau of Labor Statistics, the Manpower Administra-tion, and the State agencies cooperating in their statisticalprograms will hold all information furnished by the respond-ent in strict confidence.

LOCATION(County)(City) (State)

YEAR

AND

MONTH

(1)

I. LABOR TURNOVER DURING CALENDAR MONTH ILEMPLOYME/(One pay periocPERIOD COVERED BY

LABOR TURNOVER(Cols. 4 through Ili

Preferably one calendarmonth

All EMPLOYEESTOTAL NUMBS

who worked due-Ins or received payfor any part of thepay pericd whichInclude. the 12th

of the month(la)

SEPARATIONS (during calendar month) ACCESSIONS (during calendar month)

From(Both dates

(2)

ThroughInclusive)

(3)

TotalSeparations

(Sum of cols.6 through 8)f)

Quits

AA)

Discharges

(6)

Layoffs

(7)

Otherseparations

(8)

TotalAccessions

(Sum of cols.10 and II)

(9)

Newhires

(10)

OtherMentions

( I I)

1970Dec.

1971Jan.

Feb.

Mar.

Apr.

May

June

July

Aug.

Sept.

Oct.

Nov.

Dec.

7

- ------

1... .. /

4

YEAR

AND

MONTH

III. JOB OPENINGS for which workers from outside the firmwere actively being sought as of close of last business day ofmost recent month, (If "NONE," enter "0," do not leave blank.)

IV. YOUR COMMENTS

CURRENT JOB OPENINGS(unoccupied and ready for immdiate filling)

OPENINGS WITHFUTURE

STARTINGDATES

(not included incolumn 13)

(16)

DONOTUSE

Expl.ca.(16)

Enter main factors responsible for any significant monthto-month changes M SECTIONS I, II, and III. Examples are:

More balite*. WeatherStrike Temporary summerFIN help

Seasonal Increase(171

Number of currentJob Openings

(13)

Openings Included Incal. 13 continuingunfilled for 30 dare

Or MOTs(14)

1970Dec.

1971Jon.

Feb.

Mar.

Apr.

May

June

July

Aug.

Sop).

Oct.

Nov.

Dec.

(person to be addressed t( questions grist regarding this report) (Position)

JOB VACANCIES AND LABOR TURNOVER 39

INSTRUCTIONS FOR COMPLETING THIS FORM

I. LABOR TURNOVER

PERIOD COVERED. information on labor turnover, cols. 4 through11, is requested for the most recent entire calendar month speci-fied in column 1 or, if this is not possible, for a period whichmost closely covers that calendar month. In either case enter incols. 2 and 3 the beginning and ending dates for the monthlyperiod for which turnover data are reported.

SEPARATIONS (ALL EMPLOYEES)

Column 4. TOTAL SEPARATIONS DURING CALENDAR MONTH.. Enterin column 4 the sum of columns 5 thru 8.

Column 5. auffS.A quit is a termination of employment ini-tiated by the employee for any reason except to retire, to trans-fer to another establishment of the same firm, or for service inthe 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 hasbeen absent more than 7 consecutive calendar days.

Column 6. DISCHARGE:A discharge is a termination of employ-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 meetorganization's physical standards should be reported in otherseparations, col. 8.

Column 7. LAYOFFS.A layoff is a suspension from pay status(lasting or expected to last more than 7 consecutive calendardays without pay) initiated by the employer without prejudiceto the worker for such reasons as: 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. OTHER SEPARATIONS.This group should include onlyterminations 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 ofthe company. NOTE: 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 shouldnot be counted as separations.

ACCESSIONS (ALL EMPLOYEES)

Column 9. TOTAL ACCESSIONS DURING CALENDAR MONTH.An ac-cession is any permanent or temporary addition to the employ-ment roll whether of new or former employees, or transfers fromanother establishment of the company. Enter in column 9 thesum of cols. 10 and 11.

Column 10. NEW HIRES New hires are temporary or permanentadditions to the employment roll of (1) anyone who has neverbefore been employed in this establishment, or (2) former em-ployees you did not call back. Former employees you did callback should be included in total accessions and "other acces-sions," but not in "new hires." Persons transferred from otherestablishments of this company should be reported in "otheraccessions."

Column 11. OTHER ACCESSIONS.Include all additions to the em-ployment roll other than new hires. This includes all employeescalled back to work by the employer from a layoff as defined forcol. 7, transfers from other establishments of the company, andformer 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 becounted as accessions when they return to work.

H. EMPLOYMENT

PERIOD COVERED.Employment information, col. 12, is requestedfor one pay period (preferably one week) which includes the

12th of the calendar month for which labor turnover data arereported.

Column 12. ALL EMPLOYEES.Enter the total number of persons(both sexes) on the payrolls of the establishment (s) covered inthis report who worked full- or part-time or received pay forany part of the pay period (preferably one week). Includesalaried officers of corporations, executives and their staffs, andemployees engaged in a force-account construction but excludeproprietors, members of unincorporated firms, and unpaidfamily workers. Include persons on vacations and sick leave forwhich they received pay directly from your firm for the payperiod covered but exclude persons on leave without companypay the entire period and pensioners and members of the ArmedForces carried on the rolls but not working during the payperiod covered.

NOTE: If the number differs from the "All Employees" totalreported on the Employment, Payroll, and Hours form, explainunder Comments.

III. JOB OPENINGS

PERIOD COYERED.Job openings information, columns 13, 14 and15, is requested as of close of last business day (or nearest pos-sible day) of the most recent monthly period for which laborturnover data are reported.

Column 13. NUMBER OF CURRENT JOB OPENINGS (VACANCIES).Enterthe number of current job openings in your establishment. IF"NONE," ENTER "0." A current job opening is an existingvacant job in your establishment that is immediately availablefor filling and for which your firm is actively trying to find orrecruit some worker from outside your firm (i.e., a "new"workernot a company employee).Include such openings for all kinds of positions, classificationsand employment (full-time, part-time; permanent, temporary,seasonal), including those outstanding on orders with employ-ment agencies-and.notifications to unions.-Excltile jobs to be filled by recall, transfer, promotion, demo-igit 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 worklater; and the "openings with future starting dates" reportedin column 16."Actively trying to find or recruit" means current efforts to fillthe job opening through orders listed with public or private em-ployment agencies and school placement offices; notifications tolabor unions and professional organizations; "help wanted" ad-vertising (newspaper, posted notice, etc.); recruitment pro-grams; interview and selection of applicants.Column 14. CURRENT JOB OPENINGS CONTINUING UNFILLED FOR 30DAYS OR MORE.Enter the number of current job openings in-cluded in the figure reported in column 13 which have continuedunfilled for 30 days or more. IF "NONE," ENTER "0."Column 16. OPENINGS WITH FUTURE STARTING DATES.Enter thenumber of openings in your establishment for which your firmis actively trying to find or recruit some worker from outsideyour firm (i.e., a "new" worker), but which relate to jobs thatare currently occupied or unavailable for immediate occupancyby "new" workers for such reasons as: job unavailable until ex-pected separation of present incumbent occurs: work will notstart until some future date; new branch to be opened in futureor anticipated increase in business. IF "NONE," ENTER "0."NOTE: OPENINGS WITH FUTURE STARTING DATES

ARE NOT TO BE INCLUDED IN COLUMN 13.

IV. COMMENTS

Column 17. YOUR COMMENTS.Enter the main factors responsiblefor significant month-to-month changes in Labor Turnover (cols.4 through 11), Employment (col. 12), and Openings (cols. 18through 15). Some examples are listed in the heading of col-umn 17.

10 U.S. GOVERNMENT PAINTING /0440/.201


gram, since full coverage would be prohibi-tively costly and time consuming. The samplingplan for the program must : (a) provide thepreparation of reliable monthly estimates ofjob vacancy and labor turnover rates whichcan 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 appropriatefor the existing framework of operating pro-cedures, administrative practices, resourceavailability, and other institutional charac-teristics of the program.

In developing the sample design, the universeof establishments was stratified first by in-dustry and within each industry by size ofestablishment in terms of employment. Withineach industry, an optimum allocation designwas obtained by sampling with probabilityproportionate to average size of employmentwithin each of the strata. The total size ofsample regarded as necessary to produce satis-factory estimates of employment was dis-tributed among the size cells on the basis ofaverage employment per establishment in eachcell. In practice, this is equivalent to distribut-ing the predetermined total number of estab-lishments required in the sample among thecells on the basis of the ratio of employment ineach cell to total employment in the industry.Within each stratum, the sample members areselected at random.

Under this type of design, large establish-ments fall into the sample with certainty.Establishments with 250 or more employeesare included in the sample with certainty, al-though in some cases the cutoff is lower. Thesizes of the samples for various industries weredetermined empirically on the basis of exper-ience.

The sample design, although aimed primarilyat meeting the needs of the national program,provides a technical framework within whichState and area sample designs can be de-termined. Since, however, the rates for Statesand areas are not generally prepared at thesame degree of industry detail as the national

1 For the national sample, additional reports needed forState and area samples are added to those required by thenational design.

44

rates, the national design usually providessufficient reports for the preparation of Stateand area rates.'


Both job vacancy and labor turnover ratesare estimates of ratios. For individual indus-tries, turnover rates are computed by dividingthe number of turnover actions of each type,as reported by the sample establishments, bythe total number of employees reported bythose establishments. The result is multipliedby 100. In an industry sample, for example,623 employees quit between January 1 and 31,while 30,062 employees worked or received payduring the week of January 11-17. The Jan-uary quit rate for the industry is :

623

30,062 x100=2.1

Turnover rates for industry groups are com-puted by weighting the rates for the componentindustries 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 12thof the month, are described in chapter 2 ofthis bulletin. Rates for "all manufacturing"and for the durable and nondurable goods sub-divisions of manufacturing are weighted byemployment 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 industrystratum. The weighted number of vacancies isthen divided by the sum of employment in thestratum plus vacancies ; the quotient is thenmultiplied by 100 to determine the vacancyrate. Rates for major industry groups, for thedurable goods and nondurable goods sub-divisions, and for total manufacturing arecomputed by summing the weighted number ofvacancies in the component cells of industriesor subdivisions, then dividing by the sum ofvacancies and employment in that industry,

JOB VACANCIES AND

subdivision or division. Again, the quotient ismultiplied by 100 to obtain the rate.

As of 1971, size stratification was not usedin the preparation of job vacancy and laborturnover rates. Tests were underway to de-termine the effect on the rates of introducingsize stratification. Preliminary results of thetests suggested that size stratification wouldimprove the labor turnover rates but have littleeffect 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 laborturnover rates, reflect a regularly recurringseasonal movement which can be measured onthe basis of past experience. By eliminatingthat part of the change which can be ascribedto usual seasonal variation, it is possible toobserve the cyclical and other nonseasonalmovements in these series. Seasonally adjustedlabor turnover rates are published at the allmanufacturing industry level.

The seasonal adjustment method used forthese series is an adaptation of the standardratio-to-moving average method, with a pro-vision for "moving" adjustment factors to takeaccount of changing seasonal patterns. A de-tailed description of the basic method is givenin appendix A of this bulletin.

The 2 years of experience with job vacancydata indicate the emergence of some probableseasonal patterns. However, a minimum of 3years of data are needed to develop estimatesof the seasonal factors.

Presentation

The BLS publishes, on a national basis,monthly series of labor turnover rates for se-lected industries. These series are currentlypublished for the manufacturing division, thedurable and nondurable goods subdivisions, 21major industry groups in manufacturing, 191individual manufacturing industries, and 7categories in mining and communications.Rates are available for all manufacturing from

LABOR TURNOVER 41

January 1930 and for telephone and telegraphfrom 1943. For industry groups and individualindustries in the manufacturing and miningdivisions, all series begin with January 1958.Rates for certain highly seasonal industries,for example canning and preserving, are notnow published separately but are included inthe computation of rates for the major manu-facturing groups. Before 1958, these industriesand the printing, publishing and allied indus-tries major group were not included in therates for all manufacturing. The rates for allmanufacturing for years prior to 1958 wererevised, however, to reflect the influence ofthese industries.

Monthly rates for total accessions, new hires,total separations, quits, and layoffs are shownfor manufacturing and mining industries. Ex-cept for the new hire rates, the same items arepublished for the telephone and tilegraph in-dustries.

On a national basis, the BLS presently pub-lishes monthly estimates of total and long-termjob 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 eachof these divisions and industry groups.

Preliminary job vacancy rates for the nineselected major industry groups and turnoverrates for the 21 major industry groups inmanufacturing are published monthly in aBLS press release about a month after thereference month., and in the Monthly LaborReview 3 months after the reference month.Preliminary turnover rates for both detailedindustries and broad categories and prelimi-nary job vacancy rates for the same categoriesas are included in the press release are pub-lished in Employment and Earnings about 2months after the month of reference.

Both job vacancy and labor turnover ratesfor all manufacturing for selected States andmetropolitan areas are published each monthin Employment and Earnings. More detailedinformation is available in releases issued bythe cooperating State agencies.

National labor turnover rates (monthly dataand annual averages) back to the beginning ofeach series are published in the annual volume

45


called Employment and Earnings, UnitedStates. New editions of this volume are pub-lished annually, following each adjustment ofthe Bureau's industry employment statisticsseries to new benchmark levels.

Uses and Limitations

The two major causes of change in both jobvacancy and labor turnover rates are industrialexpansion and contraction. In prosperoustimes, job vacancy rates, quit rates, and newhires are high because of job availability; inperiods of economic recession, high layoff ratesare coupled with low vacancy, quit, and acces-sion rates. Turnover ,rates are, therefore, re-garded as good economic indicators and arewidely used by economic analysts in both gov-ernment and private industry. Together withthe turnover data and other labor force data,such as unemployment rates, job vacancy dataare also expected to provide an indicator as tothe condition of the economy.

Labor turnover rates by industry are alsovaluable for personnel planning and analysis.Employers use these rates as a yardstickagainst which to measure the performance oftheir plants. For example, they consider lowquit rates to be an indication of efficient opera-tions and good labor-management relations. Aconsideration of turnover is essential for sched-uling production and for planning the orderlyrecruitment and maintenance of an adequatemanpower supply. Labor turnover rates arealso widely used by State employment servicesto plan and appraise their operations.

Job vacancy data should also prove useful toState employment services. By identifyingemerging labor shortages, vacancy data will

461.

allow for more intelligent planning of trainingprograms and should be useful in counsellingthe unemployed and new entrants to the labormarket. The data may also make interarea re-cruitment of workers possible.

The use of turnover or job vacancy rates tointerpret changes in the BLS monthly employ-ment series is limited for the following rea-sons: (1) The labor turnover series measureschanges during the calendar month, while theemployment series measures changes frommidmonth to midmonth; and (2) employees onstrike are not counted as turnover actions, al-though such employees are excluded from theemployment estimates if the work stoppagelasts throughout the report period includingthe 12th of the month.

The Bureau publishes annual averages of jobvacancy and labor turnover rates, which arecomputed as the arithmetic means of the 12monthly rates. These can provide a usefulmeasure if a 1-month rate is not suitable forsome purposes, as for example when the ratefor a specific month is considered to be unusualor affected strongly by seasonal influences.2

-SHEILA C. WHITE

3 Because they are liable to misinterpretation, the Bureaudoes not prepare cumulative annual rates of labor turnover.For example, an annual quit rate could be obtained by divid-ing the total number of quits during the year by averageemployment during the year. An approximation of this figurecan be obtained by cumulating the 12 monthly rates. Supposethe annual rate thus obtained amounted to 50 per 100 em-ployees. This might seem to imply that 50 percent of allemployees in January voluntarily left their jobs by the endof December. However, many jobs in a given establishmentare vacated and refilled more than once during the year. TheBureau does not have information on the number of em-ployees who remained with the establishment during the en-tire year. Over short periods of time, labor turnover ratesprobably include relatively little repetitive counting of em-ployees who have held the same Jobs, while over a period ofas long as a year there is considerable duplication.

Manpower Structure and Trends

Chapter 4. Employment of Scientific and Technical Personnel

BackgroundThe 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 anumber of years, a major element in thisgrowth has been the need of the Federal Gov-ernment for increasingly complex and tech-nologically advanced national defense materialand space related research and technology. Thesurveys of scientific and technical personnelprovide a means for estimating current man-power resources and a basis for evaluatingfuture requirements for scientific an_ d engineer-ing personnel.

The surveys of scientific and technical per-sonnel were initiated by the Bureau of LaborStatistics in 1959, under the sponsorship ofthe National Science Foundation. They arepart of a comprehensive statistical program,coordinated by the Foundation, designed toyield estimates of the employment of scientistsand engineers in all sectors of the U.S. eco-nomy. The Bureau of Labor Statistics con-ducted surveys of scientific and technicalpersonnel in industry annually during the 1960's(except for 1965) and related surveys of Stategovernment agencies less frequently. Since1962, the Bureau of Labor Statistics has hadfull responsibility for the conduct and publica-tion of these surveys, which continue to beplanned in consultation with the National Sci-ence Foundation. The last survey of scientificand technical personnel was conducted in 1970.In the future, estimates of scientific and tech-nical personnel employed in private industrywill be developed as part of the Bureau's Oc-cupational Employment Statistics Program,which will provide employment statistics onall occupational fields.

Description of SurveysThe surveys of scientific and technical per-

sonnel in industry gather data on the employ-ment of engineers as a group, and on scientistsand technicians by major occupational spe-

cialty. Engineers and scientists are furtherdistributed according to whether they are en-gaged primarily in research and developmentactivities, in management and administration,in technical sales and service, in productionand operations, or in "all other" functions.Data are published separately for all majorindustry groups, and in finer industry detailfor a selected number of major industry groups.Beginning in 1966, data were developed bygeographic area.

State agencies are surveyed for informationon 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 areengaged. Data also are published by State whenappropriate.

For each professional occupation, respond-ents are asked to report the number of personswhose current positions require knowledgeequivalent at least to that acquired throughcompletion of a 4-year college course with anappropriate academic major, regardless ofwhether they hold a college degree. These sur-veys, thus, cover all persons actually workingin one of the designated occupations, specifi-cally including those who do not hold an ap-propriate degree or any degree at all, butspecifically excluding persons trained in theoccupation but currently employed in positionsnot requiring the use of such training.

Technicians are defined as persons actuallyengaged in technical work at a level which re-quires knowledge of engineering, mathematical,physical and life sciences, comparable to thatacquired through technical institutes, juniorcolleges, or other formal post-high school train-ing less extensive than 4-year college training,or through equivalent on-the-job training orexperience.

Data SourcesSources of occupational data reported by re-

spondents are personnel records and especiallyfor the small reporting units, personal knowl-

48


edge of persons completing the reports. Dis-cussions with a number of large respondentsindicate that their records typically containmuch of the data in the desired form, but thatsome adjustments by the respondent are oftennecessary because the occupational classifica-tions used in his records differ somewhat fromthose specified for the surveys.

Employment benchmarks for the survey ofscientific and technical personnel in industryare derived from employment data tabulatedfrom the first quarter reports of the unemploy-ment insurance program. The survey of Stategovernment agencies is based on a compilationof data from all State government agenciesemploying personnel in occupations covered bythe survey.

Collection Methods

Data are collected from respondents pri-marily by mail, but personal visits are made tomany large employers, and to other respond-ents who indicate particular difficulty in com-pleting the questionnaires. These visits, whichlimited resources have prevented from beingeither numerous or frequent, are carried out bysenior staff members. Normally two mailingsfollow and a sub-sample of residual nonre-spondents are contacted further by telephone.

The response to these surveys has been veryencouraging. Respondents supplying usable in-formation have constituted approximately 80percent of the reports solicited in virtuallyevery year, and have never been below 75percent.

Sampling

The basic sample for the industry survey wasdrawn from lists of establishments reporting toeach of the 51 (State and D.C.) employmentsecurity agencies for unemployment insurance(UI) purposes, and was supplemented by a listof interstate railroads and related companiessupplied by the Interstate Commerce Commis-sion. Industry classification of establishments is

based on information available to the Stateagencies.

Certain categories of establishments areeliminated from the master list before thesample is selected, either because a separatesurvey of the given category is being made orbecause the number of scientific and technicalpersonnel employed are believed to be negligi-ble. The categories of organizations omitted arethose classified according to the Standard In-dustrial Classification system in the followingmajor industry groups: 01 and 02farms ; 80medical and other health services (except807, medical and dental laboratories, which isincluded) ; 82educational services; 84mu-seums, art galleries, and botanical and zoo-logical gardens; 86nonprofit membershiporganizations; 88private households; 89miscellaneous services (except 8911, engineer-ing and architectural service, which Is in-cluded) ; 91 through 94government; and 99nonclassifiable establishments.

Establishments below a specified minimumsize, determined separately for each industrygroup, are also excluded from the listing priorto sampling. Because of the large number ofestablishments in the small size groups, mini-mum size cutoffs are essential to the efficiencyof the survey. Since excluded establishmentsemploy very few scientists, engineers, or tech-nicians, survey results are affected little bythese omissions.

Sample numbers are allocated among thevarious industry-size strata according to theprinciple of optimum allocation ; expected re-sponse rates by industry and by size are takeninto account to obtain maximum reliabilitywithin available resources. The overall samplesize is determined so that the variance (tworelative standard errors) for the estimate ofthe total number of scientists and engineers forall industries combined is about 5 percent. Inevery covered industry, all establishments with1,000 employees or more are included in thesample. In other industry-size cells, the sam-pling ratios range from 1 in 1 to 1 in 100. Ingeneral, the larger the establishment and thegreater the number of technical personnel usedby the industry, the higher is the samplingratio. All selections are made randomly withinthe designated strata.

EMPLOYMENT OF SCIENTIFIC AND TECHNICAL PERSONNEL

Since scientific and engineering employmentis concentrated to a significant degree in re-search and development laboratories, not sep-arately identifiable in the UI universe, theprobability sample was supplemented. Theessential rule for unbiased supplementation isthat the supplementary units must be drawnindependently of the probability sample ; thatis, the chance that a unit is drawn in the prob-ability sample must be independent of thechance that the unit is selected as a supplement.The initial supplementation was drawn from alist of industrial research laboratories compiledby the National Academy of ScienceNationalResearch Council and from a list of small busi-ness concerns interested in performing researchand development compiled by the Small Busi-ness Administration. Beginning in 1966, whena new sample was drawn, supplementation wasachieved by retaining in the mailing list allestablishments which reported employment of20 or more scientists and engineers in surveysbased on the previous sample. Establishmentsselected as supplements are tabulated as a sep-arate cell within their industry and size classwith a weight of 1, regardless of whether theyare also members of the probability sample.(See Estimating Procedures.)

The mailing list for State government sur-veys is not a sample, but includes all agenciesof State governments whiCh could conceivablyemploy personnel in any of the designated oc-cupations. The agencies are identified from in-formation in directories and other documentsfurnished by the States.


For the survey of scientific and technicalpersonnel in industry, estimates are obtainedfor probability cells as the ratio of primaryitem employment to total employment (of thereporting units in the cell), multiplied by atotal employment figure in that industry andsize class that is adjusted for any supplementalunits in that industry and size class to preventduplicate estimation for supplemental units.'Estimates for supplemental cells are obtainedby summing the primary item employment for

45

the supplemental reports plus an estimate fornonrespondent supplementary units.

Estimates for the survey of scientific andtechnical personnel employed by State govern-ment agencies are obtained by summing thereported data. The response rate in this surveyis extraordinarily high-96 to 98 percentandexamination of the nonrespondents shows thatthe number of scientists, engineers, and otherpersonnel employed by them is negligible.

Analysis, Interpretation, andPresentation

A report on the findings of each survey ispublished, usually within 2 years of the refer-ence date of the survey. Each report consists ofan analytical interpretation of the findings,and is supported by a statistical appendix con-taining in tabular form all of the data that canbe meaningfully derived from the survey.


Data from these surveys form the essentialstatistical base (1) for evaluating the adequacyof scientific and technical manpower resourcesof the United States in light of current orprojected demands and (2) for determining therate of growth of these resources. They havebeen used to evaluate the impact of new or en-larged Federal programs calling for substantialscientific and technical manpower. These dataalso provide the bases for projections of futuremanpower requirements in science and engi-neering. For example, data are furnished foroccupational guidance counselors and otherswho provide young people with information onwhich to base a choice of career.

These estimates must be interpreted as ap-proximations. All surveys are subject to possi-

Symbolically. PIM,where M is the cell universe total

employment, vp, is the sum of the primary item employmentof the cell respondents, e, is the sum of total employmentof the cell repsondents, and P' is the estimate. M is adjustedto prevent duplicate estimation for supplemental sample re-porters.

46 BLS HANDBOO

ble response and processing errors, althoughthese are reduced insofar as possible, throughchecking procedures and through correspond-ence with reporters whose data are internally

K OF METHODS

inconsistent or appear to involve misinterpreta-tions of definitions or other instructions. Inaddition, estimates derived from sample sur-veys are limited by sampling error.


Number1. 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 1964 (B:lletin 1557, 1967 ).

4. "Scientific and Professional Employment By State Governments" (Reprinted from theAugust 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 in State Gov-ernment Agencies, Report on a 1959 Survey. NSF 61-17 (1961).

7. Reports prepared for the National Science Foundation by the U.S. Department ofLabor, Bureau of Labor Statistics: National Science Foundation, Scientific and TechnicalPersonnel in American Industry, Report on a 1959 Survey. NSF 60-62 (1960).

8. National Science Foundation, Scientific and Technical Personnel in Industry, 1960.NSF 61 -75 (1961).

9 National Science Foundation, Scientific and Technical Personnel in Industry, 1961.NSF 63-32 (1963).

MICHAEL F. CROWLEY

50

Chapter 5. Occupational Outlook

Background

The occupational outlook program originallystemmed from a report of the Advisory Com-mittee on Education appointed by PresidentRoosevelt. This Advisory Committee recom-mended, in 1938, that an occupational outlookservice be set up in thepBureau of Labor Sta-tistics to make studies and provide informationfor use of individuals in choosing a career, andfor the use of those responsible for the plan-ning of education and training programs. In1941, the Occupational Outlook Service wasorganized under a specific authorization by theCongress. Although the first, preliminary stud-ies were begun in 1941, it was not until afterWorld War II that the occupational outlookstaff was able to devote its efforts to the prep-aration of occupational reports for use in guid-ance. In mid-1946, a manual of occupationaloutlook information was prepared for use inthe Veterans Administration (VA) counselingand rehabilitation program.

In response to a resolution by the NationalVocational Guidance Association, calling uponthe Congress to authorize this type of informa-tion for sale, and to requests by other privateindividuals and groups, the first edition of theOccupational Outlook Handbook was publishedin 1949. The favorable public response to theHandbook was a major factor in the Bureau'sdecision 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 publicrecognition of the key role of vocational guid-ance in staffing essential occupations andeffectively utilizing the Nation's manpowerresources. This resulted in the Congress in 1955providng for the mainteance of the Occupa-tional Outlook Handbook and its related publi-cations on a regular, continuing, up-to-datebasis. In 1957, the third edition of the Occupa-tional Outlook Handbook was published; alsoin 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 theHandbook.

Description of Program

Under the occupational outlook program, theBureau of Labor Statistics conducts researchin, and provides information on, future occupa-tional and industry manpower requirementsand resources. It provides vocationui guidanceinformation on expected employment oppor-tunities for the use of counselors, educators,and others helping young people in choosinga field of work. It also provides manpowerinformation for local and national trainingauthorities and policymakers for use in de-veloping programs of education and training.The results of the research are published in theOccupational 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 andanalyzed considerable manpower informationon such topics as employment trends for majorindustries of the economy and for most majoroccupations; employment effects of a greatmany long-term programs of government agen-cies, including those for defense, highways,scientific research, space technology, medicalcare, and education ; and changes in industryand occupational requirements.

Toward providing an overall framework offuture manpower requirements for the economyas a whole, projections are developed for thebroad industry and occupational groups, andhave been published regularly.' Every other

1"America's Industrial and Occupational Manpower Re-quirements. 1964 -75," prepared for the National Commissionon Technology, Automation, and Economic Progress, pub-lished in Technology and the American Economy, AppendixVol. I, February 1966, and The U.S. Economy in 1980, (BLSBulletin 1673, 1970).

47

48 BLS HANDBOOK

year, hundreds of different, detailed occupa-tional and industry statements are published inthe Occupational Outlook Handbook. In most ofthese occupational outlook statements, informa-tion is provided on : nature of work; places ofemployment, education and training require-ments; employment outlook for about 10 yearsahead, including, in most cases, estimates ofannual requirements for growth and replace-ment needs; and earnings and working condi-tions. In presenting outlook statements forindustries, information is included on natureand location of each industry and other indus-try characteristics, as well as information onthe industry's major occupations.

In presenting the employment outlook for anoccupation, information is provided not only onthe demand for workers but also on the poten-tial supply of workers from many sourcesschools and other training institutions, trans-fers from other occupations, and reentries tothe labor force. The balance between supplyand demand, in those occupations for which anassessment is possible, gives some indication ofthe nature of job competitionOn a specific fieldfacing young people in the years ahead.

In addition to the overall and detailed indus-try and occupational projections developed forthe Handbook and described earlier, specialmanpower studies are prepared, as part of theoccupational outlook program, that provide in-formation, narrower in scope and greater indepth, on the changing industrial structure and

*Employment and Changing Occupational Patterns in theRailroad Industry, 1947 -60 BLS Bulletin 1344 (1963); Employ-ment Requirements and Changing Occupational Structure inCivil Aviation BLS Bulletin 1387 (1984); and EmploymentOutlook and Changing Occupational Structure in ElectronicsManufacturing BLS Bulletin 1363 (1983).

3 Maxine G. Stewart, "A New Look at Manpower Needs inTeaching," Monthly Labor Review, June 1984, 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," MonthlyLabor Review, May 1966, pp. 503409; "Job Hopes Highfor Record College Class." U.S. Department of Labor, PressRelease, June 5, 1986 (USDL 7246), and College EducatedWorkers, 196880, BLS Bulletin 1676 (1970).

*Allan F. Salt, "Estimated Need for Skilled Workers.1985-75." Monthly Labor Review, April 1966, pp. 385-371;Joseph F. Fulton, "Employment Impact of Changing DefensePrograms," Monthly Labor Review, May 1964, pp. 508-516;Bernard Yabroff, "Trends and Outlook for Government Em-ployment," Monthly Labor Review, March 1985, pp. 285-291. and Max Carey, "The CraftsFive Million Opportunities,"Occupational Outlook Quarterly, Spring 1971, pp. 2-11.

OF METHODS

occupational composition of American indus-tries, such as railroads, civil aviation, and elec-tronics.2 Other special studies provide moretechnical information and quantitative projec-tions of manpower requirements and resourcesin specific occupations, industries, or for speci-fic groups of workerssuch as teachers, tech-nicians and nonwhite workerswhich includeconsideration of the current and future de-mand-supply relationships and their implica-tions.3 Still others discuss only manpowerrequirements trends and projections, especiallyin those occupational groups where the supplyof workers is difficult to estimate, such asskilled workers and workers in defense-relatedemployment,4

Sources of Data

The projections and other manpower infor-mation developed in the occupational outlookprogram 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 majorsources of statistical and other informationutilized in the program.

The basic statistics on current and past em-ployment in occupations and industries havebeen based mainly on Bureau of Labor Statis-tics household data from the Monthly Reporton the Labor Force (MRLF) and establishmentdata from Employment and Earnings. (Asingle publication starting with the February1966 issue). Use is made also of the scientificand technical personnel surveys conducted bythe Bureau, which contain data on scientists,engineers, and technicians. The decennial Cen-sus of Population is utilized for data on mostoccupations not covered by the limited detailpublished in the MRLF and Censuses of Busi-ness and Manufacturing are used to fill in in-dustry detail. Information from the CivilService Commission is used for data on FederalGovernment workers. These basic sources ofoccupational and industry employment statis-tics are augmented by data from Federal reg-ulatory agencies, such as the Federal AviationAgency and Interstate Commerce Commission,

OCCUPATIONAL OUTLOOK

which collect industry and occupational sta-tistics. In some cases, employment data areobtained from unions, industries, trade associa-tions, and professional societies. In most cases,however, these general statistics serve only asa starting point for the development of thespecific, current estimates needed for a. par-ticular report.

In developing analyses of past and projectedchanges in employment requirementswhichwill be described laterthe outlook programutilizes statistics of output, hours of work, andoutput per man-hour. The major sources of thestatistics used are Bureau of Labor Statisticsstudies of productivity and technological de-velopment, the Federal Reserve Board produc-tion indexes, and the 'U.S. Department ofCommerce output data from the Annual Surveyof Manufactures and the Census of Manu-factures. Industry associations and unions alsooften provide similar types of data.

Estimates of the past and probable futuresupply of workers utilize entirely differentsources of information. U.S. Office of Educationdata on enrollments and degrees in high school,post-secondary schools, and colleges and uni-versities form a major component of the supplyestimates. Bureau of Apprenticeship and Train-ing statistics on apprenticeship, as well asinformation on company training programs,provided through company reports and per-sonal interviews, provide other inputs into thesupply estimates.

Special studies of various aspects of the sup-ply of workers provide essential informationfor the development of the estimates andprojections of supply. A few examples of thesetypes of statistical source materials are occu-pational mobility studies (Bureau of LaborStatistics) ; Tables of Working Life (Bureauof Labor Statistics) ; followup studies of col-lege graduates (National Science Foundation) ;and many other specific types of studies, oftenof a one-time nature. Earnings information,which appears in many of the outlook publica-tions, is drawn primarily from wage and earn-ings survey.; conducted by the Bureau of LaborStatistics, supplemented with additional infor-mation on many occupations from Federal reg-ulatory agencies. Studies of union wage scales

49

are also used. Information is also drawn fromreports 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 othersclosely associated with an industry or occupa-tion; (2) reports and interviews with profes-sional or trade associations and licensingagencies ; (3) union publications and officials;and (4) periodicals, trade journals, annual re-ports, and so on.

Methods of Analysis

The projections of requirements and re-.sources developed for the occupational outlookprogram require varying methods of analysis,usually because of differences in the factorsaffecting a particular occupation or industry,but also because of differences in the amount ofdata available for analysis. The broad patternof research, however, is generally the same inall of the detailed, comprehensive occupationaland industry studies.

The starting point in most studies is ananalysis of the factors affecting the demand forworkers in the occupation, and an assessmentof how these factors may operate in the future.Occupational employment is affected by a hostof factors. Technological change is the mostoften discussed factor affecting occupationalemployment, but occupational changes are alsoinfluenced by other factors, such as growth inpopulation and its changing age distribution,as in the case of teachers. Government policyrelating, for example, to the magnitude of thedefense and space programs and to expendi-tures for research and developmentalso playsa major role. Occupational employment is alsoinfluenced by institutional factors, such asunion-management relationships and practices,as in the case of railroad workers, or by therelative supply of workers in other occupa-tions, as for example, the substitution effectresulting from shortages of engineers and theirreplacement by technicians. Also influencingoccupational employment are changes in the

53


total demand for the employing firm's product ;changes in the level of income and distributionof income among consumers, industry and gov-ernment; and changing patterns of consump-tion.

It is apparent, in view of this multitude offactors, that no one technique can be usedsuccessfully to project manpower requirementsin all occupation and industries. The growthand decline of each occupation is affected byits Own complex of factors. The number ofteachers required, for example, is affected bythe number of pupils (related to birth ratesand trends in the proportion of children at eachage who attend school) and by trends in theratio of teachers to pupils, which depend uponeducation practices and available financing.

Projections of requirements for scientists,engineers, and technicians require considera-tion of different factors. These must take intoaccount such factors as the growing utilizationof technical personnel, the increasing tech-nological complexity of industrial products andprocesses, changes in levels of expenditures fordefense, and growing research and develop-ment activities. Requirements for automobilemechanics are related to the number of newautomobiles and accessories and the age ofautomobiles ; for radio and TV repairmen, tothe number of radios and TV's sold, and theirage and complexity ; for policemen, to popula-tion and urbanization ; for truckdrivers, to im-proved equipment and highways, and forcompeting methods of transportation.

For many occupations., the significant factorsinfluencing employment are the prospectivelevels of demand for the products of the variousindustries in which the occupation is found,and the effect of these changes in demand onemployment in the industries. Among the gen-eral factors which must be considered in ananalysis for an industry study are expectedchanges in the total domestic production ofthe industry's product or service, competitionwith other products or services, expected tech-nological changes, output per man-hour, andchanges in hours of work. More specifically, inprojecting the activity or production level ofan individual industry, it is necessary first toestablish the nature of the demand for an in-

54

dustry's products or services and the relation-ship of this industry to the growth of the wholeeconomy. Obviously, an industry producingproducts directly for consumers will have a dif-ferent type of demand function than an indus-try which is making raw materials to be usedas a component for further manufacturing.

In projecting the production of steel, for ex-ample, consideration must be given to theexpected increase in population and the trendin steel output per capita. The total require-ments for steel depend on the requirementsprojected for each of the principal steel-usingindustries, such as the automobile, construc-tion, electrical appliances, machinery, and con-tainers industries; competition to steel fromother materials, such as aluminum and plastics;and the import-export balance for steel. Ineffect, it is necessary to project the output ofboth domestic and foreign users of steel inorder to estimate total steel requirements.Future industry production or activity levelscan then be translated into overall manpowerrequirements by estimating changes in manhours per unit of output for each industry, andby making assumptions as to changes in hoursof work.

Because of the tremendous amount of re-sources necessary to make an extensive studyof each industry in the economy, a more globaltype of analysis has been used to fill the gapsand to provide an overall framework for theoccupational and industry projections. In mak-ing the analysis for the products of each indus-try, the usual starting point is the total demandof the economy for goods and services ; this canbe apportioned among the requirements foreach major product or service (classified byindustry). The estimates of production canthen be translated into requirements for work-ers in total and by occupation.

The general approach in the development ofthis industry framework is to begin with thepopulation and labor force projections developedby the Bureau of the Census and the Bureau ofLabor Statistics. Assumptions are made as tothe size of the Armed Forces, the level of un-employment, annual hours of work, and outputper man-hour. Multiple correlations are madewhich take into account past employment

OCCUPATIONAL OUTLOOK

trends and relationships, and variables such asunemployment, size of the Armed Forces, grossnational product, and population. By this tech-nique, preliminary projections of manpowerrequirements are developed for each industryfor which adequate data are available.

The results of the multiple regression analy-sis are then used as the vs,:sis for further judg-ment decisions as tc the level of manpowerrequirements in the projected period, par-ticularly for those industries for which detailedindustry studies have been made. The consider-able amount of information on individual in-dustries developed in the occupational outlookprogram over many years and through discus-sion with representatives of industry andunions provides essential background in mak-ing these judgments. Analysis of trends andprojections for the economy as a whole, or forindivdival industries made by other groups,such as the National Planning Association,Stanford Research Institute, State and localgovernments, and universities, also contributeto these judgments. The adjusted overall in-dustry projection framework is then utilizedas a basis for occupational requirements. Theserough 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. (Seedescription of Industry-Occupational Matrix,chapter 7.) In general, it may be said that theprojections are based heavily on judgment asto the effect of the demand factors on specificoccupations.

Projections of changes in manpower require-ments by occupation and industry provide onlyone part of the information on the total numberof job openings which will need to be filled inthe years ahead. In most occupations, moreworkers are needed yearly to fill positions leftvacant by those who leave the occupation toenter other occupations or because of retire-ment or death, than are needed to staff newpositions created by growth of the field. In esti-mating the total number of openings likely toarise in an occupation, the occupational outlookprogram analyzes studies of occupational mo-bility among selected groups of workers, andtables of working life.

51

These tables of working life, which are sim-ilar to the actuarial tables of life expectancyused by insurance companies, provide a basisfor assessing future rates of replacement re-sulting from deaths and retirements, which arein turn affected by differences in sex and aver-age age of the workers in particular occupa-tions. Where men comprise the great majorityof workers, estimated replacement rates fordeath and retirement usually average between1 and 3 percent a year. In occupations in whichwomen predominate, the rate is usually muchhigher, and allowance must be made for thelarge numbers of women who leave paid em-ployment to get married and assume familyresponsibilities but who return to paid employ-ment after marriage or raising a family. Thereplacement rate among elementary schoolteachers, for example, is estimated at 4.8 per-cent a year; many of these teachers return toemployment at a later date.

In appraising the overall employment oppor-tunities in an occupation, estimates are alsomade of the future supply of personnel, at leastin those fields in which the supply is identifi-able. Statistics on high school and college en-rollments and graduations are the chid sourcesof information on the potential supply of per-sonnel in the professions and in occupationsrequiring extensive formal education. Data onnumbers of apprentices and graduates of voca-tional and technical training programs providesome limited information on new entrants intoskilled trades. However, in many occupationsmost new entrants are trained informally,through on-the-job training or company train-ing programs.

It is not enough to know, of course, howmany persons are being formally trained foran occupation, ;ince not all those completingformal training or education in a particularfield enter that field upon completion of theircourses. As a result, special surveys are utilizedto provide additional information on the actualnet supply of workers from a training programor a field of study. These include studies of em-ployment plans of college seniors, job place-ments of college graduates, and jobs enteredafter completion of MDTA and other types oftraining. Limited data on transfers out of an

55

52 BLS HANDBOOK

occupation or re-entries into an occupation arealso utilized, although, in general, data on oc-cupational mobility are available for only a fewoccupations.

The estimates of the future demand in anoccupation is then related to estimates of thefuture supply to develop the employment out-look in that field and to provide information topolicy makers, educators, and others on theimplications of these relationships.

Presentation

The Occupational Outlook Handbook is themajor publication of the occupational outlookprogram. Oriented toward vocational guidance,the Handbook is a basic reference source, pub-lished every other year, which includes com-prehensive and non-technical job informationon approximately 800 occupations and 30 major

56

OF METHODS

industries, covering the entire spectrum ofwhite-collar, blue-collar, and service occupa-tions. An occupational outlook report seriesprovides reprints of individual statements fromthe Handbook.

The Occupational Outlook Quarterly providesa continuous flow of current occupational andjob information between editions of the Hand-book, together with the most recent informationavailable on earnings, training requirements,and other related topics. In addition to thesetwo publications, developed mainly for use invocational guidance, the occupational outlookprogram conducts technical and detailed stud-ies on specific occupations and industries inorder to furnish information to manpower ex-perts, personnel departments, and others in-terested in the more technical aspects of theNation's future manpower needs.

-NEAL H. ROSENTHAL

Chapter 6. Projections of the Labor Force

Background and Uses

Projections of the future size of the laborforce are needed for a variety of planning pur-poses. They provide a basis for establishing theamount of employment growth the economymust generate to maintain high levels of em-ployment. They serve as the basis for one ap-proach in setting goals for a general economicgrowth rate consistent with full utilization ofhuman resources. Projections help to gain in-sight into the characteristics and numbers ofworkers who will be available for industry, andto see what this implies for education, training,and personnel policies. In addition, labor forceprojections, 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 Laboris particularly concerned with the relationshipbetween the expected labor supply and the needfor various skills and training created by ourchanging technology.

Method

Projections of the labor force as a whole andof the separate age-sex groups are made forquinquennial dates usually for about 15 yearsahead. The schedule for preparing the projec-tions has been irregular owing, in part, to thetiming of new projections of the population ofworking age. Labor force projections for 1975,1980, and 1985 were published in May 1970.

Because social and economic factors affectthe supply of labor, certain assumptions needto be made about conditions surrounding anyset of labor force projections. Generally, projec-tions have been made on the basic assumptionsthat past trends in labor force participationwould continue into the future, and that theeconomy would continue to expand and main-tain high levels of employment opportunityconsistent with an unemployment rate of about4 percent. Another usual assumption is thatthere would be no major war or significant

ii

change in the size of the Armed Forces whichmight substantially alter the previous workpatterns of the population. It also presupposesthat the trend toward increased school enroll-ment beyond the high school level, which hasa direct bearing on labor force activity of youngpersons, would continue, supported by adequateschool facilities, staff, and aid to students.

The general approach used in preparing theBureau's labor force projections is to projectthe proportion of the population in each age-sex group or subgroup that is expected to bein the labor force, i.e., the labor force partici-pation rate at the specified future date, and toapply these rates to the expected population ineach group.

In making projections for a given age-sexgroup or its subgroup, the standard procedureis to fit a line or curve to a series of pointsrepresenting the labor force participation ratesfor that group for the years since 1947, and toextrapolate the line or curve into the periodcovered by the projection. The procedure ismodified, as appropriate, to discount the tem-porary effect of factors judged to be operativefor only short periods.

The population projections used in project-ing the labor force are prepared by the Bureauof the Census on the basis of analyzing pasttrends in birth rates, death rates, and net im-migration and projecting these trends. Sincethe birth rates pose the most uncertainty inprojecting the population, the Bureau of theCensus prepares several series of populationprojections on the basis of varying assumptionswith respect to birth rates. The uncertainty ofprojecting birth rates does not directly affectthe level of the labor force projections 15 yearsahead, since everyone of working age (16 yearsand over) at that future date has already beenborn when the projections are made. However,the birth rates do have a bearing on projec-tions of the labor force participation rates ofyounger married women, because mothers ofyoung children are less likely to work. Becauseof this indirect effect, it was necessary to selectthe one series of population projections which

68

-57


seemed most reasonable on the basis of an in-dependent evaluation of past trends in birthrates. For recent labor force projections, series"C" of the population projections published inthe Census Bureau's Current Population Re-ports, P-25, No. 381 was chosen.

The overall size of the labor force is built upby age and sex, not only because the composi-tion is needed for many of the purposes notedearlier, but also because the degree of laborforce participation varies among the differentage-sex groups, and the historical trends inthese rates also vary. Some of the factors whichhelp to explain the behavior of the labor forceparticipation rates and which affect particulargroups include school attendance, maritalstatus, birth rates, and the availability of socialsecurity benefits and the expansion of privatepension plans. The method of projecting thelabor force participation rates for the variousage-sex groups takes into account the influenceof a number of these specific demographic andsocial factors. For example, projections of theproportion of persons enrolled in schools in thevarious young ages are used to subdivide thefuture population of young persons into thosewho are expected to be in school and those notin school. The population of married women inages 20 to 44, by age, is grouped by those whoare expected to have children of preschool ageand those with no children under 5, on the basisof projected trends in fertility and child spac-ing. Similarly, projected marital status distri-butions of older adult women are used toprovide the future numbers in each maritalcategory within each age.

For each of these subgroups, the projectedlabor force participation rates are applied totheir respective future populations and the re-

suiting labor force summed to provide the totallabor force for each age-sex group and for allages.

Sources of Data

The source of the basic historical data onlabor force participation rates by age and sexused to project the labor force is the monthlystatistics on the labor force. These are pub-lished by the Bureau of Labor Statistics andare based on the Current Population Survey ofthe Bureau of the Census. Historical data onlabor force activity by various categorieswithin several of the age sex groups are ob-tained from the recurring supplementary laborforce surveys also based on the Current Popu-lation Survey. These include information fromthe October surveys of the employment ofschool-age youth and the March surveys of themarital and family characaterisitics of workers.

The population projections are the latestavailable projections made by the Bureau of theCensus and published in their Current Popula-tion Reports, Series P-25. Projections of schoolenrollment and marital status, by age, are basedon published and unpublished data of the Bu-reau of the Census. Data used in projecting theproportion of women in each age group whowill have children under age 5 years includepublished and unpublished data on birth rates,by age of mother and order of birth, from theDivision of Vital Statistics of the Public HealthService ; fertility and marriage data from re-ports of the Bureau of the Census, CurrentPopulation Reports, Series P-20, and data fromthe decennial censuses of population.

Technical ReferencesNumber

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).

SOPHIA COOPER TRAVIS

5s

Chapter 7. Industry-Occupational Matrix

Background

The Bureau of Labor Statistics has developeda comprehensive set of data on the occupationalemployment composition of all industry sectorsin the economy. Presently, industry-occupa-tional matrices are available for 1960, 1967,1970, 1975, and 1980. These data are set up toform a matrix, or table, of 162 specific occupa-tions plus groupings of occupations cross-classified with 116 industries. Thus, the occupa-tional pattern of each industry is showni.e.,the proportion of each occupation to total em-ployment in an industry. Looked at anotherway, the tabulation shows how total employmentin an occupation is distributed by industry.

Initially, work on the Industry-OccupationalMatrix grew out of concern by the Departmentof Defense for anticipating the economic prob-lems that might arise from various defenseprograms. The first set of tables related to1950 and were prepared by the Bureau as apart of the interindustry program of the early1950's, sponsored by the U.S. Department ofthe Air Force. That program was terminatedin 1953, but the 1950 matrix and its successorscontinue to provide the basic information foremergency manpower planning, now carried onby the Office of Emergency Preparedness. Inrecent years, a strong interest has developed indetermining manpower needs for other pur-poses. The latter have included training newworkers, retraining workers displaced by auto-mation, and providing information to highschool counselors and to students making careerdecisions. The change has focused increased at-tention on the need for estimates of numberspresently employed in specific occupations andthe likely future employment requirements byoccupation. The Industry-Occupational Matrixprovides a systematic approach to developingthe desired information.

Sources of Data

Data for the Industry-Occupational Matricesare brought together from a wide variety ofsources. A major source for the development ofthe 1960 matrix was the Occupation by Indus-.try report from the 1960 Census of Population.The Current Population Surveys (CPS) are thesource for total employment, employment forbroad occupational groups, and for a few large,specific occupations.1 Other sources of occupa-tional employment data included the Bureau ofLabor Statistics annual surveys of occupationalwage rates in metropolitan areas and selectedindustries ;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 occupationsbased on licensing data and membership rec-ords of professional societies; and surveys ofemployers by the Bureau and other agencies toobtain estimates of employment in a limitednumber of highly important occupations suchas scientists, engineers ,3 teachers, and police-men.

Specific estimates from sources other thanthe Census were incorporated into the cells ofthe matrix for about 16 million workers, orone-fourth of all those who were ,employed in1960. The remaining details in the-matrix werederived by forcing 1960 population census esti-mates for detailed cells (published in Occupa-tion by Industry) into agreement with controltotals for occupational groups and industriesfrom sources other than the Census. The occu-pational control totals were average annualemployment by occupational group taken fromthe CPS. Most of the industry employment

See chapter 1.I See chapter 14.

See chapter 4.

59'55


totals were based on BLS estimates of privatewage-and-salary workers adjusted to includethe self-employed, unpaid family workers, andgovernment workers, and to exclude the sec-ondary jobs of dual job holders. Total employ-ment in agriculture and private households wasbased on CPS estimates. The adjustments ofthe matrix to consistency with CPS estimatesof total employment and industry employmentestimates, derived as described above, bringsthe matrix for 1960 into agreement with dataused as the basis for the Bureau's projectionsof total employment and occupational employ-ment by industry. The Bureau's occupationalprojections are reflected in and developed inpart through matrix techniques. (See sectionon analysis and uses.) The 1975 matrix * wasdeveloped by examining a variety of historicalstatistics on the changing occupational struc-ture of industries and data from the 1950 and1960 censuses and evaluating the factors likelyto influence changes in the future such as ex-pected new technology, changes in product mix,and the general organization of industries. Asimilar procedure, together with use of infor-mation for the 1960-67 period, was used inpreparing the 1980 matrix.

The 1960 matrix provided the base for the1967 and 1970 matrices.5 Where available, oc-cupational data from other sources, such asthose cited above, were incorporated into theupdated matrices as fixed cells. For the re--maining cells, first approximations of the oc-cupational patterns for 1967 were made byinterpolating between the p4terns of the 1960and the 1975 matrices. The resulting patterns(in mining and manufacturing) were thenbrought into consistency with data on produc-tion worker trends available from the Bureau'sCurrent Employment Statistics program. Thepatterns were then applied to individual indus-try employment controls and summed up toarrive at occupational totals. These occupa-tional control totals were then compared todata from the CPS and other sources of infor-

,For a detailed description of the procedures followed indeveloping the 1975 matrix, see Occupational EmploymentPatterns, 1960 and 1975 (BLS Bulletin 1599, 1968).

For a more complete discussion of the methods used indeveloping the 1967 and 1970 matrices, see OccupationalEmployment Statistics 1960-67, BLS Bulletin 1643.

mation. When necessary, certain occupations(except for fixed cells) were then forced on aprorated basis to predetermined occupationalcontrol levels. This iterative forcing procedurewas repeated until the internal matrix cellswere consistent with both the industry and theoccupational controls. A similar procedure wasfollowed in preparing the 1970 matrix with theexception that the 1967 matrix provided thebasic data file of occupational ratios used forthe iterative forcing procedure. Thus, boththe 1967 and the 1970 industry-occupationalmatrices were consistent with' (a) national em-ployment by industry, (b) broad occupationalemployment levels from the CPS, (c) trends inproduction (and nonproduction) worker em-ployment by industry, (d) anticipated trendsin occupational structure within industries, and(e) reliable estimates of detailed occupationalemployment available from the CPS and othersources.

Analysis and Uses

A basic objective of the project is to haveavailable a comprehensive set of data on in-dustry-occupational relationships which can beused in projecting manpower requirements byoccupation. Although statistics on employmentby occupation are relatively thin, particularlybetween decennial censuses, there is a greatdeal of information on total employment in de-tailed industries. Each industry utilizes aunique combination of occupational skills, to-gether with other factors of production, in itsefforts to achieve least cost for its output. Oc-cupational patterns may be markedly differentfrom 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 inrestaurants is, largely made up of waiters,waitresses, cooks, and owner-managers. Overperiods as short as a decade, the occupationalstructure of many industries is relativelystable. Consequently, if good information isavailable on the occupational composition ofindividual industries for a base period, it can

INDUSTRY-OCCUPATIONAL MATRIX

be used together with the available statisticson changing employment in each industry todevelop estimates of current employment byoccupation for later periods. Further, if pro-jections of output and employment are avail-able by industry, the base period occupationalratios applied to the industry employment pro-jections will yield initial estimates of employ-ment requirements by occupation for futureperiods.

Although the occupational patterns of manyindustries are relatively stable over periods of5 to 10 years, it is clear that occupational pat-terns change with the advance of technologyand changes in the supply of workers in eachoccupation. Hence, information on how tech-

57

nology and labor supply are changing the oc-cupational pattern in each industry is used tomodify the initial estimates. This improves theestimates of current employment by occupationand of future employment requirements by oc-cupation, developed by applying base periodindustry-occupational ratios to industry em-ployment estimates. Changing technology andother factors which affect skill requirementsare constantly being studied in order to esti-mate the future occupational structure of eachmatrix industry. The adjusted occupationalpatterns are then used, together with projec-tions of employment by industry, to prepareestimates of future employment requirementsto 1980.

61

-RICHARD DEMPSEY

Prices and Living Conditions

Chapter 8. Consumer Expenditures and Income

Background

Consumer expenditure surveys 1 are specializedfamily living studies in which the primary em-phasis is on collecting data relating to familyexpenditures for goods and services used in day-to-day living. Expenditure surveys of the Bureauof Labor Statistics also include information onthe amount and sources of family income, onchanges in savings or debts, and on major demo-graphic and economic characteristics of familymembers.

The Bureau's studies of family living conditionsrank among its oldest data-collecting functions.The purpose of the first nationwide expendituresurvey in 1888-91,2 in line with the legislationcreating the Bureau, was to study the worker'sconsumption habits and living costs as elementsof production costs, with special reference to com-petition in foreign trade. It emphasized theworker's role as a producer, rather than as a con-sumer. Purposes and coverage changed in suc-cessive surveys, and problems caused by risingprices led to the second survey, during the year1901. The index of prices of food purchased byworkingmen, with weights based on the 1901 data,was used generally, as a deflator for workers' in-comes and expenditures for all kinds of goodsuntil World War I. The third major survey,spanning 1917-19, provided weights for comput-ing a "cost-of-living" index, now known as theConsumer Price Index (CPI). (See chapter 10.)The next major study, for 1934-36, was madeprimarily to revise these index weights and cov-ered only urban wage and clerical workers.

However, in the severe economic depression ofthe 1930's, interest in consumer surveys expandedfrom study of the welfare of selected groups togeneral economic analysis. Thus, almost simul-taneously with its 1934-36 investigation, the Bu-reau cooperated with four other Federal agenciesin a fifth survey, the Study of Consumer Pur-chases, in 1935-36, which undertook to show con-sumption of all segments of the population, bothurban and rural.' The Bureau's sixth major sur-

vey, for 1950, covered all urban consumers. Itprovided the basis for revising the Consumer PriceIndex (CPI) and also supplied abundant materialfor broader types of eemomic and market analysis.The remainder of this chapter deals with the1960-61 surveythe latest in the series* describingchanges in the consumption habits of the Az iericanpeople.


The basic orientation of the Bureau's most recentsurvey was to obtain detailed information for re-vising the CPI. The increasing need for consumerexpenditure and income data for other purposeswas taken into account in planning the survey ofurban families in 1960 and 1961. Then, in coop-eration with the U.S. Department of Agriculture(USDA), the 1961 coverage was extended to ruralareas. Thus, for the first time since 1941, informa-tion on spending habits became available for across-section of the entire noninstitutional popula-tion in urban, rural nonfarm, and rural farm areasof the United States. Concepts, techniques, andpublications for the 1960-61 survey were plannedto provide as much continuity and comparabilityas possible with the Bureau's 1950 and earlier ex-penditure surveys.

11n this chapter, the initials CBS are used to refer to theprogram of consumer expenditure surveys.

The State of Massachusetts conducted the first expendituresurvey in the 'United States. This investigation of living con-ditions In 1874-76 undertook to measure the welfare of theworkingman's family before and after migration to the 'UnitedStates. For a more detailed account of expenditure surveys inthis country, see 'U.S. Bureau of the Censue, Historical Statisticsof the United StatesOolonial Times to 1957, Chapter G (1980).

The Bureau oleo cooperated with the 'U.S. Department ofAgriculture In a smaller scale nationwide survey of urban andrural families in 1941-42 to obtain facts nn which to base deci-sions for the civilian economy during wartime. In addition, theBureau conducted a Survey of Prices Paid by Consumers in 1944among a nationwide sample of urban families ; as a byproductof the reports required for the analysis of prices, certain data onfamily 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 .forspecialised studies.

,%/69

60 CONSUMER EXPENDITURES AND INCOME

All data were collected through the voluntarycooperation of families." The family, or consumerunit (CU), referred to (1) a group of peopleusually living together who pooled their incomeand drew from a common fund for their majoritems of expense," or (2) a person whose incomeand expenditures were not pooled with others,whether living alone or in a household. However,never-married children living with parents alwayswere considered as members of the parents'consumer unit.

Information was recorded for the family as com-posed in the survey year, including part-yearmembers. Family members were not eligible forperiods 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 netchanges in the family's assets and liabilities dur-ing the year. The estimated value of goods andservices received as gifts or otherwise, withoutdirect expenditures by the family, was requestedalso. To supplement the annual data, familieswho prepared meals at home provided a detailed7-day record of expenditures for food and relateditems purchased frequently.

For selected items of clothing, housefurnishings,and food, the record of expenditures was supple-mented by information on quantities purchasedand prices paid. Characteristics of the housingoccupied by homeowners and renters and an inven-tory of the major items of housefurnishings theyowned were recorded.

To permit more meaningful analysis of thespending habits of American families, limiteddemographic information was obtained. This in-formation included the sex, age, years of schoolcompleted, occupation, race, and marital status ofeach family member.

Data Sources and Collection Methods

All data were collected by personal interview.The BLS was responsible for collecting data fromall residents of urban places. The BLS andUSDA shared this responsibility in the rural areas

of Standard Metropolitan Statistical Areas(SMSA's) 8 and the USDA had sole responsibilityfor interviewing rural households in nonmetro-p olitan area.s.5

Field Organization

To reduce the size of the staff to be recruitedand trained and to utilize this staff over a longerperiod, it was decided that the urban survey wouldcover 2 years, 1960 and 1961. For similar reasons,the surveys for each year were conducted in two"waves." As field work was completed in thelargest SMSA's, supervisors were reassigned tosmaller places. Interviews for the 1960 and 1961CES were conducted in the spring and summerof 1961 and 1962, respectively.

The supervisory field personnel were recruitedby the BLS Regional Offices " and brought toWashington for 6 weeks of intensive training onthe purposes of the survey, survey techniques, andschedule content. These supervisors went to anassigned city where they, in turn, recruited inter-viewers, whom they trained for about 8 days.

Questionnaires

The detailed questionnaires used by the BLSagents in interviewing families in the 1960-61 sur-vey had been tested in surveys for 1959 in threecities. They i.ncorporated modifications based onthis experience. Three forms were used in thenonfarm surveys. Schedule A was a two-pageform to determine the family's eligibility forthe survey and, for families who refused or wereunable to participate in the survey, it provided arecord of minimum data for the analysis of non-response. Schedule B, on which the interviewerentered the complete annual record of the family'sliving arrangements, income, spending, and

*See discussion of bureauwide policy on voluntary reportingand confidentiality, wider Introduction.

This category includes children temporarily away from homeat school or college.

*The tabulations published in the CBS reports listed in thetable at the end or this chapter include only inn-year consumerunits. i.e., units with at least outs member who Was eligible overthe entire survey year. In addition, the Bureau obtained sched-ules from approximately 400 part-year CV's, for special analyticalresearch.

See appendix B for description of SMSA classification system.See discussion of 'sample design, p.

" The following description refers to procedures of the BLS,but USDA procedures were similar.

63

BLS HANDBOOK OF METHODS 61

ambiguous entries or to complete a record." Onthe average, the interviewer spent 7 to 8 hours witha family in a series of visits arranged at thefamily's convenience.

changes in savings, consisted of 59 pages andformed the basic framework of the survey. Sched-ule C provided 15 pages for a detailed report ofthe family's purchases of food, beverages, tobacco,personal care, and household supplies in the weekpreceding the interview.". The coding system forsummarizing and classifying the data was devisedwhile the 1960-61 schedules were being designed,and tabulating codes were printed on the schedules.The precoded data were then readily transferredto punch cards and magnetic tapes for tabulating.

Information in schedule 73 was grouped in 23sections, placed in a sequence so as to establishand maintain rapport between the interviewer andthe respondent. The detailed checklists of items 12in each section were included not only to facilitaterecall, but to provide the specific informationneeded to determine CPI weights. However, allsections 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 recordswhenever possible.

Reported receipts and disbursements were sum-marized and reviewed in the field to determine thecompleteness, consistency, and balance of the fam-ily account. Families were reinterviewed whenthe field supervisor deemed it necessary to clarify

31 A modification of schedule B, providing for detailed reportingof farm receipts, disbursements, changes in farm assets, andvalue of home-produced food, but otherwise paralleling the non-farm schedule. was used by the U.S. Department of Agriculturein rural farm areas. Schedule C was not used in interviewingfarm families.

**For a discussion of global versus detailed questions and"free-listing" versus "check-listing," see report by Helen HumesLama le, Methodology of the Survey of Consumer Expendituresin 1910, Philadelphia, Pa., University of Pennsylvania, WhartonSchool of Finance, /959, pp. 18--19.

**Prior to editing the schedules for machine tabulation, theschedules were reviewed in the Washington oftleee of the BLSand USDA primarily to determine that entries conformed to thesurvey concepts and methodology. This review was concentratedon sections of the schedule which had proved most difficult inprevious surveys, and on unusual situations which requiredspecialized instructions. Occasionally, schedules were returnedto the field for clarification or additional information, and someschedules considered complete in the field were rejected in Wash-ington. If a schedule met the test of the review and editinginstructions with reepect to internal completeness and com-!latency of expenditures with each other and the family's reportedmanner of living, the record was used even though there was asubstantial lack of balance between the family's reported totalreceipts and disbursements.

14 This procedure involves the probability selection of a sample"pattern" from a let of patterns purposively established so that,taken as a group, they give each primary sampling unit itsproper chance of appearing in the final sample. The selectionof the city sample is described in "Technical NoteThe RevisedCity Sample for the Consumer Price Index," Monthly LaborReview, October 1960, pp. 1078-1083. (BLS Reprint 2352.)

Sampling

Separate stratified samples were selected forurban areas, rural areas in metropolitan counties,and rural areas in nonmetropolitan counties. Athree-stage sample design was used within eacharea to obtain a sample of consumer units repre-sentative of all U.S. consumer units as definedfor this survey.

In developing urban sampling plans, continuedrepresentativeness of the sample for measuringnational changes in consumer prices was of para-mount importance. Consideration of probable re-sources led to setting 66 as the maximum numberof cities for the CBS sample. Tests of the ef-fectiveness of some of the more obvious modes ofstratification indicated that no elaborate stratifi-cation was justifiable for so small a sample ofcities. In general, classifications by geographicregion and size of city seemed to be most effective,especially since an important objective in selectingspecific cities was to achieve good geographic dis-persion. For this purpose, the BLS utilized the"controlled selection" procedure." The primarysampling unit was the Standard MetropolitanStatistical Area (SMSA) in the metropolitan seg-ment of the United States and the individualurban place in nonmetropolitan areas.

Sample Design for Three Urbanizations

In the first stage of the design for the urbansample, all SMSA's and nonmetropolitan urbanplaces were classified by population size andregion. A sample of 66 places listed in the tablewas selected to represent all urban places in the50 States. All of the 12 largest areas in the UnitedStates automatically were included. For NewYork and Chicago, the Standard ConsolidatedAreas, rather than the constituent SMSA's, wereused as piimary sampling units. However, in thecollection and analysis of the data, the New York-.Northeastern New Jersey Standard ConsolidatedArea was divided into two subareasNew York,N.Y. and Northeastern New Jersey. All of theselargest areas were surveyed in both years with data

64


SURVEY OP CONSUMER MCP/MD=11ES, 1980-61

(SUMMARY OF SAMPLE SIZE AND AVAILABILITY OF DATA FOR URBAN AND RURAL AREABt BY GEOGRAPHIC REGION, POPU-LATION STRATUM, AND SMSA OR OTHER URBAN PLACE)

Region, population stratum, SMSA, or other urban place

Number of-

Assignmentaddresses

Usableschedules

Total urban and ruralUnited StatesNortheast.

Northh Central

West.

Rural (arm United StatesNortheastNorth CentralSouthWest_.

Rural nonfarmUnited StatesNortheastNorth CentralSout.Wost

UrbanUnited StatesNortheast.

SAHA, population 1,400,000 and overBoston, MassNow York, N.YNortheastern New Jersey.Philadelphia PaPittsburgh, 1 a

SM8A, population 250,000 to 1,400,000Buffalo, N.YRadford, Conn

MBA, population 50,000 to 250,000Portland, MaineLancaster, PIL.

Nonmetropolltan urban place, population 2,500 to 50,000Burlington, VtKingston, N.YLewistown, Pa--Athol, MassMUlvUle, NJSouthbridge, Masa

North Central133d8A, population 1,400,000 and over

Chtmgo, IllCleveland, OhioDetroit, ktMSt. Lonis, Mo

SM 8A, population 250,000 to 1,400,000Ind apolls, IndDayton, OhioWichita, Zang

SM8A, population 50,000 to 260,000Cedar Rapids, Iowa

U rChampaign-bana, 111Green Bay, Wis

Nonmetropolitan urban place, population 2,500 to 50,000Devils Lake, N.DakFindlay, OhioLaSalle, UlNiles, MichOwatonna, MinnCambridge, OhioCrookston, MinnLogansport, IndManhattan, KensMenasha, *Ls

SouthMBA, population 1,400,000 and over

Baltimore, MclWashington, D.0

SMSA, population 250,000 to 1,400,000Atlanta, GsDallas TerNashville, Tenn

SMSA, population 60,000 to 260,000Austin, Tat ._Orlando, Fla.Baton Rouge, LaDurham, MO

See footnotes at end of table.

1

1

17,283

4,1 1275,875, 100

! 2,869

2,581214980

1,181206

2,497453702999343

12,2053,460

375626

850075375

260260

160160

656565656565

8,505

875.875875

260250250

16600

160

65656565656565656565

2,920

875876

250250260

160160160160

1 13, 7283,2234, 0024,180

12,228.

1,967145742925155

2,245406628948803

9,4762,677

268448366

323813

199175

136151

5247

so41

as63

2022

371294290319

173180189

125126130

49asas61484361ao4558

2,807

818323

198

210781

110106112165

65

Survey year Publications

1960 1961 BLS reportnumber

Supplements

!SI

(5)

()

(5)

------

()

(5)

(a

i:s

(5)

()

237-93237-89237-90237;91237-n

(11)

237237--84

88

237-85237-88237-87

237-38237-34

237-7 and 57237-4 and 54237-18 and 68

_ 237 -8 and 68237-11 and 61

237-18237-68

237-1428744

237-24237-24237-24237-71287-74237-74

237-35

237-6 and 85237-21 and 71237-1 and 51237-15 and 65

237-102374023747

237-17237-2.1237-73

237-27237-27237-27237-27237-27237-77237-77237-77237-77237-77

287-36

237-16 and 66237-3 and 53

237-6237-20237-56

237-12287-1923742287-69

1, , 31,22,8112,31.2,31,2,8

1,,1,22,3

3

42,21,2,3

42,31,2,3

$111s1$111

s111

1181

1,2,8

111111

1,2,31111


SURVEY OF CONSUMER EXPENDITURES, 1960-61Continued

(SUMMARY OF SAMPLE SIZE AND AVAILABILITY OF DATA FOR URBAN AND RURAL AREAS, BY GEOGRAPHIC REGION, POPU-LATION STRATUM, AND SMSA OR OTHER URBAN PLACE) Continued

Region, population stratum, SMSA, or other urban place

Number of Survey year Publications

Assignmentaddresaes

Usablesobechfles

1960 1961 BLS reportnumber

Supplements

Urban United StatesContinuedSouthContinued

Nonmetropolitan urban place, population 2,600 to 50,000Cleveland, TennGriffin, GoMcAllen, Tex.Reserve, LaUnion, s.cVicksburg, MissFlorence, AlaGainesville, TexMangum, OklaMartinsville, VaOkmulgee, OklaSebring, Fla

WestSMSA, population 1,400,000 and over

Los Angeles-Long Beach, CalifBan Francisco-Oakland, Calif

SMSA, population 250,000 to 1,400,000Seattle, WashDenver, ColoHonolulu, Hawaii

EalSA, population 50,000 to 250,000Bakersfield, Calif

Nonmetropolitan urban place, population 2,600 to 60,000Anchorage, AlaskaGallup, N.MexKlamath Falls, OregEureka, CalifOrem, Mali

es056565esaseses6563eses

12,820

500

250250260

160

27565as65es

430188645065546650554857

11,770

888302

209204215

120

184

444264

()()()

()

i'l

(()()

(,)()

()

(I)

()()

237-25237-25237-25237-25237-25237-25237-75237-75237-76237-75237-75237-75

23747

M7-29 and 72237-2 and 62

237-9237-69237-78

237-70

22N-269237-26237-76237-76

4/:1

2,2,0

s Is 1

1I 1s 1

n

Asterisk indicates year of survey'Includes Anchorage, Alaska, whle.11 was surveyed for 1959.

Published by the U.S. Department of Agriculture.I Supplements 2 and 3 not to be publisbed; for a limited time photocopies of

tables may be obtained from the Bureau of Labor Statistics at a nominal cost.4 No supplements available.

Survey for 1903.

collected from half the sample of consumer unitseach year. Half of the remaining sample ofsmaller SMSA's and urban places was surveyedeach year, i.e., for 1960 and 1961.

At the second stage, a sample of housing unit(living quarter) addresses was obtained in eachcity and in the entire urban part of each SMSAfrom the Comprehensive Housing Unit Survey(CHITS) conducted by BLS in the fall of 1960 16or 1961. The BLS housing unit surveys ,werebased on area block samples designed to representall nonirstitutional living quarters (including non-transient accommodations in hotels and roominghouses) in the urban area.

Because of the shortage of time, the BLS did not con3uct aCH178 in 1960 in places with population of 2,500 to 50,000. The1960 sample for these cities was selected from 1960 Census, ofHousing and Population listings of living quarter addressesArranged by Enumeration Districts (ED's).

se These classifications, which were on the basis of Censusdefinitions, made it possible for the BLS to refer addresses of allhouseholds meeting the Census definition of farm operator orfarm resident to the USDA for inclusion in the rural sample.

NOTE: The Bureau also has published reports for the following special-cityenrveys: Cincinnati, Obio, 1959 (Report 237-28); Fairbanks, Alas1 1969(Report 23740); Xeicbikan, Alaska, (Rvort 237-31); Juneau, alca,1960 (Report 2137-39); Las Vegas, Nev 1962 (Report 237-33); Houston, Tex.,1963 (Report 237-83); Kansas City, Kans.-Mo., 1963 (Report 237-70); Mil-waukee, Vas., 1963 (Report 23740); Minneapolis-St. Paul, Minn., 1963;Report 237-81); and Ban Diego, Calif., 1963 (Report 237-82).

In the third stage, a subsample of addresses wasselected from the housing unit addresses obtainedin the OHUS or Census, arranged by a numberof characteristics, e.g., household size.

The first stage in selecting the rural sample inmetropolitan areas utilized all 34 SMSA's selectedfor the urban sample. In the second stage, BLSconducted 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 smallersegments or blocks in the designated ED's. Eachhousing unit was visited and classified as farm ornonfarm, and as to whether the family includeda farm operator." In the third stage, subsampiesof rural nonfarm and rural farm housing unitaddresses were selected from the RHUS listing byapplying a ratio based on census data on ruralfarm and rural nonfarm households in eachstratum.

66


In the first stage of the USDA's design for therural sample in nonmetropolitan area8, countieswere grouped by State Economic Areas 27 into 126strata equal in weighted counts of rural farm andrural nonfarm dwellings, as the same sample ofcounties was to be used for both farm and nonfarmhouseholds. For each stratum, one county waschosen at random with a probability proportionalto its weighted count. Counties were selectedfrom 41 States. At the second stage, within eachsample county, a selection of rural segments wasmade separately from rural places (100 to 2,500inhabitants) and the open country. Addresses ofall housing units in these segmerits were listed andclassified as farm and nonfarm. \Farm operatorsalso were identified. In the third stage, sub-samples of nonfarm and farm housing unitaddresses were selected from the survey listings.

Sample Size

The master sample for the total urban and ruralpopulation included 17,283 living quarter ad-dresses which were assigned to the interviewers."Usable schedules were tabulated for 13,728 con-sumer units. The distribution of assignmentaddresses and usable schedules by urbanization,geographic region, and for individual metro-politan areas or cities in the urban sample is shownin the table.


To describe the spending and saving of all fami-lies. in the United States, data from the CESsamples have been combined to obtain regionaland U.S. levels." This information was sum-marized for each level of urbanization and for theentire population by using a system of weightsbased on the 1960 Census of Population.

To obtain the weights, adjustments were made inthe Census total of persons in the population onApril 1, 1960, to correct for definitional differencesbetween the Census and the CES universe. Theinstitutional population and on-post military per-sonnel, which were not included in the CES, werededucted from the Census population. Since theCES data apply to the full survey year and familysize is measured in year-equivalent persons, whilethe Census data are a count of persons on April 1,

67

1960, the Census data were adjusted to take accountof births, deaths, and net civilian migration dur-ing 1960. For the 50 States, the net effect of theadjustments was to lower the population total from179,325,671 to 177,391,360.

The total adjusted population was distributedamong the sampling stratein accordance with thedistribution of the unadjusted population. Thepopulation represented by each surveyed area wasdivided by the average family size in the area, asdetermined from the survey, to obtain the totalnumber of families represented by each area. Theestimated number of consumer units in the universefor the United States was 55,306,253.

The adjusted 1960 population was used as urbanweights for both 1960 and 1961. Weights werecomputed for 67 urban strata,2° including Anchor-age, Alaska, which was surveyed for 1959. (Seetable.) Rural nonfarm weights were computedfor 42 strata-34 SMSA's and a farm operator andnonoperator stratum for each of the four regions.The rural farm sample was designed to be self-weighting within regions. Sample averages forthe four regions were combined to U.S. levels bythe application of weights consistent with thoseweights used in the urban and rural nonfarmparts of the CES.

In applying the weights to the stratum aver-ages, to obtain U.S. and/or regional averages, theblow-up factor for each class (e.g., income group,family-size class, etc.) was the number of consumerunits in the universe represented by each samplefamily in a stratum multiplied by the number offamilies in the sample for that class. The result-ing numbers of consumer units became the multi-pliers in calculating stratum aggregates which

11 See appendix B."The address following each address in the maker sample was

picked ae an alternate. The alternate address was substitutedif the master address could not be located or the unit wasvacant, if no one was at home after at least 2 visits, or if theoccupant refused or was unable to give the minimum informa-tion required for classifying the family.

If the interviewer found more than one consumer unit livingat an address, each unit was included in the sample. In hotelsand apartments, the address given to the interviewer indicateda specific housing unit or units within the structure.

"The samples were not designed to provide tabulations byState.

ag For a single year such as 1901, the city weights differedfrom the 1900-81 weights, in that cities surveyed in that yearcarried the entire weight for their respective region city -sirestratum in the 1981 tabulations. In combining 1900 and 1901,each year's sample represented approximately half of the ad-justed population. The 1959 data for Anchorage were weightedinto the combined 1900-01 tabulation for the West and theUnited Staters, but not into the tabulations for 1960 or 1981.

BLS HANDBOOK OF METHIMS 65

were combined to regional and U.S. levels. Theregional 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 andregion listed in the table. All dollar values (in-come, expenditures, and changes in savings) wereshown as averages per family (i.e., consumer unit).As city tabulations were completed, they wereexamined for reasonableness and internal consist-ency, and were compared with tabulations forother cities in the 1960-61 CES sample and withinformation from independent sourcesprinci-pally the 1960 censuses of population and housing.Similar analytical comparisons were made for theregional and U.S. tabulations at each level ofurbanization. For the urban sample, trends since1950 were analyzed for cities which were in theCES sample for both periods. Each report con-tained brief analytical and interpretative text, plusdefinitions and statements on methodology.

The basic reports (see column 5, table) pre-sented averages for major components of familyaccounts for consumer units classified by fivecharacteristics : Family income after taxes, familysize, age of family head, occupation of the End,and housing tenure. Supplement 1 presented thesame information, classified by four additionalcharacteristics: Education of the head, race, fam-ily type, and number of full-time earners. At theregional 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 cityand other location.

Data for eight family characteristics in theabove summaries were cross-classified (two vari-ables) with each of other selected characteristics,as follows :1. Family size with income, age of head, family type and

location and size of place.2. Age of bead with income, occupation of head, and

tenure.8. 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.

The two-variable tables are published as Supple-ment 2 to the basic reports.

Supplement 3 presents in detail the componentsof consumer expenditures, income, and changes insavings, which were summarized in the basic re-ports and Supplements 1 and 2. To illustrate, thecategory "automobile transportation" is brokendown into 10 subgroups of expenditures. Thesedetailed tabulations provide data for consumerunits cross-classified by family size and incomeafter taxes, and by family size and location of thefamily's residence inside or outside SMSA's.


From its inception, the 1960-61 CES wasplanned to serve a variety of purposes. To thisend, questionnaires and tabulation plans Werecirculated among a number of Federal agenciesfor comment. The BLS also formed the Con-sumer Expenditure Advisory Committee, repre-senting academic, private research, and marketingusers, and consulted with this Committee on a widerange of decisions affecting the CES. To date,the results have been used for only some of thecontemplated purposes.

As stated earlier, the primary objective of the1960-61 CES was to revise the CPI (chapter 10).The Bureau also uses the data to revise and expandits work in deriving standard quantity budgets forselected types of families (chapter 9) . As timeand resources permit, the Bureau proposes to drawupon this fund of consumer information to developa broad program of family living conditionsstudies.n

The 'U.S. Department of Commerce relies onthese family expenditure studies as the sole sourceof information for revising its benchmark esti-mates for a number of components in the householdsector of the national accounts. The InternalRevenue Service used the published 1960-61 datato revise its tables of State sales tax payments, forguidelines to taxpayers in filing their Federal in-

"The Bureau has initiated a series of special analytical re-ports (BLS Report 288-1, et. seq.) based oh the 1980-81 CEP.A list of reports in this series is available imon request.

The Bureau also prepares methodological', monographs for theexpenditure surveys which compare national aggregates of in-come, vxpenditures, and savings derives from the CDS withaggregates developed for the national income and product ac-counts by the U.S. Department of Commerce and with data fromother 618ax.


come tax returns. Currently the Department ofHealth, Education, and Welfare is studying thedata in connection with numerous welfareanalyses.

A number of universities utilize the Bureau'sbasic data in a variety of research projects. Thepotentialities of the survey results in consumermarket analysis are evident from reports issuedby the National industrial Conference Board, theChamber of Commerce of the United States, and avariety of trade publications and businessorganizations.

Data obtained from a sample survey as complexas the survey of Consumer Expenditures are sub-ject to many types of errors. These includesampling, recording, and processing errors, anderrors due to the refusal or inability of some fami-lies to give the information requested.

All data were reviewed, edited, and screened tominimize processing errors. Chance variationsdue to sampling can be measured statistically, andthe BLS has published preliminary rough esti-mates of sampling error in the urban sample, in itsreport for the urban United States. Preparationof more detailed estimates for both the urban andrural segments will depend upon the availabilityof resources. Each report contains a section alert-

ing users to the cautions that must be exercised inusing averages based on small samples and pro-vides either the actual number or the basis fordetermining the number of families on which theaverages in each table are based. Approximately78 percent of the national sample of urban andrural families furnished usable schedules andsome of the nonrespondents supplied limited infor-mation on family characteristics." Among theparticipating families, inaccurate reporting is asource of error, despite continued research inschedule design and intensive training of the inter-viewers. Such inaccuracies result from memoryerrors, misunderstanding of a question or reluc-tance to answer it, and incorrect entries by theinterviewer. Although the BLS and USDA haveaccumulated substantial knowledge about suchreporting errors and will continue research inthis field, these errors cannot be quantifiedsatisfactorily.

"This information is used in the monographs on methodology(see footnote 21) to evaluate the nature of the sample lossesdue to nonresponse. It has been the Bureau's practice not toattempt to introduce in the basic CBS tables adjustments fornonresponse or for the underreporting which is common in vir-tually an compilations of income statistics. Such adjustmentspresent many difficulties when data are cross-classified as ex-tensively as are the CBS tabulations.


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 aa the role of the consumergained importance in, economic theory in the mid-1930's. Discusseel problems of concepts,definitions, sampling, and data collection that emerged as emphasis shifted from the analysisof 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 StudiesA Symposium, Studies and Reports, NewSeries No. 63, Geneva, 1961.

A oolleoton of recent case studies selected to draw attention to different types of familyliving 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," TheAmerican Economic Review, May 1958, pp. 291-299.

Relates changes in criteria for income adequacy to uconomic and social developments inthree broad periods: 1860 to 1900the "Subsistence" or "Break-Even" concept; 1900 to 1935the "Living-Wage" concept; sines 1935the "Social" concept.

69


Technical ReferencesContinued

Number

5. Lama le, Helen H. Study of Consumer Expenditures, Incomes and Savings Methodology of the Surveyof Consumer Expenditures in 1950. Philadelphia, Pa., University of Pennsylvania, Wharton Schoolof Finance and CoMmerce, 1959.

A comprehensive statement of the purposes, procedures, and reliability of results of the 1950survey, 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, and1950.

6. Stigler, George J., "The Early History of Empirical Studies of Consumer Behavior," The Journalof Political Economy, April 1954, pp. 95-113.

Examines the relationship of the empirit al work on consumer behavior to the theoreticalwork on income and demand prior to the twentieth century.

7. United Nations Statistical Office, Handbook of Household Surveys: A Practical Guide for Inquirieson 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 householdsurveys with the aim that their adoption or adaptation would enhance the value of the surveysfor national purposes and facilitate international comparison of results.

8. U.S. Bureau of the Census, Historical Statistics of the United StatesColonial Times to 1967, (1980)Chapter G. The same, Continuation to 19693 and Revisions, 1985.

A compilation of tabular data from major Government and private statistical reports onconsumer income and expenditures in the United, States. Text describing each report is keyedto 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 technicalcomments on consumption statistics and an extensive bibliography.

10. Williams, Faith M. and Zimmerman, Carle C., Studies of Family Living in the United States andOther 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 in52 countries. Includes statements on the history and methodology of consumption studies sincetheir inception.

A continuation of this bibliography, covering 1946 and later years, is published in MissWilliams chapter, "International Comparisons of Patterns of Family Consumption," in ConsumerBehaviorResearch on Consumer Reactions (Lincoln H. Clark, ed.) New York, Harper andBrothers, 1058.

11. Zimmerman, Carle C., Consumption and Standards of Living, New Yak, 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.

KATHRYN R. MURPHYOffice of Prices and Living Conditions

70

67

Chapter 9. Family Budgets

Background

"Standards of living" refer to the goals ofconsumers and workers in their consumption ofgoods and services, use of leisure time, andconditions of work. Standard budgets, also de-scribed as family budgets, measure the totalcosts or amounts of income required to achievethe levels and manner of living implicit in thesegoals.1 Cost estimates are developed by trans-lating the generalized concept of a living stan-dard into a list of commodities and serviceswhich can be priced. Thus, standard budgets arenormative, or benchmark, estimates of livingcosts. They do not represent the ways in whichfamily incomes should be spent, or the waysaverage families actually spend their incomes.

The first standard budgets prepared by theBureau of Labor Statistics were developed forthe specific purpose of evaluating living condi-tions of cotton-mill workers in Fall River, Mass.and in the South in 1908-09. These budgetsdescribed two standards of livinga minimum,including only bare necessities ; and a fair stan-dad, including some allowance for comfort. An-other budget defining a standard of health anddecency was developed in 1919. In the late1930's, BLS cooperated with the Works Prog-ress Administration in pricing two budgets: Amaintenance budget, described as above the

'In 1954, at the request of the United Nations Economic andSocial Council, a committee of experts from six differentcountries (including the United States) recommended that thefollowing distinction be maintained between the terms "level'and "standard" of living: The "level of living" relates to theactual living conditions of a people. The "standard of living"relatcs to the aspirations or expectations of a people, that is,the living conditions which they seek to attain or regain, orwhich they regard as fitting and proper for themselves toenjoy. However, laymen and the general public frequentlyrefer to the high levels of living" actually achieved by theaverage American worker as the "high standard" of Americanliving.

3 Published sources for thcse budgets are described in Tech-nical Reference No. 3, listed on p. 73.

a See Technical Reference No. 18. For description of a com-panion budget for an elderly couple, see Technical ReferenceNo. 19.

4 See Technical Reference Nos. 7 and 10.3 See Technical Reference Nos. 11, 14, 15, and 16.

minimum subsistence level but approaching asatisfactory American standard of living; andan emergency budget, derived by mating themaintenance budget for emergency conditions"with the least harm to the individuals and thesocial group." The information available at thetime concerning the requirements for nutrition-ally adequate diets and healthful housing wasincorporated into the definitions of the livingstandard in each of these budgets. For othercomponents of family living, the "require-ments" were formulated primarily on the basisof the personal judgment of the budget makers?

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 judgmentsof scientists and experts where these were avail-able; for other components, they depended onstatistical analyses of consumer choices.8 Thesame method, with some refinements, was usedin 1959, in the interim revisions of the CityWorker's Family Budget and the Budget for aRetired Couple."' ilese procedures were usedagain, with additional refinements, in the mid-1960's to develop the "intermediate" budgets(initially described as "moderate" budgets) fora four-person family and a retired couple. Sub-sequently, the costs of the intermediate levelbudgets were scaled downward and upward bya variety of techniques to produce a "lower"and a "higher" budget for each family type.5Procedures for the intermediate budgets of the1960's and the scaling techniques are describedin the remainder of this chapter.

Description of the Budgets

All normative estimates of living costs mustbe based on specific family situations. The con-struction of a family budget, therefore, requiresa set of assumptions, i.e., specifications, whichmust be formulated explicitly by the budgetmaker at the outset. These relate to the age,


size, and type of family ; the manner of livingappropriate for the specified family compositionand 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 effecton spending patterns, manner of 'living, andfamily needs. The budgets ;or a younger, four-person family, specifies that the family consistsof an employed husband, age 38, who has a wifenot employed outside the home, and two chil-dren, a girl of 8 and a boy of 13. This familytype represents a middle stage in the life cycle,and it has been widely used as the unit for otherbudgets compiled for earlier periods. The familyin the budgets for a retired couple consists of ahusband and wife, age 65 or over, who areassumed to be self-supporting, in reasonablygood 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 innational policy formation over the last threelicades. Budget quantities and budget com-ponent cost estimates for other family typescannot be derived as fractions or multiples ofthe quantities or cost estimates for food, shel-ter, clothing, transportation, etc. for the four-person family or the retired couple.6

Both types of families were assumed to livein an urban area. Assumptions also were madeconcerning the living arrangements and tenureof the families ; inventories of housefurnish-ings, household equipment, and clothing; meansof transportation; ownership of life insurance ;provisions for- medical care; savings positions,etc. In making these assumptions, the budget

°Extensive analyses of consumption data dating back overmore than a century have provided a variety of measures ofgeneral welfare; e.g., the relative adequacy of diets, the pro-portion of income spent for various categories of goods, or theproportion of income saved. These measures, either singlyor 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 spendingthe same proportion of income on food have attained equallevels of living. Although the scale is useful in estimatingequivalent costs of goods and services, or net income re-quirements after income taxes and occupational expenses, itcannot be applied to individual items or major components ofbudget costs.

?For a discussion of the relativity of living standards, seeTechnical Reference Nos. 5 and 6.

°For a description of the Bureau's surveys of consumer ex-penditures, see chapter 8.

makers were guided by data on the prevalenceof ownership of particular types of assets in theurban metropolitan population, and the avail-ability of goods and services provided by gov-ernments for collective consumption or providedunder collective bargaining agreements betweenemployers 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 levelsof adequacy were obtained by varying the as-sumptions concerning the manner of living andby providing different quantities and qualitiesof the necessary goods and services.

The content of the budgets is based on themanner of living and consumer choices in the1960'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 proportionof retired couples who may own their ownhomes), relies heavily on public transportation,supplemented, where necessary, by the use ofan older car, performs more services for itself,and utilizes free recreation facilities in the com-munity:, Compared with the intermediatebudget, the life style in the higher budget, ismarked by more homeownership, high levels ofnew-car ownership, more household appliancesand equipment, and more paid-for services. Formost items common to all budgets the quanti-ties are greater and the qualities higher in theintermediate than in the lower budget, and inthe higher than in the intermediate budget.'

Data Sources

Budget quantities and pricing specificationswere derived from two sources : (1) Scientificor technical judgments concerning the require-ments for physical health and social well- being;and (2) analytical studies of the choices ofgoods and services made by consumers in suc-cessive income groups, as reported in the Bu-reau's surveys of consumer expenditures,8 todetermine by statistical procedures the incomeclass whose spending pattern would be used asthe "norm" for a specified budget level.

Scientific standards for nutritionally ade-

I-N(1tti

FAMILY

quate diets for individuals in different sex-agegroups have been developed by the Food andNutrition Board of the National Research Coun-cil, and translated by the 'U.S. Department ofAgriculture into food plans at different costlevels. These food plans were used as the basisfor 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 adoptedfor the budgets. These standards relate to sleep-ing space requirements, essential householdequipment (including plumbing facilities), ade-quate utilities and heat, structural condition,and neighborhood location of the dwelling units.Fuel requirements were derived by analyzingactual fuel purchases of families in the specifiedtypes of dwellings in relation to degree-days toprovide an adjustment for differences in cli-mate. Estimates of electricity and utility serv-ices required for the appliances specified for thebudgets were furnished by utility companiesand associations.

The widespread use of insurance to cover thecost of major illness was accepted as a basis fora 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 coveredby insurance were derived from data on utiliza-tion rates provided by the 1963-64 U.S. Na-tional Health Survey and the 1960-61 ConsumerExpenditures Survey. Major medical provisionswere specified for the higher budget.

No generally accepted scientific standardsare available for other components of the budg-ets (clothing, housefurnishings, transportation,personal care, household operation, reading,recreation, tobacco, education, gifts and contri-butions, and miscellaneous expenses). There-fore, for most of these components a techniquewas developed which relied on the choices of con-sumers as the basis for a standard. Purchaseswere examined at successive income levels todetermine the income level at which the rate of

9 For a description of the Consumer Price Index, see chapter10.

10 Technical Reference Nos. 13, 15, and 16 provide additionaldetails on the methods, as well as lists of goods and servicesPriced. Dollar cost estimates are provided in these reportsand also in Technical Reference No, 9.

BUDGETS 71

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 atthis income level became the quantities andqualities specified for the intermediate levelbudget. In general, income classes below andabove the classes used for the intermediate levelwere specified as the source of quantities forthe lower and higher budgets, respectively.

For the transportation component, quantitiesfor the intermediate and higher budgets werebased on the average consumption pattern offamilies of each budget type. For the lowerbudget, average patterns of renter familieswere used. Except for the higher budget wherecosts include a car for all families, automobileownership was specified in inverse relationshipto the availability of mass public transporta-tion. Mileage allowances were adjusted by theuse of automobiles for work.

In determining budget costs, levels of pricespaid for items are as important as the numbersof items bought. Items in the intermediatebudget were priced in the types of stores andprofessional and service establishments custom-arily patronized by urban families. Prices, pric-ing procedures, reporting stores and serviceestablishments, and price calculation methodswere those used by the BLS for the ConsumerPrice Index,9 except that additional quotationswere obtained in some cases to calculate aver-ages and different qualities were priced in othercases to represent the intermediate budgetlevel. For some items in the lower and higherbudgets, special prices were collected directlyfrom stores and establishments. In the main,however, prices for those two levels were esti-mates in a variety of ways. 10


In the methods described, a family budgetis the end result of a multitude of decisions bythe budget maker, based on standards formu-lated by scientists or experts or on analyses ofdata on consumption patterns from a variety ofsources. The budgets are not simply the prod-

\J


ucts of a survey of ways families at particularincome levels actually spend their money. Thejudgment of the budget maker is involved inselecting among the family types and mannersand levels of living to be represented ; in deter-mining the most appropriate sources of data tobe used in deriving budget quantities ; and ininterpreting actual family Consumption in termsof norms or benchmarks. The appropriatenessof the operating assumptions can be evaluatedonly by the budget users in relation to the pur-poses to be served."

Budget estimates may be analyzed in fourways: (1) Costs are compared with income.However, costs for a specific family type shouldbe compared only with average incomes, or in-come distributions, for families of the sametype. This kind of analysis has been restricted,therefore, by the availability of cost estimatesfor only two family types. However, familyequivalence scales may be used to develop esti-mates for comparable benchmark levels forfamilies of other types. (2) Budget costs in oneplace are compared with costs in another, i.e.,the budgets provide a basis for calculating anindex of locality differences in living costs. TheBureau has provided this type of analysis inconjunction with its published reports. (3)Costs are compared over time to measurechanges in living standards. The sporadic char-acter of the Bureau's family budget researchprogram imposes serious limitations on thistype of analysis. Also the judgment factor indeveloping budgets introduces a serious bias forevaluating changes in the levels and living stan-dards of families from decade to decade. (4)Finally, budget estimates of different levels arecompared to provide a measure of the aggre-gate addition to income required to raise con-sumption to particular levels. The developmentof budgets for three different levels facilitatesthis type of analysis.


Family budgets are used in economic researchto appraise the economic condition of the popu-lation and to evaluate the need for, and theeffect 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 incomesufficient to purchase the manner and content ofliving used to define a specified budget level; toestimate aggregate costs of consumer goods asa basis for developing public policies ; or to pre-pare estimates of the number of families livingbelow the specified budget level. Budgets alsoprovide benchmarks for administrative deter-minations, as required by a number of existinglaws or policies of social, welfare, and educa-tional agencies ; e.g., to establish criteria ofeligibility for public assistance, public housing,support services for individuals in job develop-ment programs, subsidized medical or mentalhealth, guidance services, or college scholar-ship aid.

In addition to their primary use as tools inevaluating income adequacy, family budgets areused to measure place-to-place differences in liv-ing costs, as a basis for family counseling, inwage negotiations, and as an aid in consumereducation.

Locality indexes based on the BLS budgetsreflect differences in costs of established resi-dents in a community. Rental costs, for ex-ample, are based on the averages for occupieddwellings and are not a valid measure of thecosts 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 relativecosts of purchasing homes in current markets.The cost of food reflects not only differences inprice levels but also, and more important, differ-ences in regional preference patterns in thechoice of food to meet nutritional standards."The indexes, therefore, are more appropriate asresearch tools in analyses of the relationshipbetween income and costs of established resi-dents in different locations than as measures ofdifferences in costs for families moving fromone location to another.

II For a discussion of the uses of family budgets, see Tech-nical Reference Nos. 2, 4, 8, and 17.

" For a measure of the effect on food costs of price-leveldifferences versus regional differences in the choice of foods,see Technical Reference No. 1.

),

FAMILY BUDGETS


1. 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 setof weights representing urban U.S. food patterns, or different weights for each city reflectingthe food preferences of the region in which the city is located.

2. Brackett, Jean C., "New BLS Budgets Provide Yardsticks for Measuring Family LivingCosts," Month ki Labor Review, April 1969, pp. 3-16.

Provides a nontechnical description of the concepts and procedures used to develop thebudgets for a f(mr-person family at three levels of living, a summary of the spring 1967 costestimates and iocality indexes, and a discussion of the appropriate uses of the budgets.

3. Clorety, J:,6eph A., "Consumption Statistics: A Technical Comment," How American BuyingHabits Change, Chapter X, 1959, pp. 217 -242.

Presents a representative cross-section of budgets compiled in this country during the20th century. Shows average dollar cost figures for the total and for the major components ofeach budget.

4. Hawes, Mary H., "Measuring Retired Couples' Living Costs in Urban Areas," Monthly LaborReview, November 1969, pp. 3-16.

Includes estimates of the costs of budgets for three levels of living for a retired couple inspring 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 ownspending habits.

5. 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 UnitedStates and the goals or standards of living which have been accepted in different historicalperiods and for different purposes.

6. Lamale, Helen H., "Poverty: The Word and the Reality," Monthly Labor Review, July 1966,pp. 822-827.

Discusses the standard budget approach to the evaluation of income adequacy for differentfamily types and in different geographical locations and estimation of the extent of poverty inthe United States.

7. Lamale, Helen H. and Stotz, Margaret S., "The Interim. City Worker's Family Budget,"Monthly Labor Review, August 1960, pp. 78 5-808.

Estimates the cost of a "modest but adequate" standard of living for a husband, wife,and two children (living in rented housing), at autumn 1969 prices, in 20 large cities andtheir suburbs. Includes the detailed list of the goods and services used to define the livingstandard for the 1950's, and describes the way this list was developed and priced.

8. Orshansky, Mollie, "Budget for an Elderly Couple: Interim Revision by the Bureau of LaborStatistics," Social Security Bulletin, December 1960, pp. 26-36.

A summary report on "The BLS Interim Budget for a Retired Couple." (See ReferenceNo. 10.) Includes a discussion of various conceptual problems encountered in developing nonn-ative living costs estimates for a retired couple, and some limitations of this particular bud-get.

9. Ruiz, Elizabeth, "Spring 1970 Cost Estimates for Urban Family Budgets," Monthly LaborReview, 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-personfamily consumption costs to obtain comparable estimates for other family types.

10. 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 66 orover and his wife (living in rented housing), at autumn 1959 prices, in 20 large cities andtheir suburbs. Includes the detailed list of the goods and services used to define the livingstandards for 1950's; and describes how this representative list was developed and priced.

11. U.S. Department of Labor, Bureau of Labor Statistics, "City Worker's Family Budget for aModerate Living Standard, autumn 1966" (Bulletin 1570-1, 1967).

Describes changes in this budget over the last two decades, and gives autumn 1966 costs

75

73



for urban United States and costs and comparative indexes for 39 metropolitan areas, and4 nonmetropolitan regions.

12. U.S. Department of Labor, Bureau of Labor Statisticsrl'Revised Equivalence Scale for Esti-mating Equivalent Incomes or Budget Costs by Family Type" (Bulletin 1570-2, 1962).

Includes scale values for selected family type which can be used to approximate total costsof consumption for the three budget levels. Also includes a summary and discussion of thestatus of research on family equivalence scales.

13. 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. urbanaverage prices and averages for five metropolitan areas for selected items priced for thebudget.

14. 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 costsfor urban United States and costs and comparative indexes for 39 metropolitan areas andfour nonmetropolitan regions.

15. U.S. Department of Labor, Bureau of Labor Statistics, "Three Standards of Living for anUrban 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 theconcepts, procedures, data sources, and estimating methods, and provides lists of goods andservices priced. Includes spring 1967 costs and locality indexes.

16. 'U.S. Department of Labor, Bureau of Labor Statistics, "Three Budgets for a Retired Couplein 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.

17. 'U.S. Department of Labor, Bureau of Labor Statistics, "Report of the Advisory Committeeon 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 themajor uses of standard budgets.

18. U.S. Department of Labor, Bureau of Labor Statistics, "Workers' Budgets in the UnitedStates: 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 1946-47cost estimates for 34 cities. Also an historical survey of family budgets, and summary dataon State budgets for single women workers.

19. "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 techniquesused to compile this budget were parallel to those employed in developing the original BLSCity Worker's Budget. See Reference No. 18. )

ti

JEAN C, BRACKETT

Chapter 10. Consumer Prices

Background

The Consumer Price Index was initiated duringWorld War I when rapid changes in living costs,particularly in shipbuilding centers, made such anindex essential in wage negotiations. To provideappropriate weighting patterns for the index,studies of family expenditures were conducted in92 industrial centers in 1917-19. The Bureau ofLabor Statistics began publication of indexes for32 individual cities in 1919. Regular publicationof 'U.S. city average indexes was not begun until1921, but indexes were estimated back to 1913.1Since that time the weighting factors, the list ofitems included in the market basket, and the citiesin which price data were collected for calculatingthe index have been updated several times.

Because people's buying habits had changedsubstantially by the mid-1930's, a new study wasmade covering expenditures in the years 1934-36which provided the basis for a comprehensivelyrevised indey introduced in 1940 with retroactivecalculations back to 1935.

During World War II, when many commoditieswere scarce and goods were rationed, the indexweights were adjusted temporarily to reflect theseshortages. Again in 1950, the Bureau made in-terim adjustments, based on surveys of consumerexpenditures in seven cities between 1917 and 1949,to reflect the most important effects of immediatepostwar changes in buying patterns.2. This ad-justment, was followed by the first cominehensivepostwar revision of the index, which was com-pleted in January 1953' At that time, not onlywere the weighting factors, list of items, andsources of price data updated, but many improve-ments in pricing and calculation methods were in-troduced. Also, coverage of the index wasextended to small cities so as to represent all urbanwage-earner and clerical-worker families. Themost recent revision was completed in 1964, with

the introduction of new expenditure weights basedon spending patterns in 1960-61, and updatedsamples of cities, goods and services, and retailstores and service establishments.

The manner in which the index has been usedand its acceptance by the public have changedfrom time to time. It has seen many appraisals,criticisms, and investigations. Perhaps the mostfar-reaching study was conducted during WorldWar II by the President's Committee on the. Costof Living! The House Committee on Educationand Labor conducted a detailed examination of theindex in. 1951.' The most recent study was madeby the Price 'Statistics Review Committee, ap-pointed by the National Bureau of EconomicResearch, at the request of the Office of StatisticalStandards of the Bureau of the Budget, to reviewall government price statistics,

As a result of these investigations and theBureau's continuing efforts to improve the index,changes in coverage, collection, and calculationprocedures have been introduced at various dines.Examples of these changes include the additionof medium and small cities to the city sample in1953, the extension of coverage to include singleworkers in 1964, and institution of direct pricingof restaurant meals in 1953.

I Collection of food prices back to 1890 bad been Initiated In1903. During the course of the 1917-19 expenditure survey,retail prices for other articles were collected in 19 cities forDecember of each year back to 1914 and in 18 other cities backto December 1917 only. Retail prices of food and wholesaleprices of other items were used to estimate price change from1914 back to 1918.

See Interim Adjustment of Consumers' Price Index (BLSBulletin 1089, 1951),

s see Consumer Price, in the United States 1958-58 (BLS Bul-letin 1256).

'Report of the President's Committee on the Cost of Living,Office of Economic Stabilisation, Washington, 1945.

Consumers' Price IndexReport of a Special Subcommitteeof the Committee on Education and Labor. House of Representa-tives, 82/1, Subcommittee Report No, 2, Washington, 1951.

Government Price Statistics Hearings before the Subcom-mittee on Economic Statistics, Joint Economic Committee, Con-gress of the United States, 87/1, Part 1, Washington, Janu-ary 24,1961.

7775


Description of the Index

Concept and Scop The Consumer Price Index(CPI) i ical measure of changes in pricesof goods and services bought by urban wageearners and clerical vorkers,a including familiesand single persons. The index often is called the"cost -ofliving index," but its official name is Con-sumer Price Index for Urban Wage Earners andClerical Workers. It measures changes in prices,which are the most important cause of changes inthe cost of living, but it does not indicate howmuch families actually spend to defray their liv-ing expenses. Prior to January 1964, the com-plete name for the index was: Index of Change inPrices of Goods and Services Purchased by CityWage-Earner and Clerical-Worker Families toMaintain 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 PriceIndex the procedure is to measure price changeby repricing at regular time intervals and com-paring aggregate costs of the goods and servicesbought by consumers in a selected base period.The quantities of these goods and services are keptconstant except at. times of weight revisions. Sincenew weights .are introduced without affecting theindex level, any change in aggregate costs is due toprice change. The quantities represent not onlyannual consumption of the goods and servicesactually priced for the index but also consumptionof related items for which prices are not obtained,so that the total cost of the market basket repre-sents total consumer spending for goods andservices.

The index represents price change for every-thing people buy for livingfood, clothing, auto-mobiles, homes, housefurnishings, household sup-plies, fuel, drugs, and recreational goods; fees todoctors, 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. Itdeals with prices actually charged to consumers,including sales and excise taxes, since these arean inherent part of the market price the consumermust pay for goods and services subject to suchtaxes. It also includes real estate taxes on ownedhomes which are part of the price of homeowner-

ship. However, it does not include income or otherpersonal taxes, since they are not associated withprices of specific goods and services although theymay have an indirect impact" Since 1953, it hastreated the purchase of a home in the same wayas the purchase of such durable goods as automo-biles, refrigerators, etc.

In the 1964 rer:sion, the index coverage wasextended to include single consumer units 10 inaddition to families of two or more." The averagesize of families represented in the index is about3.7 persons, and their average family income in1960-61 was about $6,230 after taxes. The averageincome after taxes of single persons representedin the index was about $3,560.

Weighting Structure. The annual consumptionpatterns represented in the index since January1964 were determined in the Survey of ConsumerExpenditures (CES)U in 66 Standard Metropol-itan Statistical Areas (SMSA's) and smaller cities

*See article by Sidney A. Jaffe, "The Statistical Structure ofthe Revised CPI", Monthly Labor Review, August 1984, pp. 918-924.

The definition of wage earners and clerical workers is based onthe occupational classification used by the Bureau of the Censufor the 1980 Census of Population and listed in the AlphabeticalIndex of Occupations and Industries. The group includes crafts.men, foremen, and kindred workers, such as carpenters, book-binders, etc.; operatives and kindred workers, such ae apprenticesin the building trades, deliverymen, furnacemen, smelters, andpourers, etc.; clerical and kindred workers ; service workers, ex-cept private household, such as waitresses, practical nurses, etc.;sales workers; and laborers, except farm and mine. It excludesprofessional, technical, and kindred workers, such as engineersand teachers; farmers and farm managers ; managers, officialsand proprietors, except farm ; private household workers; andfarm laborers and foremen. A consumer unit included in the1980-01 Survey of Consumer Expenditures was classified in theindex group if more than half the combined income of allfamily members was obtalf.ed in a wage-earner and clerical-worker occupation and at least one family member was a full-time earner (i.e., worked 87 weeks or more during the surveyyear). xh

'For a more detailed discussion, see "Taxes and the Con-sumers' Price Index," Monthly Labor Review, January 1958, pp.58-57. (Reprint 2090.)

20A person living alone or in a household with others fromwhom 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 singleconsumer units and not to unmarried persons.

11Prom January through November 1984. the Bureau computeda separate index applying only to families of two or more, forcomparability with the previous series. Since the same itemswere priced for families and singles, the overall movement ofthe separate index was identical with the index including singleworkers and it was discontinued.

32 The Survey of Consumer Expenditures is discussed inChapter 8 of this bulletin. The selection of the expenditure sur-vey and CPI city samples is described in detail in an article byMarvin Wilkerson, "The Revised City Sample for the ConsumerPrice Index," Monthly Labor Review, October 1980, pp. 1075-1083. (Reprint 2352)

CONSUMER PRICES 77

TABLE 1. CITIES, POPULATION WEIGHTS, AND PRICING SCHEDULE FOR. THE REVISED CON61711ER PRICE INDEX

City and site stratum

Pricing schedule

Populationweight

Food

Other items

SamplesSchedules

M 1 2 3

A. Standard metropolitan statistical areas of 1,400,000 or more in 1960:Baltimore, MdBoston, MassChicago-Northwestern Indiana 4Cleveland, Ohio

Detroit, Mich.Los Angeles-Long Beach, CalifNew York-Northeastern New JerseyPhiladelphia, Pa

Pittsburgh, PaSt. Louis, MoSan Francisco-Oakland, CalifWashington, D,C

B, Standard metropolitan statistical areas of 250,000 to 1,399,9991 '1960:aAtlanta, OaBuffalo, N.YCincinnati,Dallas, TexDayton, Ohio

Denver, ColoHartford, ConnHon oluln, HawaiiHouston, TaxIndianapolis, Ind

Kansas City, Mo.-KansMilwaukee, WisMinneapolis-St. Paul, Mimi

Nashville, TennSan Diego, CalifSeattle, WashWichita, Kane

C. Standard metropolitan statistical areas of 50,010 to 249,999 in 1960:Austin, TexBakersfield, CalifBaton Rouge, LaCedar Rapids, IowaChampaign-Urbana, Ill

Durham, N,CGreen Bay, WisLancaster, PaOrlando, FlaPortland, Maine

D. Urban places of 2,500 to 49,999 in 1980:Anchorage, AlaskaCrookston, MinnDevils Lake, N.Dak.Findlay, OhioFlorence, Ala

Kingston, N.YKlamath Falls, OregLogansport, IndMangum OklaMartinsville, Va

McAllen, TexN.J

Niles, MichOrem, UtahSouthbridge, Mass

Union, S.0Vicksburg, Miss

1.4021.9305. M21.325

2.8055.017

12.6772.703

1.6681.4292.3721.255

2.9342. 347.42.7934

0

1.095

1.8382. 348

1:

no850

1.042

2.933.72

8371.6

1.096

1.2501.3231.2501.2841.284

1.2501,2841.8031.2501.803

. 0651.8521.8521.3521.227

1. 1711.3381.352

-1:2%-1.227

1.2271.1711.3611.3391.170

1.2271.226

111, 1B, 2A, 2B

1A,2B

122

1A, 2B1A, 2B

1A, 2B

11A, 2B

2

1,2121

21212

22121

12

1A,2B1A, 2B

1A, 1B, 2A, 2B1A, 2B

1A, 2B1A,2B1A, 2B1A, 2B

1A,2B1A, 2B

1A, 1B, 2A, 2B1A, 1B, 2A, 2B

1A, 2B1A, 2B1A, 2B1A, 2BIA, 2B

1A,2B-1

1A, 2BIA, 2B

2

1A, 2B1A, 2B1A,2B

21A, 2B1A, 2B1A, 2B

122

1A,2B1A, 2B

1A, 2B1

1A, 2132

1,21211

21212

22121

12

x

xxxx

x

X

X

X

xx

X

x

x

x

X

X

X

x

x

X

X

xxx

X

xx

X

x

X

X

X

X

X

X

X

xx

x

xx

x

x

x

x

x

'The 18 largest Standard Metropolitan Statistical Areas as defined forthe 1960 Census of Population were selected on a certainty basis and rem.sent themselves only in the population weight patterns. The other sampleselections carry not only their own population weights but also prorateshares of the population weights of all cities in their region in the samepopulation class. .

I Item samples are identified as samples "1" and "2." Outlet samples areidentified as samples "A" and "B." The determination as to the extent

. of sampling within an area depended on plans for publishing separate areaindexes and on plans for developing estimates of sampling error and itscomponents.

I Foods, fuels, and several other items are priced every month in all cities.Prices o a few items are collected semiannually or annually in all cities.Prices of other goods and services are obtained on tho schedule indicated:

M -Every month.2- January, April, July, and October...February, May, August, and November.

3 -March, June, September, and December.4 Standard Consolidated Areas.

Population weights revised for this group beginning January 1068

79


covering the period 1960-61, except for Anchor-age, Alaska, which was surveyed for 1959. Ex-penditure records were obtained from the 4,343urban families of two or more persons and fromthe 517 single workers included in the survey whowere classified as wage earners or clerical workers.

Sixteen cities in the smallest size class whichwere included in the 1960-61 surrey 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 includedwith those for the 16 small cities priced for theindex, so that the resulting weights are based onthe total sample of N small cities.

In establishing index weights from the detailedexpenditure data, about 400 items were selectedobjectively to compose the "market basket" forcurrent pricing, beginning withethe January 1964"new series" indexes. Not all items are priced inevery SMSA or city. In order to make possibleestimates of sampling error, two subsamples ofitems have been established. These are priced indifferent areas and in different outlet samples, asindicated in table 1. The population weightsshown in the table are used to combine price datafor the 56 individual areas in the CPI sample intoa U.S. city average. They were derived from the1960 Census of Population but adjusted to repre-sent the wage-earner clerical-worker coverage ofthe CPI. For the 18 largest SMSA's, which areincluded in the sample with certainty, the weightsare based on their respective populations only. Forthe remaining cities, which were selected by prob-ability sampling methods to represent all otherurban places, the weights represent not only thespecific city population, but also the populationof all cities in ths same region and size class. Thus,every city in the same region and size class (otherthan the 18 largest) has identical populationweights."

The list of items priced, includes all the mostimportant goods and services and a sample of the

s Six additional B size Standard Metropolitan Statistical Areaswere added to the national index In January 1966. Since theywere selected outside the probability framework, ,they were as-signed only their own population weight which was substractedfrom that of other stratum E cities in the same region.

1' Minor weight revisions are introduced by linking.For a more complete discussion, see The Consumer Price

Index: Pricing and Calculation Procedures, unnumbered paperby Doris P. Bothwell, Burma of Labor Statistics, II.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. Thecontent of this market basket in terri..4 of items,quantities, and qualities is kept essentially un-changed'* in the index calculation between majorrevisions so that any movement of the index fromone month to the next is due solely to changes inprices. A comparison of the total cost of the mar-ket basket from period to period yields the measureof average price change.

In the selection of the item sample for the re-vised CPI, except for the choice of the particularquality or variety of the item to be priced (speci-fication), probability sampling techniques wereused, as described later. The more important itemsare included in the sample with certainty. Theremaining items within a given expenditure classwere selected with probability. The relative im-portance of a certainty item represents consumerspending for that item only. "Probability" itemsrepresent all other items within the expenditureclass. The total weight of these items is dividedequally among the probability items within an ex-penditure class. Table 2 shows relative im-portances in the national index as of December1963. Individual relative importances are notshown in the table for probability items; rathertheir combined- importance is shown as "otherpriced items" in each expenditure class.


Prices are obtained in the 56 area CPI sample bypersonal visit to a representative sample of nearly18,000 stores and service establishments wherewage 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."Rental rates are obtained from about 40,000 ten-ants. Reporters are located both in the cityproper and in suburbs of each urban area. Co-operation is completely voluntary.

Prices are' collected' in each urban location atintervals ranging from once every month to onceevery 3 months, as indicated in table 1, with a fewitems surveyed semiannually or annually. Be-cause food prices change frequently, and because

SO

CONSUMER PRICES 79

TABLE 2. CONSUMER PRICE INDEX NEW SERIES)' RELATIVE IMPORTANCE OF MAJOR GROUPS, SPECIAL GROUPS ANDINDIVIDUAL ITEMS SELECTED WITH CERTAINTY DECEMBER 1963

ComponentsPercent

temsof

ail iDecember 1963

ComponentsPercent ofall items

December 1963

All items

Mein ambersFoodHousingApparel and upkeepTransportationHealth and recreation

Medical care.Personal careReading and recreation..Other goods and services

100.00

22.4333. 2310.6313.8819.453.702.755.915.06

HShelter

RentHotels and motelsHomeownership

Purchase and financingHome purchaseMortgage interest

Taxes and insuranceReal estate taxesProperty Insurance

Maintenance and repairsCommoditiesServices

Fuel and utilities.Fuel oil and coal

33.2320.156.50.38

14.279.118.282.82.13

3

1.72.41

3.9988

2.055.26.73

Semler, Gaouvs Fuel oil .57Coal. .06

All items less shelter. 79.85 Gas and electricity 2.71All items loss food 77.57 Gas 1.30

Electricity 1.41Commodities 65.97 Other utilities 1.82

Nondurables 4719 Telephone 1..38Durables 18..78 Water and sewerage 44

Services 34.03 Household furnishings and operation. 7.82Commoiliti ass food. 43.54 Textile housefurnbhings .61

Nondura foodApparel commodities

Apparel commodities less footwearNondurables less food and apparel

24.769.157.65

15.60

FurnitureBedroom suite.Living room suiteOther priced items

1.44..28.88

Household durables 5.09 Floor coverings .48Housefurnishings 4.72 Rugs, soft surface .34

Services less rent Et 53 Other priced items .14Household services less rent 1347. Appliances 1.36Transportation services 4.88 Refrigerators .28Medical care servicesOther services

4.565.64

Other priced itemsOther bousefurnisbings.Housekeeping suppliesHousekeeping services

1.83.08

1.551.6s

INDIVIDUAL ITEMS Domestic service .26

Food 22.43BabyF age

sitter... .29.23

Food at home. 17.89 Other priced items .77Cereals and bakery products 2.45

Cereals 085 Apparel and upkeep 10.63Bakery products 1.6 Men's and boys' apparel 2.88

White bread 60 Men's apparel 2..3821

Other priced items 1.05 Suits, year roundMeats, poultry, and fhb 5.63 Other priced items 1.85

Meats 4.45 Hoye' apparel .85Beef and veal 2.21 Women's and girls' apparel 4.08

Hamburger .57 'Women's apparel. 3.23Steak 80 Winter coats .23Other priced items 84 Street dresses. .60

PorkPork chops.BaconOther priced items

1.30363064

Hose, nylon.Other priced items

Girls' apparelFootwear

.392.06.83

1.51Other meats Street shoes, men's .26

Poultry 73 Street shoes, women's .26Frying chickens .51 Other priced items .99Other priced items. 22 Other apparel 2.18

Fish .45 Commodities. .71Dairy products 2.80 Services 1.49

Milk, fresh (grocery) 85 Dry cleaning .79Milk, fresh (delivered)ButterOther priced items

e6

1.02

Men's suitWomen's dress.

Other priced items

.44as

.68Fruits and vegetables 3.02

Fresh fruits .76 Transportation 13.88ApplesBananas

.17

.15Private transportation

Autos and related goods12.649.02

Oranges Auto purchase 5.02Other priced items

Fresh vegetables.24at

New cars.Used can

2.662.47

Lettuce .16 Gasoline and motor oil 3.28Potatoes 24 Gasoline 3.05Tomatoes .14 Motor oilOther priced items .40 Auto parts .72

Processed fruits and vegetables 1.32 Automobile services 3.62Other food at home 3.99 Auto repairs and maintenance .96

Eggs 64 Other automobile expenses 2.04Fats and oils 55 Auto insurance 1.42

Margarine .15 Registration tees .37Other priced items .40 Drivers' license .04

Sugar and sweets 64 Parking fees. .18Nonalcoholic beverages 1.01 Auto financing charges' .63

Coffee, can and bag .40 Public transportation 1.24Other priced items at Local transit .78

Prepared and partially prepared food 1.15 Taxicabs .14Food away from home 4.54 Train fares .07

Restaurant meals 3. 75 Airplane fares .20Between meal snacks 79 Intercity bus fares 05


81


TABLE 2. CONSUMER PRICE INDEX (NEW 8E81E8)1 RELATIVE IMPORTANCE OF MAJOR GROUPS, SPECIAL GROUPS ANDINDIVIDUAL ITEMS SELECTED WITH CERTAINTY' DECEMBER 1963-Continued

Components- --

Percent ofall items

December 1063Components

Percent oall items

December 1963

Health and recreation 19.45 Health and recreation-ContinuedMedical care 5.70 Reading and recreation-Continued

Drugs and pharmaceuticals 1.14 Recreation-ContinuedOverthe-counter Items .50 Recreational services 1.58Pr ascriptions .64 Movies (indoor) .38

Professional services 2.59 Bowling fees .35Family doctor, house visit_ .12 Other priced items .84Family doctor, office visit .77 Reading and education. 1.58Optometric examination and eyeglasses .29 Newspapers_ .50Dentists' feesOther priced items

.88

.55College tuitionOtber priced items

.23

.85Hospital services .36 Other goods and services 5.06Health insurance I 1.61 Tobacco products 1.89

Hospital services :66 Cigarettes 1.74Nonbospital services .71 Other priced items .15Overhead .24 Alcoholic beverages 2.64

Personal care 2.75 Deer 1.06Toilet goods 1.52 Whiskey and wine

,.78

Services 1.23 Away from home .80Men's haircut .51 Personal expenses .53Other priced items .72 Funeral services .28

Reading and recreation 5.94 Bank service charges .12Recreation 4.36 Legal services .13

Recreational goods 2.78TV sets .63 Miscellaneous 1 i .38Other priced items 2.15

3 For a description of the new series, see The Coneumer Price Index,( Rerised January 1984). A Short Description.

s The list of items priced includes all the more Important goods and servicesand a sample of the less important ones. In combination, these representall items included in the OPI. Weights for Individual certainty items areshown separately. Some of them, however, are represented by more thanone specification, but the weights for the individual specifications am notshown. The remaining weight of each expenditure class having both car.flinty and probability items was sbared equally by the probability items as

foods are a significant part of total spending, foodpricing is conducted every month in each urbanlocation. Prices of most other goods and servicesare collected every month in the five largest urbanareas and every 3 months in.all other places. Pric-ing of food is done each month on 3 consecutivedays early in the month; rents and items for whichprices are obtained by mail are reported as of the15th of the month; pricing of other items extendsover the entire calendar month. The Bureau usesmail questionnaires to obtain data on streetcar andbus fares, public utility rates, newspaper prices,and prices of certain other items which do notrequire personal visit by Bureau agents. For anumber of items, e.g., home purchase, college tui-tion, used cars, magazines, etc., data Collected byother Government: agencies or private organiza-tions are used.

To insure that. the index reflects only changes inprices and not changes due to quantity or qualitydifferences, the Bureau has prepared detailed spec-ifications which describe the physical character-istics of the items in the market basket. Speciallytrained Bureau representatives examine merchan-dise in tin stores to dc,! Trmine whether the goodsand services for which they record prices conform

82

of December 1963, except In a few cases where weights for duplicated Itemsbare double weights.

a Not actrelly priced; imputed from priced items.4 Cost of health insurance is imputed to price changes for representative

services plus the cost of overhead. For a more complete discussion, seesnick by James C. Daugherty, "Health Insurance in the Revised CPI,"Monthly Labor Reriew, November 1964, pp. 1200 -1300.

Personal financing cbarges other than mortgage interest and auto ammo.ing.

to the specifications. Where the precisely specifieditem is not sold at a particular retail establishment,the Bureau's representative quotes prices andobtains a detailed technical description of the itemnearest to the physical characteristics of the spec-ification, in order to insure that prices will bequoted on the same quality and quantity fromtime to time. At the first pricing in an outlet theagent selects the volume selling item meeting spec-ification, making sure that it is regular merchan-dise in good condition and available in a customaryassortment of colors, patterns, etc.

At the subsequent pricings the agent prices theidentical item if it is still available in a reasonableassortment and selling in substantial volume. Ifit is not, she must substitute another volume sellingitem, meeting specification, if possible. If shecannot, she prices an item deviating from specifi-cation. Prices of substitute items meeting specifi-cation are compared directly. Prices of deviatingitems are introduced by linking or splicing insuch a way that the difference in price betweenthe specification and the deviating item is not re-flected as a price change. If it is possible to obtainan 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 thecurrent month.

When the sample of reporters is changed for anyreason, pricco from the new reporter also are intro-duced by linking.

Sampling

A complicated index such as the CPI must bebased on a whole complex of samples." A sampleof cities or areas is required in which expendituresurveys 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, butnot essential, that the city sampling points be thesame for price collection as for the Consumer Ex-penditure Surveys.

Further, since it is impossible to price all thethousands of items which consumers buy, it isnecessary to select a sample of items for pricing,to represent price movement of all items. Samplesof outlets are needed ai; each sampling point inwhich price quotations are obtained for the selecteditems. Finally, pricing usually is done at aspecific time of the month or quarter so there is,in effect, a sampling of time.

In the 1964 revision, the Bureau of LaborStatistics used probability sampling to a greaterextent than had been done previously, despite thewell-known difficulties involved. At the sametime, the Bureau attempted to include in the CPIsampling design some method for obtaining an

:estimate, even if only a crude one, of samplingerror. Probability sampling is a necessity, ofcourse, if this is to be done in a conventional man-ner. However, even if probability sampling couldbe followed rigorously through all the complicatedCPI structure, the mere computational load wouldbe 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,itemsamples, and replicated outlet samples. In addi-tion to the minimum program, designed to producean 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 theerror, that is, variation in sampling results due tosampling of cities, items, and outlets.

City Sample. A core sample of 50 SMSA's (seefootnote 18) or smaller cities for the index, sup-plemented by 16 additional D size cities for thefamily expenditure surveys was the maximum sizeconsistent with available budget. These addi-tional D size cities were surveyed because expendi-ture patterns are more variable among small citiesthan among large cities. The primary samplingunits (PSU's) are Standard Metropolitan Sta-tistical Areas as they were defined by the Bureauof the Budget prior to the 1960 Census, except thatthe Standard Consolidated Areas for New Yorkand Chicago were used, plus individual urbanplaces outside the SMSA's. Because 1960 Censusdata were not then available, the measure of sizeused in sample selection was the estimated urbanpopulation as of January 1, 1959. The populationweights actually assigned are based on 1960 data.The PSU's were stratified by broad region and bysize into 12 regional-size strata. The 12 largestSMSA's were selected with certainty, that is, theyrepresent themselves in the sample design. SinceAlaska and Hawaii have been added in the revisedCPI, one sample selection has been allocated toeach of these two States. The remaining 36 selec-tions are allocated to the 12 regional-size strata onthe basis of relative population and relative costsof pricing cities of different size. Four size strataare defined as follows :

A. The 12 largest SMSA's on the basis of urbanpopulation, in effect those with population over1,400,000;

B. Other large SMSA's with urban populationgreater than 250,000;

C. SUSA.'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 generallyas "controlled selection" which was described byRoe Goodman and Leslie Kish in the September

3 For a more detailed discussion, see Sampling Aspects of theRevised OPI, unnumbered paper by Marvin Wilkerson, Bureau ofLabor Statistics, B.O. Department of Labor, 1964.

83


1950 issue of the Journal of the American Statis-tical Association (pp. 350-372). This methodaccomplishes a good geographic dispersion ofsampling points across the country.

After the initial 50-area sample was selected,the BLS received funds to prepare city indexes forsix additional large SMSA'sCincinnati, Hous-ton, Kansas City, Milwaukee, Minneapolis-St.Paul, and San Diegoas part of a plan to publishindexes for each SMSA with 1,000,000 total popu-lation in 1960. These areas were added to thenational index in January 1966.

Samples of Consumer Units. The CES sampleswere chosen as subsamples of housing units enu-merated in advance Comprehensive Housing UnitSurveys (CHUS) conducted in each area late inthe year preceding the actual survey date."' TheCHUS also serve as the source of the samples ofrental dwellings for measuring price change inrents, and of owner occupied units for measure-ment of changes in property taxes. They alsoprovide data for weights for home purchase. Theactual size of the CHUS sample in an area isdetermined primarily by the rental sample desiredand by the proportion of renters in the given area,as estimated from Census data. The number ofaddresses enumerated in tho CHUS is usuallymany times larger than it is in the samples forsurveys of consumer expenditures.

Sampling of Items. A. classification system hasbeen developed to provide a logical publicationframework containing the traditional major ex-penditure groups, subgroups, etc., but, in a broadersense, to divide the thousands of goods and servicespurchased by consumers into meaningful andmanageable components of the universe. It pro-vides the framework for the selection of the itemsample and for the derivation of index weights.

Two levels of the classification systedt-are-ofcritiolt1 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 towhich allocations of items are made and withinwhich probability samples of items are selected.

4The term "expenditure clkss" (EC) is given to this

" The selection of the CES sample er.aeussed in chapter 8of this bulletin.

level. The expenditure classes are primarilygroupings of items which serve similar humanneeds. Items are grouped within an EC so thatthey are as homogeneous as possible with respectto their physical characteristics. It is not possibleto confine groupings to items which are similarwith respect to price movements.

Within an expenditure class base period expen-diture weights will be held constant; that is, theEC expenditures serve as a way of defining thelevel of living which is to be held constant untilthe next major revision of the CPI. The Bureauplans to resample items within an EC betweenmajor revisions whenever there is evidence of amajor redistribution of relative expenditures orindications that the previous sample of priceditems does not adequately represent the class. Theconnotation of "item" in the sampling frame isnecessarily fairly broad and the items are not ofequal 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 tocarry 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 1959pilot survey in Cincinnati, a final sampling framecontaining 52 EC's and 812 items was developed.The list of EC's and the number of items in eachare shown in table 3.

The first step in the selection of the item samplefor the revised index was to make a roughly opti-mum allocation of the total number of items to bepriced to each EC. Factors considered were therelative importances of the EC's and a roughmeasure of variability of price movement.

As in past revisions of the CPI, the sampleswere selected on a national basis. Selection ofindependent samples, city-by-city, is not practicalsince it would result in a huge list of items to bepriced in at least one city and an impossible burdenof writing and keeping up with changes inspecifications.

The two replicated samples of items of the re-vised CPI have been selected with "probabilityproportional to size," size being defined as therelative importance of the expenditures for theitem to total 'expenditures for all items. The gen-

84

CONSUMER PRICES

eral procedure was to array items within astratum and by using a random start to make regu-lar selections along the array. Each of the tworeplicated samples this contains "certaintyitems;" that is, items which are certain of inclu-sion because their relative importance is greaterthan the selecting interval. The replicatedsamples also contain some duplicates of itemsselected but not with certainty. Table 4 containsa listing of the items priced in each sample.

For the final selection, relative importances (inthe family expenditure pattern) for the condensedsampling frame (52 EC's and 812 items) wereobtained from expenditure data for nine of thecities surveyed for 1960. Ideally, of course, thedata should have covered all 66 cities, but such datawere not available in time i'or use in selection ofitems. Expenditure data for these nine placeswere weighted together to give preliminary esti-mates of U.S. average expenditures. (Finalindex weights of course are based on complete datafor all cities.)

The selection of one or more specifications or"specified-in-detail" items to represent the itemsselected from the sampling frame has been madein most cases by commodity specialists from expertknowledge of the item. Factors taken into con-sideration are the importance and representative -ness of particular qualities and the feasibility ofdescribing a selected item clearly enough to permitrepetitive price collection. In a few cases wheresufficient data existed, it is possible to make asecond stage prok,bility selection of specifications.

Outlet Sampling. The first big problem encount-ered in attempting probability sampling of outletswas to obtain information about the universe ofretail 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 fairlyspecific information as to types of merchandisecarried would have been desirable.

Comprehensive establishment data were ob-tained from a list of firms which report to th(:Bureau of Old-Age and Survivors Insurance(Social Security Administration, U.S. Depart-ment of Health, Education, and Welfare). Usingsampling ratios furnished by;BLS, master sam-ples of retail and service outlets were selected by

83

BOASI. These were supplemented with listingsfrom other sources.

In the larger SMSA's, a two-stage sampling pro-cedure has been followed. Samples of neighbor-hood and suburban. localities and shoppingcenters have been selected in which pricing outsidethe downtown area is conducted. These wereselected with probability proportional to sales vol-ume, using the best available sales data. The list-ings of sample outlets were limited to those fallingwithin the sampled areas.

The number of food stores priced varies fromless than 10 in the smallest cities to about 80 inNew 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 meansthat for the cities in which both item samples arepriced eight is the maximum sample size even foritems appearing in both item samples. In a few"A" cities, the sample sizes are set at 5 per sampleor a maximum of 19- At the U.S. level, however,the number of quotations is sizeable.

In selecting the sample, allocations of quotationswere made for each item by type of outlet, basedon available sales data, "where bought" surveys,etc. As a specific example, if eight quotationsare required for a particular woman's shoe specifi-cation, the allocation might be three quotations todepartment stores, two to women's specialty shops,two to women's shoe stores, and one to family shoestores. Specific allocations also are made by loca-tion within the SMSA (central business district,neighborhood centers, and suburbs) and, insome cases, to multiunit and independentestablishments.

In addition to the pricing in regular retail andservice outlets, there are a number of special itemswhose nature requires separate samples of specifictypes of "outlets;" for example, samples of phy-sicians and other medical specialists, restaurants,dairies, hotels, property owners, etc. Each ofthese offers its own particular problems.

When the original samples, which were selectedin Washington, were sent to the regional officesa great many practical problems were encountered,and many expedients and compromises with strictprobability procedures were required to completethe initiation of pricing for the revised index.However, even though some deviations from prob-ability sampling were inevitable and had been an-

E5


TABLE 4. LIST OF ITEMS PRICED FOR THE REVISED CONSUMER PRICE INDEX AS OF DECEMBER 1963

Croups, subgroups, expenditure classesPriced Items

Sample A Sample B

Food:Food at home:

Cereals and bakery products:EC-1 Cereals and grain products

EC-2 Bakery products

EC-3MeatsM, poultry, and fish:es:

. 3A Beer and veal

3B Pork

3C Other meats

EC-4 Poultry

EC-li Fish

BC-d Dairy products

EC -7

EC-8

Fruits and vegetables:Fresh fruits

Fresh vegetables

E0-9 Processed fruits and vegetables..

:

EC-11

EC-12

EC -13

EC-14

Othergfoodsp at home:

EFats and oils.

Sugar and sweets

Nonalcoholic beverages

Prepared and partially pre-pared foods.

EC-15 Food away from home. -

Housing:Shelter:

EC-18 Rent.

Homeownership:Home purchase and financing...

Taxes and insurance

EC-17

EC-113

EC-19

EC-20

Maintenance and repairs:Commodities

Services


Corn flakesRico, long and short grainWhite broadWhole wheat breadLayer Cake, plain

Hamburger, pregroundSteaks, round, bone -InSteaks, porterhouse boneinRump roasts, standingChuck roasts, bone-inVeal cutlets, bone-haPork chops, center cutBacon, elMedPork roads, loin halvesPicnics, smoked.Lamb chops, loinSalami sausage, slicedFrankfurters, *MineFrying chickens, ready-to-cookChicken breasts, fresh_Fillets or steaks, fresh or frozenTuna fish. c.bunk styleMilk, fresh, groceryMilk, fresh, deliveredMilk, fresh, skimIce cream, prepackagedButter, salted

Apples, all purposeBananas, yellow varietyOranges, except Temple or KingGrapes, Thompson seedlessGrapefruit, fresh, pink cc whiteOrange juice, freshHead lettucePotatpes, whiteTomatoesAsparagus, greenCarrots, topped, prepackagedCucumbersSpinach, prepackagedPears, Bartlett, can or jarLemonade, concentrate, frozenBeets, sliced, can or jarTomatoes, can or farDried beans, Navy or Great Northern

Eggs, fresh, large, Grade AMargarine, coloredSalad dressing, ItalianSugar, white, granulatedChocolate bar, plain milkCoffee, can or bagCarbonated driMa., fruit-flavoredTea brgs, orange pekoe and pekoe tea

Bean soup, canned, condensedSpaghetti, in tomato sauce, cannedMashed potatoes, instantPotatoes, French fried, frownRestaurant meals:

lamaBreakiset

Between meal snacks:Coffee, cupCarbonated beverages, cupFrankfurter on rollIce cream, dish

Rent of house or apartmentHotel, motel room rates

Home purchaseMortgage interest ratesProperty taxes, residentialProperty insurance rates:

Fire and extended coverageComprehensive homeownership policy

Exterior house paintFurnace air filtersPackaged dry cement mix-

Reef houses.Rath g roofsRep sinks

86

Flour, white, all-purpose.Cracker mealWhite bread.Cookies, cream filled.Cinnamon rolls, frosted.

Hamburger, preground,Steaks, round, bone-in.Steaks, sirloin, bone-in.Rump roasts, standingRib roasts, bone-in.Beef liver, sliced.Pork chops, center cut.Bacon, sliced.Pork roasts, loin halves.Ham, whole.Bologna sausage, sliced.Liverwurst sausage, sliced or whole.Ham, canned, domestic or imported.Frying thickens, 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 Ring.Grapes, Thompson seedless.Strawberries, fresh.Watermelons, whole or sliced.Head lettuce.Potatoes, white.Tomatoes.Cabbage all varieties except red.Celery, Reseal, stalk'Onions, yellow.Peppers, sweet, green.Fruit cocktail. canned.PlneappleGrape fruit juloe 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:

Coffee, cup.Carbonated beverages, cup.Pie, slice.Candy bar.

Rent of house or apartment.Hotel, motel mom 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.

Reed houses.

Ringpain llving and dining rooms.Repair

ri

CONSUMER PRICES 85

TABLE 4. LIST OF ITEMS PRICED FOR THE REVISED CONSUMER PRICE INDEX AS OF DECEMBER 1963Continued

Groups, subgroups, expenditure classes:Priced items

Sample A Sample 13

HousingContinuedEC-21 Fuel and utilities

EC-22

EC-23

Household furnishings and operation:Textile housefurnisbings

Furniture

EC-24 Floor coverings

EC-25 .. Appliances.

EC-26

EC-27

EC-28

Other housefurnishings

Housekeeping supplies

Housekeeping services

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

EC-29 Men's apparel

.1 EC-30 Boys' apparel

Women's and girls' apparel:Women's apparelEC-31

EC-32 Girls' apparel

EC-33 Footwear

Other apparel:EC-34 Commodities

EC-35

t.

Services


Fuel oil and coal:Fuel oil, 02Coal, anthracite or bituminous

Gas and electricity:Gas, 3 bills per cityElectricity, 3 bills per city

Other utilities:Residential telephone servicesResidential water and sewerage services.

Pillows, bed, polyester or acrylic fillingCurtains, tailored, polyester marquisetteDrapery fabric, cotton or rayon/acetateBedroom suites good or inexpensive qualityLiving room suites, good and Inexpensive qnalityLounge cbairs, upholsteredSofas, dual purposeSleep sets, Hollywood bed typeAluminum folding chairsRugs, soft surface:\ Broadloom, wool

\ Broadloom, nylonRugs, hard surfaceRefrigerators or NM -freezers, electricWashing machines electric, automatic.Ranges, free standing gas or electricClothes dryers, electric, antomatioRoom heaters, electric, portableDinnerware, eartbeuwareCarpet sweepers, manually operatedVenetian blinds, white, steel or aluminum elate._Electric drills, hand heldDetergent, liquid, laundryLaundry soap for fine fabricsScouring pads, steel wool.Toilet tissueDomestic service, general houseworkBaby sitter servicePostal services

'Laundry flatwork, finished serviceLicensed day care service, preschool child....Washing machine repairs

Suite, year round weight, 2 qualitiesTopcoats, woolSuits, tropical weightSlacks, wool or wool blendShirts, work, cottonShirts, sport, cotton, short sleevesShirts, sport, cotton, long sleevesT-shirtCoats, all purpose, cotton or cotton blendDungarees, cotton or cotton blend.

Coats, heavyweight, wool or wool blend, 2 quall-:ties.

Carcoats, heavyweight, cottonSkirts, wool or wool blend.Skirts, cotton or cotton blendDresses, street, chiefly manmade fiber, 2 qnalitiesDresses, street, wool or wool blendDresses, street, cottonHousedresses, cottonSlacks, lightweight, cotton and carded cottonSilpe, nylonBrassieres, cottonHose, nylon, full fashioned and seamless, 2 styles..Anklets, cottonHandbags, rayon faille or plasticRaincoats, vinyl plastic or chiefly cottonSkirts, wool or wool blend..Blips, cotton blendHandbags, plastic.

Men's:Shoes, street, oxford, 2 qualities

Women's:Shoes, street, pump, 2 styliz.Shoes, evening, pumpShoes, casual, pumpHouseslippers, scuff

Children's:Sneakers, boys', oxford type

Diapers, cotton ganseYard goods, oottonEarrings, Pearl, simulated ccDry cleaning, men's wits and women's dresses.Shoe repairs, women's heel liftLaundry, men's shirts

Fuel oil, r2.Coal, anthracite or bituminous.

Gas, 3 bills per city.Electricity, 3 bills per city.

Residential telepbone services.Residential water and sewerage services.

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, npholstered.Box springs.Cribs.Rugs, soft surface:

Broadloom, wool.Broadloom, nylon.

Tile, vinyl.Refrigerators or refrigerator-freesers, electric.Washing machines, electric, automatic.Vacuum cleaners, canister type.Air conditioners, demountable.Garbage disposal units.Flatware, stainless steel.Table lamps, with shade.Lawn mowers, power, rotary type.Nails, 8d (penny) common.Detergent, grannies or powder.Air deodorisers, spray type.Paper napkins, embossed.Stationery, envelopes.Domestic service, general housework.Baby sitter service.Postal services.Laundry fistwork. finished service.Reupholstering furniture.Moving expenses.

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

Sport coats, wool or wool blend.Undershorts, cotton.

Coats, heavyweight, wool or wool blend, 2 qnalitiesCoats, 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, oottoa.Blouses, cotton.Bathing suits, one piece.

Girdles, manmade blend.Panties, acetate.Hose, nylon full fashioned and seamless, 2 styles.(Movie, 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 snits and women's dresses.Antomatie laundry service.Tailoring (Martell, WM adjustment.


TABLE 4. LIST OF ITEMS PRICED FOR THE REVISED CONSUMES PRICE INDEX AS OF DECEMBER 1963Continued

Groups, subgroups, expenditure classesPriced items

Sample A Sample B

Transportation:Private:

EC--86 Auto purchase

EC-37 Gasoline and motor oil

E0-88 Auto partsAutomobile services:

EC-32 Auto repairs and maintenance

EC-40 Other automobile expenses.

EC-41 Public transportation

Health and recreation:Medical care:

EC-42 Drugs and prescriptions

EC-48

E0-44

Professional services.

Hospital services and health In-surance:

Hospital servioes

Health insurance

f>

Personal care:EC-46 Toilet goods

EC-46 Personal care services


-New care:

Chevrolet, Impala, 2-door, hardtopChevrolet, Chevello, 2-door hardtop .Ford, Galaxie 603, 2-door hardtopPlymouth, Fury III, 4-door sedanRambler, Classic 660, 4-door sedan

Used cars:: years old, Chevrolet and Ford3 years old. do....4 years old_ do-h years old- do..

Gasoline, regular and premium..Motor oil, premium.Tires, tubeless, retread

Chassis lubrication, complete.Motor Maul'Automatic transmission re rAuto insurance ratea, I bility and physical

damage.Auto financing charges IAuto reestration and inspection leesDriver's license feesPLocal

trarking ansifees,tprivate and municipal

Taxicab faresRailroad fares, coachAirplane fares, chiefly coach..Bus fares, intercity

Overthe-counter items:Multiple vitamin concentratesLiquid tonicsCold tablets or capsules

Prescriptions:Anti-infectives:

Penicillin 0 buffered tabletsSultisomuole tablets.

Sedatives and hynotics:Phenobarbital tablets.

Ataractic:a:Chlordiarsposide-hydrochloride capsules_

Antiepasmodice:Propantheline Bromide tablets

Cardlovascuicre and antihypertensives:Reserpine tablets

AntlarthritiosCough preparations:

Terpin hydrate with codeine, elixir.Family doctor, office visitsFamily doctor, house visitePediatric care, office visitsPsychiatrists, office visitsRoutine laboratory testsExamination, prescriptions and dispensing of

eyeglasses.Filftp, adult, amalgam, one surfaceDentures, full upper

Daily service charges:Semiprivate room_Private room

Hospital services:- Daily service charges, semiprivate room

Daily service charges, private roomOperating room

Nonboepital services:Family doctor, office visitSurgeon's lees, tonsillectomyladenoldectomy

Prescriptions and drugsRetained earnings (overhead)

Toothpaste, standard dentrifleeHand lotions, liquidpew powder, pressedCleansing tissuesMen's haircutsShampoo and wave sets, plainWomen's haircuts

88

Chevrolet, Impala, 2-door hardtop.Ford, Falcon, Future, 4-door eedan.Ford Galaxie 600, 2-door hardtop.Pontiac, Catalina, 4-door sedan.Volkswagen, Deluxe, 2-door hardtop.

2 years old, Chevrolet and Ford.8 years old, Do.4 years old, Do.8 years old, Do.Gasoline, regular and premium.Motor oil, premium.Tires, tubeless, new.

Water pump replacement.Replacing muffler.Front end alignment.Auto Insurance rates, liability and physical damage.

Auto financing charges.'Auto registration and inspection fees.Driver's license Meas.Parking fees, private and municipalLocal transit Wes.Taxicab fares.Railroad fares, coach.Airplane fares, chiefly coach.Bus fares, intercity.

Aspirin compounds.Cough ps.Adhesive

syruWindage'. package.

Tetracycline capsules.

Secoharhital sodium capsules,

Meprobamate tablets.

Phenobarbital and belladonna extract.

Crystalline digitoxin tablets.Chlorotbiazide tablets.Prednisone, tablets.

Family doctor, oillfse Welts.Family doctor, house visits.Obstetrical came.Chiropractors and podiatrists, office visits.Herniorrhaphy, adult.Examination, prescriptions Bud dispensing

eyeglasses.Fillings adult, amalgam, one surface.Extractions, adult.

Daily service charges:Semiprivate room.Private room.

Hospital services:Daily service charges, semiprivate room.Daily service charges. Private room.X-ray diagnostic series, ripper G.I.

Nonhoepital services:Family doctor, office visit.

Surgeon's fees, hernlorrhapby, adult.Obstetrical cases.Prescriptions and drags.

Retained earnings (overhead).

Toilet soap, hand milled.Shaving cream, aerosol.Deodorants, cream or rollon.

ffome permanent refills.Men's haircuts.Shampoo and wave Mts. plain.Permanent waves, 0014.

CONSUMER PRICES

TABLE 4. LIST OP ITEMS PRICED FOR THE REVISED CONSUMER PRICE INDEX AS OF DECEMBER 1963Continued

Groupe, subgroups, expenditure classes

87

Priced items

Sample A Sample B

EC-47

E C-48

EC-4g

EC-50

E C-51

E C-52

Health andrecreatIonContinuedReading and recreation:

Recreation:Recreational goods TV sets, portable and console

Radios, portable and table models, AM band only.TV replacement tubesSports equipment:

Golf ball% liquid centerBasketballs, rubber or vinyl cover

Outboard motors.Tricycles._DollsStuffed animalDog food, canned and boxed_

Recreational services Indoor movie admissions:AdultChildren's

TV repairs, picture tube replacementBowling fees, eveningGolf green foes

Reading and education Newspapers, street sale and deliveryCollege tuition and fees, undergraduateMagazines, single copy and subscriptionCollege textbooks, undergraduate

Tobacco products Cigarettes, nonfilter tip, regular size, packCigarettes, filter tip, king size, carton -Cigars, domestic, regular size

Alcoholic beverages Beer, at home local and national brandsWhiskey, spirit blended and straight bourbon__Wine, dessert and tableBeer, away from home

Financial and nal.::ollaneour; personalexpenses. Funeral services, adult

Bank service charges, checking account

Other goods and services:

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, 85 mm, color.

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.

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.

Funeral services adult.Legal services, short form will.

1 Two of the largest volume tellers among the following typos of fish arePriced within each city, since within any given city, all varieties of fish arenot evadable: Frozen ocean porch and haddock; fresh cod, catfish, kingsalmon, halibut, sole, and haddock.

ticipated, the final samples adhere to the originalbasic 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 type's of outlets, sections of eachSMSA, etc.

Calculation Procedures

The index is a time series. As previously ex-plained, it is a weighted average of price changesfor 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 consumerexpenditures, are kept constant from month tomonth. The index measures changes as they oc-cur. It is not adjusted for seasonal variation."

The Bureau began publication of seasonally ad-justed indexes in 1966, for selected componentswhich show a significant seasonal pattern of pricechange."

Not actually priced; imputed from priced items.Four items are priced only for healthinsurance: Operating room, X-ray,

tonsillectomy and retained earnings; prices for the remaining items are alsoincluded as directly priced professional and hospital services.

Formula. In the absence of major weight revi-sions, and ignoring the problemstof sampling, theindex formula is most simply expressed as:

(1) ifio X100Z(Pigo)(P.0

or by its algebraic equivalent, the dollar weightedaverage of price relatives:

fa I [(Pogo) 10](1a), /i, X100

(Pogo)

This is the customary, oversimplified way ofwriting a price index formula to show that theq's are held constant between major revisions.In actual practice, the basic data for weights arevalues which include allowances for unpriced

la For a discussion of the problems involved in using varyingseasonal weights. see "Use of Varying Seasonal Weights in Price-Index Construction," by Doris P. Rothwell, in the Journal ofthe American Statistical Association, March 1058,' pp. 60-77.

° Factors need to compute seasonally adjusted indexes are-avail.able on request.

69


items, and the current index is computed by achain computation procedure, as shown below:

(2) .r0--z(A-iox (1');qa) xioo2(pogo) (7"-10

where q is a derived composite of the annual quantitiespurchased in a weight base period for a bundleof goods and services to be represented by thespecific item priced

p and p' are the average prices of the specificcommodities or services selected for pricing (thesuperscript indicates that the average prices arenot necessarily derived from identical samples ofoutlets and specifications over long periods)

i -8 is the month preceding a weight revision(most recently, December 1963)

i is the current montha is the period of the most recent Consumer Ex-

penditure Survey (1960-61) from which therevised weights are derived

o is the reference base period of the index (1957-59).

The (Pogo) or (p;_,D) base "weights" for agiven priced item are the average annual expendi-tures in a weight base period represented by thatitem and other similar non-priced items. Al-though constant physical weights are implicit inthe index, in. reality the constant es are notcalculated separately.

In actual practice, the base expenditure for eachitem is projected forward for each pricing periodby the price relative for the priced item!

pc l(7)14.)--=

pc -1

in practice, then, the index formula is as follows:(Dec. 1963 (Change from Dec.

Index) 1963 to month i -1)

(3) L..= (vi-sqO xZ (74-1q0(MD l(71t--.4.)

(Change from monthf -1 to month I)

iga) (24)X `1-1 X100

(7)i-iga)

Illustrative Calcul ation. Average price changesfrom the previous pricing period to the currentmonth are expressed as relatives (or ratios) foreach item, and the price changes for the variousgoodo and services are combined, using weightingfactors based on the importance of the item inconsumer spending and that of other items whichit represents. This composite importance is calledthe cost weight of the market basket item. Thereis a selg separate cost weights for each of the 56

urban locations included in the index. The fol-lowing hypothetical example for pork illustratesthe index procedure:

Sampleitem

Septemberprice

Octoberprice

RatioOctober+

September

Septembercost

weight

Octobercost weight

(Sept. Xratio)

Pork elope.-HamBacon

Total

$0.75.80

1.00

$0.779.5.82

1.02

1.031.0251.02

$15.008.00

10.00

$15.458.20

10.20

83.00 83.85

Identical results could be obtained for pork bymultiplying prices each period by the impliedphysical quantities included in the market basket,as the following illustrates.

Sample itemImpliedquantity(pounds)

Septemberprice

Septembercost

weight

Octoberprice

Octobercost

weight

Pork chopsRamBacon

Total

201010

$0.75.80

1.00

$15.008.00

. 10.00

$0.7725.82

1.02

$15.458.20

10.20

33.00 33.85

The average change in pork prices is computedby comparing the sum of the cost weights in Octo-ber with the comparable sum for September, asfollows:

October cost weight $33.85 X 100 =-402.6September cost weight $33.00

This means that pork prices in October were 102.6percent of (or 2.6 percent higher than) pork pricesin September.

Although the second method may appearsimpler, in reality it is not. Deriving the impliedquantity weights is an extra operation, and theseimplicit quantities change as revised samples arelinked in. Furthermore, the second formulationgreatly complicates the handling of the numeroussubstitutions of reporters and items whichoccur constantly in repetitiVe index work. Con-sequently, the first method is the one actually usedfor the CPI. The second illustration, however,'may assist the user to understand the meaning ofthe index mechanism.

After the cost weights for each of the items havebeen calculated, they are added to area totals forcommodity groups and all items. The U.S. totalsare obtained by combining area totals, with eacharea total weighted according to the proportion of

CONSUMER PRICES 89


the total wage-earner and clerical-worker popula-tion which it represents in the index based on 1960Census figures. Finally, the U.S. totals for thecurrent and previous months are compared to com-pute the average price change.

Reference Base Period. Since 1962 the index hasbeen calculated on the reference base of 1957-59= 100.!0 This means that current prices areexpressed as a percentage of prices for the averageof the 3 years - 1957, -1958, and 1959. An indexof 110 means that prices have increased 10 percentsince the base period; similarly, an index of 90means a 10-percent decrease. The index can beconverted to any desired base period for which theindex is available. This is done by dividing eachindex number to be converted by the index for thedesired base period. Tables of conversion factorsare provided on request for most series, enablingusers to convert indexes for prior periods on otherbases to the current 1957-59 base. Since the 1957-59 base was adopted in 1962, some indexes havebeen continued on the bases of 1947-49=100 and1939=100. These are calculated by the applica-tion of appropriatkpnversion factors to indexescomputed on the 1957-59 base.

Imputation Procedures. Although prices are notobtained in all 56 cities every month (see table 1for pricing cycle), it is necessary to represent all56 cities in each monthly index computation. Be-tween quarterly survey dates, for every item ex-cept new automobiles, the weights are held at thelevel of their last pricing, which in effect meansprices are estimated unchanged. For new auto-mobiles, a price change is imputed to the unpricedcities on the basis of changes in cities surveyedevery month

For food and apparel items which are sold onlyat certain seasons of the year, the indei calcula-tion is made in the off-season as if prices of theseitems changed proportionally with prices of itemsof a similar nature which are available all year.For example, prices for strawberries when not inseason are carried forward on the basis of changesin prices of all other fresh fruits. When the itemreturns to the market the current price is, in effect,compared with the estimated price implicit in theprocedure described.

Average Prices. In the calculation of averagefood prices for publication, the prices used in theindex are given special editing since they are notnecessarily restricted to a single specified qualityand size. Procedures have been devised to calcu-late city and U.S. prices for publication which useindex values and price relatives extensively.These procedures employ benchmark prices fordefined specifications for each of the 56 cities, inwhich quotations not meeting the specified qualityare excluded. Benchmark prices are computed inan ind:wdent operation, pooling prices for alloutletztirither than as an average of average pricesfor the two subsamples. The benchmark pricesare then adjusted month by month by the pricechanges reflected in the index. The first bench-mark calculation was for April 1964, from whichdate prices were estimated back to December 1963_and forward to December 1964. New benchmarkcalculations are planned as of each January. City--prices are combined to U.S. averages by the use ofthe 1960 index population weight -s?'

Average bills for specified quantities of gas andelectricity and average prices of fuel oil, which arepublished for the largest cities, are the same asthose used in the index calculation. Since theseare for identical quantities and qualities from'month to month, no special editing is required.

Item indexes. Indexes for selected items andgroups (c,ommonlycreferred to as item indexes),were published semiannually during 1964 and1965. Quarterly publication was resumed in 1966.Although the published indexes refer to March,June, September, and December, some prices forearlier months must be used in their calculation forcities not priced in these months. In June, forexample, the indexes are based on (a) June pricesin the 5 cities surveyed every month and the 17cities surveyed in June on a quarterly cycle; and(b) April and May prices in the 17 cities surveyedin each of those months on a quarterly cycle, car-ried forward to June as in the composite indexcalculation.

sine index base was 1947-49=100 from 1958 through 1961,1985-89=2100 from 1985 through 1952, and 1918=100 from1918 through 1144.

" For a more detailed discussion, see article by Doris P.Bothwell, "Calculation of Average Itot:Al Food Prices," MonthlyLabor Review, Tannery 1965, pp. 61-te.



The CPI is made available first at a press con-ference, usually held near the end of the monthfollowing that to which the data relate. On thesame 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 pricechanges during the month and several tables ofmajor group and subgroup indexes and percentagechanges 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 textwith some additional tables is published about twoweeks after the date of the press release. In addi-tion each of the Bureau's six regional offices pre-pares and mails a press release for each of the, citiesin its region for which CPI figures are published.These releases are timed to coincide with the na-tional release. Other monthly reports containaverage prices of selected foods and fuels in thelargest metropolitan' areas. A quarterly publi-cation presents the U.S. city average indexes forindividual goods and services.

The CPI for the United States and for selectedareas is published also in the Monthly Labor Re-view in the issue dated two months later than theindex. The annual Statistical Supplement to theMonthly Labor Review contains indexes for indi-vidual goods and services (item indexes) as wellas the rplative importance of the items in the totalindex as of December.

Average prices for foods and fuels are publishedin Estimated Retail Food Prices by Cities andRetail Prices and Indexes of Fuels and Electricity.

Uses of the Index

The most widespread use of the CPI is in wageadjustments and collective bargaining negotia-tions. Although this was the primary reason forits beginning, use of the CPI for this purposedeclined during the post-World War I and de-pression periods. Its use in this way was revivedduring World War II, but escalation by the indexdid not receive widespread acceptance until theprinciple was written into a contract between theUnited Automobile, Aircraft, and Agricultural

Implement Workers of America and the GeneralMotors Corporation in 1948. The number ofworkers covered by such contracts in 1965 wasabout 2 million. However, movements of theindex have an indirect effect on Wages and salariesof many more workers.22

The CPI is used extensively to measure changesin purchasing power of the consumer dollar. Itis the basis for most estimates of changes in realearnings of labor, and for comparison with pro-duotivity measures. Changes in purchasing powerare used for such diverse purposes as adjustingroyalties, 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 isas a guide to broad economic policy. It is one ofthe most widely used measures of inflationarypressures. During wartime periods the index 'andits components have served an important adminis-

. tredve function in connection with determinationof 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-pricerelationships and to judge the advisability of mak-ing monetary or tax adjustments. It is one of thechief statistical tools for conversion of the nationalaccounts to constant dollars.

Limitations of the Index

The CPI is not an exact measure of pricechanges. It is subject to sampling errors whichmay cause it to deviate somewhat from the resultswhich would be obtained if actual records of allretail purchases by wage earners and clericalworkers could be used to compile the index. Theseestimating or sampling errors, are limitations. uponthe precise accuracy of the indei rather than mis-takes in the index calculation. The accuracy couldbe increased by using much larger samples; butthe cost is prohibitive. Furthermore, the index isbelieved to be sufficiently accurate for most of thepractical uses made of it. With the changes insampling techniques introduced in 1964, the Bu-

See article by Francis S. Cunningham, "The Use of PriceIndexes in Escalator Contracts," Monthly Labor L'evlem. August1963, pp. 948-952. Reprint 2424.

92

+IP

CONSUMER PRICES 91

TABLE 5. SUMMARY OF CHARACTERISTICS OF THE CPI BEGINNING 1964

DEPINITION Or TEE INDEXTitle

Formula (Simplified expression).

Base periodDurationDefinition of index expenditure weights

POPULATION COVERAGE OPEXPENDITURE 13171tVar

Place of residenceFamily elmOccupation

Length of employment

Income

CITY COVERAGE

Population weights

Sample of priced cities

Published Indexes

ITEM SAMPLE

Basis of sample selectionBasis for allocation to priced items

Commodity coverage

Number of Items priced

Pricing cycle

LocationNumber of reporters

Number of quotations

Pricing technique

REPORTER COVERAGE

Consumer Price IndexU.S. City Average for Urban Wage Earners and Clerical Workers.

Z(p-,g.) ()If%

1937-69..100. Series also published on 1047-49 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 changesbetween the survey dates and 1963 except for 6 cities added in 1966.

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 individualsliving alone. (More than hal fof tots I famil,,

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-earnsr or clerical-worker occupations.No criterion as to income except the qualification above.

Based on 1960 Population Census; Alaska and Hawaii included. Proportion of population in wage-earnerand clerical-worker group covered by index was based upon BLS expenditure surveys.

50 metropolitan areas and cities selected originally to represent all nrban places in the U.S. Including Alaskaand 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 morelarge metropolitan areas available in the latter part of 1965.

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 in re-

maining cities.

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 875.000

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 seermilary sources.Specification pricing but agent is permitted to price deviations from specification under prescribed con-

ditions.

reau is attempting to measure the sampling errorin the index_ ,23

Another kind of error occurs because people whogive information do not always report accurately.The Bureau makes every effort to keep these errorsto a minimum, obtaining prices wherever possibleby 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. Thefield representatives who collect the price data andthe commodity specialists and clerks who process

Preliminary estimates of sampling error were computedand published in Measurement of Sampling Error. in the Con.sumer Price Indee: First Results, by Marvin Wilkerson, paperpresented at American Statistical Aseociation meetings, Decem-ber 29, 1984. Additional estimates will be made available aswork continues on this project.

.7

them are well trained to watch for unusual devia-tions in prices which might be due to errors inreporting.

The CPI represents the average movement ofprices for urban wage earners and clerical workersas a broad group, but not the change in prices paidby any one family or small group Of families. Theindex is not directly applicable to any other occu-pational group or to non-urban workers. Somefamilies may find their outlays changing becauseof changes in factors other than prices, such asfamily composition. The index measures only thechange in prices and none of the other factorswhich affect family living expenses.

In many instances, changes in quoted prices areaccompanied by changes in the quality of con-sumer goods and services. Also new products are

93


CONSUMER PRICES

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. Technicalspecifications and highly trained personnel are re-lied on to insure comparability of quality of itemscompared from period to period." Nevertheless,some residual effects of quality changes on quotedprices undoubtedly do affect the movement of theCPI either downward or upward from time totime.25

Another important limitation of the index isthat it measures only time-to-time price change ina given area. City indexes do not show intercitydifferences in either prices or living costs. They

show only differences in rates of price change fromone time to another. Other types of measures arerequired to show place-to-place differences in liv-ing costs. The most recent such measure is "TheInterim City Worker's Family Budget" whichshows the estimated dollar costs of a "mulest butadequate" level of living in 20 large cities and theirsuburbs in the fall of 1959, which is described inChapter 9.

is Mover, Ethel D., "The CPI and Problems of QualityChange," Monthly Labor Review, November 1961, pp. 1175-1185.Reprint 2378, and Larsgaard, Olga A,. and Louise J. Mack,"Compact Cars la the Consumer Price Index," Monthly LaborReview, May 1981, pp. 519-523. neprint 2368.

See testimony of Ewan Clague in Hearings before the Sub-committee on Economic Statistics, Joint Economic Committee,Congress of the United States, 87/1, Part II, p. 588, Washing-ton, May 1, 2, 3, 4, and 5, 1961.


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 Iwo major price indexes published by theBureau of Labor Statistics--the Consumer Price Index (CPI) and the Wholesale Price Index(WPI). Examines the basic elements of an escalator clause and procedures for carrying outthe agreement.

2. Daugherty, James C. "Health Insurance in the Revised CPI," Monthly Labor Review, November1964, pp. 1299-1300.

Explains and justifies the- Major change in the treatment of the health insurance componentof medical care as initiated in the recent revision of the Consumer Price Index. Compares theformer method of pricing actual premium rates with the new method of pricing the benefitsreceived for hospital and professional services combined with a measurement for retained earnings.

3. Hoover, Ethel. D. "The CPT. and Problems of Quality Change," Monthly Labor Review, November1961, pp. 1175-1185.

Explains and illustrates problems of quality measurement met in the index calculationprocedures. 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 argumentthat the index is not a true measure of mice change because of not fully eliminating the effectof quality changes.

4. Humes, Helen and SchirOl Bruno. he Rent Component of the Consumers' Price Index.Part IConcept and Measure t," Monthly Labor Review, December 1948, pp. 631-637."The Rent Component of ths., onsumers' Price Index. Part IIMethodology of Measure-ment," Monthly. Labor &vitro, January 1949, pp. 60-68. Combined in Reprint No. 1947.

Part I discusses toe basic concepts underlying the rent index. Part II explains the methodsof obtaining and calculating rental data, the "new unit bias" which existed during World War IIand the problem of compedsating for depreciation of quality caused by aging.

5. Jaffe, Sidney A. "The Statistical Structure of the Revised CPI," Monthly Labor Review, August1964, pp. 916-924.

Describes the concept and formulation, populatiou and expenditure coverage, statisticaltechniques and problems of the revised index. Examines some operational aspects, especiallysample replication. Presents the index formula in general, simplified, and in operational form,

6. . The Consumer Price IndexTechnical Questions and Practical Answers. Paper presentedbefore the American Statistical Association Meeting, Washington, D.C., December 30, 1959(Mimeographed by the U.S. Department of abor, 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


Numbs7. Lamalc, 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 usedto measure changes in the prices of the various items of shelter coat.

8. Lnrsgaard, Olga A. and Mack, Louise J. "Compact Cars in the Consumer Price Index," MonthlyLabor Review, May 1961, pp. 519-523. Reprint No. 2368.

Summatizcs and explains the methodology used to link compact cars into the ConsumerPrice Index in 1961. Discusses the historical treatment of quality changes in standard sizecats.

9. National Bureau of Economic Research. The Price Statistics of the Federal Government: Review,Appraisal, and Recommendations, (Washington, National Bureau of Economic Research, GeneralSeries, Number 73, 1961), 496 pp. Also appears in Government Price Statistics: Hearings,Subcommittee on Economic Statistics of the Joint Economic Committee, 87th Cong., 1st sees.,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 NBERin 1059 of the main price indexes compiled by the Federal Government: The Consumer PriceIndex; 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. Twelvs staff reports appended.

10. Rothwell, Doris P. "Calculation of Average Retail Food Prices," Monthly Labor Review, January1965, pp. 61-66. _

Explains the BLS methods of collecting prices, and computing indexes and avreage pricesfor food items in the index. Emphasizes the unsuitable nature of index data for comparisonof prices between cities. Presents estimated retail prices of food from December 1963 throughNovember 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.

DiseuggEs 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 AmericanStatistical Association, March 1958, pp.

Describes a formula based on vary. qg seasonal weights for month-to-month measurementsof 'price change which does not exhibit the "biases" of chain indexes and whicb satisfies classicalindex theory with respect to year-to-yew. comparisons. Results of experimentation withalternative 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 PriceIndex. Presents details of history, uses, and method of construction of the index,, Recommendscontinued support of the index by the Congress.

14. U.S. Congress, Joint Economic Committee. Government Price Statistics: Hearings, Subcommitteeon Economic Statistics, 87th Cong., 1st seas., 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 findin:1: of an investigation by the Price Statistics Review Committee ofthe National Bureau of Economic Research in 1959-60 of all government price statistics. Alsoincludes 12 staff papers on specific subjects. The detailed technical report includes recommenda-tions for improvement of all indexes and, specifically for the Conbumer Price Index, suggestsextended coverage to include single consumers, probability sampling techniques, establishmentof a research division, and regularly scheduled weight revisions. Part 2 presents testimonybefore the subcommittee of members of the Price Statistics Review Committee, governmentofficials, and other interested parties concerning the committee report and recommendations.

95

93



Numba15. U.S. Congress, Joint Committee on the Economic Report. The Consumers' Price IndexReport

of the Joint Committee on the Economic Report on the Consumers' Price Index of the United StatesBureau 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 January1984): A Short Description, (U.S. Department of Labor, Bureau of Labor Statistics, September1964).

A nontechnical description of the index, its scope and computation. Explains the marketbasket, formula, uses and limitations of the index. Tables show cities included, populationweights, pricing schedules, groups :1 goods and services priced, their relative importance, andthe number of items priced as of December 1963.

17. . Seasonal Factors, Consumer Price Index: Selected Series, June 1953-May 1981, Bulletin 1366(1963), 47 pp.

Provides basic data with which Consumer Price Index old series indexes can be adjustedfor seasonal variation. Users are cautioned that the 1964 revision may have a very differenteffect 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 tablesproviding indexes and seasonal factors for 86 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 andexzise taxes.

19. --. Interim Adjustment of Consumers' Price Index: Correction of New Unit Bias in Rent .Componentof 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 in1952. The failure to reflect the difference between rents for new dwellings when they first enterthe market and comparable dwellings already on the market during and after the Second WorldWar is discussed and the method of.adjustment presented. Tabulation of adjusted indexes from1940-50, 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 band on the 1934-36expenditures survey, with the adjusted index based on the 1947-49 expenditures survey in 7cities and the revised index baited on the 1950 expenditures survey. Explains the 1952 revisionin 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 ConsumerPrice Index for measurement of the change in the cost of living during wartime. _Includesdetailed 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., December1964 (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 firstattempt 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.

CONSUMER PRICES

Technical References--ContinuedNumber

23.. Wilkerson, Marvin. Sampling Aspects of the Revised CPI (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 Indexat 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 samplingframe 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, October1960, pp. 1078-1083. Reprint No. 2352.

Describes the selection procedures used to derive the core sample of 50 cities used in therevised Consumer Price Index computation. Tables show probability patterns for the selectedareas and basic and alternate city samples for the Consumer Price Index and the ConsumerExpenditures Survey.

DORIS P. BOTHWELL AND CARLYLE P. 'STALLINGSOffice of Prices and Living Conditions

95

Chapter 11. Wholesale Prices

Background

The Wholesale Price Index (WPI) is theoldest continuous statistical series published bythe Bureau of Labor Statistics (BLS) and oneof the oldest in the Federal Government. It wasfirst published in 1902, and covered the years1890-1901. The origins of the index are as-sociated with a resolution of the U.S. Senatein 1891, which authorized the Senate Com-mittee on Finance to investigate the effects ofthe tariff laws "upon the imports and exports,the growth, development, production, and pricesof agricultural and manufactured articles athome and abroad. "'

The index published in 1902 on the base1890-99 was an unweighted average of pricerelatives and included from 250 to 261 com-modities. Since that time, many changes havebeen made in the sample of commodities, thebase period, and in the method of calculatingthe index. The first major change was com-pleted at the end of 1914, when a system ofweighting was introduced and the index wasrecalculated back through 1890.2 By 1940, thenumber of commodities had increased to ap-proximately 900, based on about 2,000 individualprice quotations. Then, in 1952, the most ex-tensive revision in the history of the index wascompleted.3 The number of commodities andquotations was doubled, weights were based on1947 Censuses and changes were made in thecalculation method. Some changes in classifica-

I Wholesale Prices, Wages, and Transportation, Senate Re-port No. 1394, "The Aldrich Report," Senate Committee onFinance, Congress of the United States, March 3, 1893, PartI, (52d Cong., 2d session), Government Printing Office (1893);and Course of Wholesale Prices, 1890-1901, Bulletin of theDepartment of Labor, No. 39, March 1902, pp. 205-209.

s See also. Allan D. Searle, "Weight Revisions in the Whole-sale Price Index, 1890-1960," Monthly Labor Review, February1962, pp. 175-182.

s A large number of the newly introduced commodityprices were carried back to 1947. The presently publishedindex contains the new commodities for the period 1947-51and displaces th! older, less comprehensive index on the 1926base published for the same period.

tion were made also, including expansion to thepresent 15 major groups. A major reclassifica-tion was implemented in January 1967, whenthe 8-digit classification structure was initiated.Also at that time, new weights from the 1963industrial censuses were introduced.

By January 1971, the number of commoditieshad increased to more than 2,500, the numberof price quotations had increased to over 8,000,and the index had become increasingly repre-sentative of general primary market pricechanges.


Concepts

Throughout its history, the WPI had been ameasure of price changes for goods sold inprimary markets in the United States. "Whole-sale".as used in the title of the index refers tosales in large quantities, not prices received bywholesalers, jobbers, or distributors.

From its inception, the index has been con-sidered a general purpose index designed tomeasure the general price level in other thanretail markets. From the beginning of the in-dex, however, attention was directed to somespecific needs of ueers, and indexes for indi-vidnal commodities and for major commoditygroups were published. As early as 1903, twospecial group indexes by stage of processingRaw Commodities and Manufactured Com-moditieswere published "to meet the wishesof students of price statistics." In recent years,emphasis has been placed on the developmentof more subdivisions within major groups andspecial combinations of indexes such as bystage of processing and by durability ofproduct.

Most of the quotations reported to the Bu-reau are the selling prices of representativemanufacturers or producers, but some prices

f38


are those quoted on organized exchanges (spotprices) or at central markets. Prices for im-ported commodities are those received byimportersthe first commercial transactioninvolving the commodity in the United States.Since the index is intended to measure "pure"price change, that is, not influenced by changesin quality, quantity, shipping terms, productmix, etc., commodities included in the index aredefined by precise specifications which in-corporate their principal price-determiningcharacteristics.' So far as possible, prices aref.o.b. production point, and refer to sales for'immediate delivery. Prices applicable to long-run contracts and "futures" are usually notincluded.

Universe

The WPI universe consists of all commodi-ties sold in commercial transactions in primarymarkets of the United States, including Alaskaand Hawaii. Commodities produced in theUnited States are included, as well as thoseimported for sale. The universe covers manu-factured and processed goods and the output ofindustries classified as manufacturing, agricul-ture, forestry, fishing, mining, gas and elec-tricity, public utilities, and goods competitivewith those made in the producing sector, suchas waste and scrap materials. All systematicproduction is represented, but individuallypriced items, such as works of art, are excluded.Also excluded are goods transferred betweenestablishments owned by the same company(interplant or intra-company transfers). Goodssold at retail by producer-owned retail estab-lishments also are excluded because theyconceptually belong to a retail (consumers')universe, rather than to primary markettransactions.

Civilian goods normally purchased by theGovernment are in the universe, but militarygoods are not. Government sales of some com-

An example of a commodity specification for steel stripis: "Strip, cold-rolled, carbon steel, coils, No. 4 temper, No.2 finish, No. 3 edge, base chemistry, 6" x .050", in quantitiesof 10,000 to 19,999 lb., mill to user, f.o.b. mill, per 100 lb."

The prices used in the index through 1951 were the simplearithmetic averages of prices for all Tuesdays in the month.From January 1952 through December 1966, Tuesday of theweek containing the 15th was the pricing date.

modities (e.g., electric power) are included ifthey can be considered competitive with freemarket sales.

Prices

To the extent possible, the prices used inconstructing the index are those that apply tothe first significant commercial transaction inthe United States. Transactions for the sameitem at later stages of distribution are notincluded. 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 thanTuesday is used because it is considered morerepresentative.5

The Bureau attempts to base the WPI onactual transaction prices. Companies are re-quested to report prices less all discounts, al:-lowances, rebates, free deals, etc., so that theresulting net price is the actual selling priceof the commodity for the specified basis of quo-tation. The Bureau periodically emphasizes toreporters the need to take into account all dis-counts and allowances. However, list or bookprices are used if transaction prices are un-obtainable.

Prices are generally f.o.b. production or cen-tral marketing point to avoid reflection ofchanges in transportation costs. Deliveredprices are included only when the customarypractice of the industry is to quote on this basisand the Bureau cannot obtain a price at theproduction point. Subsidies to the producer andexcise taxes are excluded since they are notconsidered part of the price, but import dutiesare 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 toprovide a means for bridging over changes indetailed specifications (or descriptions of itemspriced) so that only real price change will bemeasured. 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 helpinsure that only price changes influence theindex. A new price series resulting from aphysical change in an article or a change in itsselling terms is substituted for the earlier seriesby direct comparison or by linking. The objec-tive of the linking procedure is to insure thatthe index will reflect only those changes dueto actual price differences." Each time a changein the item priced occurs, the Bureau appraisesthe significance of the specification change toascertain whether an actual price change oc-curred. If the specification change is minor anddoes not involve price-making factors, the sub-stitution is effected by direct comparison, andany reported price change between the old andthe new specification is reflected in the index.If changes in specification are major, and ifeither no real price change occurred or no in-formation can be obtained concerning the valueof the difference in specification (perhaps in-dicative of a change in quality), the substitu-tion is made by linking and no change isreflected in the index. In this case, any reported

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 modelof the machine was introduced, priced at 12,562.60. The newmodel was considered essentially comparable with the old,except that it had a more powerful motor and larger tires.These were valued at $188.20 more than the value of thoseused on the former model. For linking, the September priceof the new model was estimated at $2,533.70 ($2,347.50 Sep-tember price of former model plus $186.20 increase in valueof motor arid tires). The price comparison between Septem-ber and October was based on the estimated September priceof $2,533.70 and the reported October price of $2,562.60. Thusa 1.1-percent increase was reflected in the October index, butthe price change due to quality improvement (more powerfulmotor and larger tires) was not reflected.

tit

99

difference in price level is not permitted toaffect the index level.

When differences are major, an attempt ismade to obtain data from the reporters on thevalue of the additional (or deleted) featuresand to adjust the price index accordingly. Thisis particularly important in the case of somedurable goods, such as automobiles, which haveperiodic model changes. Also, price increaseswhich result from the addition of features thatformerly sold at extra cost are not reflected inthe index. Conversely, price changes attrib-utable to deletion of equipment which wasformerly standard are not treater ls decreases.

In the event production of a specified com-mMity is discontinued by a reporter, or its im-portaace is reduced, the Bureau collects pricedata for a similar or a replacement item. Pricesare obtained for the new and the discontinuedseries for a 1-month overlap period. The indexis extended by linking, and the difference, ifany, between the new item price and the orig-inal price is taken as a measure of the qualitydifference between the two items.

Linking is also used for the addition to ordeletion 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 thesource of price. Whenever a new commodityis added to an existing commodity group, link-ing of the new item to any one of the existingitems is not pertinent. Instead, the weights ofthe entire group are redistributed to includethe new item and the link is made at the grouplevel instead of at the commodity level. A similarprocedure is used to handle items that drop outof the index.

100


WHOLESALE PRICE INDEXRELATIVE IMPORTANCE, NUMBER OF ITEMS AND PRICE QUOTATIONS FOR MAJOR GROUPS AND SUBGROUPS

Grouping

Relative importancein total 1963 weights

Number of itemsand price quotations,

January 1971December 1970 December 1966 Items 'Price quotations

All commodities01 Farm products01-1 Fresh and dried fruits and vegetables01-2 Grains01-3 Livestock01-4 Live poultry01-5 Plant and animal fibers01-6 Fluid milk01-7 Eggs01-8 Hay, hayseeds, and oilseeds01-9 Other farm products02 Processed foods and feeds02-1 Cereal and bakery products02-2 Meats, poultry, and fish02-3 Dairy products02-4 Processed fruits and vegetables02-5 Sugar and confectionery02-8 Beverages and beverage materials02-7 Fats and oils 102-8 Miscellaneous processed foods02-9 Manufactured animal feeds03 Textile products and apparel03-1 Cotton products03-2 Wool.,products

-03-3 Manmade fiber textile products03-4 Silk' products 103-5 Apparel03-6 Textile housefurnishings03-7 Miscellaneous textile products04 Hides, skins, leather, and related products04-1 Hides and skins04-2 Leather04-3 Footwear04-4 Other leather and related products05 Fuels and related products, and power05-1 Coal05-2 Coke05-3 Gas fuels05-4 Electric power05-6 Crude petroletal and related products105-7 Petroleum products, refined06 Chemicals and allied products06-1 Industrial chemicals06-2 Paint and paint materials106-3 Drugs and pharmaceuticals06-4 Fats and oils, inedible06-5 Agricultural chemicals and chemical products06-8 Plastic resins and materials06-'1 Other chemicals and allied products07 Rubber and plastic products07-1 Rubber and rubber products07-2 Plastic products08 Lumber and wood products08-1 Lumber08-2 Millwork08-3 Plywood08-4 Other wood products09 Pulp, paper, snd allied products09-1 Pulp, paper, and products, excluding building

paper and board09-2 Building paper and board10 Metals and metal products10-1 Iron and steel10-2 Nonferrous metals10-3 Metal containers10-4 Hardware10-5 Plumbing fixtures and brass fittings10-8 Heating equipment10-7 Fabricated structural metal products10-8 Miscellaneous metal products11 Machinery and equipment11-1 Agricultural machinery and equipment11-2 'Construction machinery and equipment11-3 Metalworking machinery and equipment _11-4 General purpose machinery and equipment11-6 Special industry machinery and equipment11-7 Electrical machinery and equipment11-9 Miscellaneous machinery12 Furniture and household durables12-1 Household furniture12-2 Commercial furnitureSee footnotes at end of table.

.101

100.000 100.000 2,50310.038 10.637 10111.176 1.171 3k3

.198 1.357 a2.851 7.088 12

.255 .332 4

.437 .553 192.081 2.001

4.510 .576.694 .780

97

.835 .781

8,017143

48a

124

2984

1317

16.298 16.533 193 6842.017 2.038 22 794.153 4.404 40 t*" 992.300 2.275 16 42

.888 ,856 35 1721.268 1.192 11 232.118 2.047 18 104 i.627 .603 17 24

t1,185 1.183 20 67 1

1.765 1.935 15 54 a

519 li

j

io

6.875 7.149 1831.088 1.152 38.346

20 1.40023 8

17451.3,48

3.631 3.562 67.359 .384 11.132 .139 5

1.239 1264 49.077 .097 12.179 .196 11.694 .667 17.289 .304 9

9339

108

235 7

1292

10012125521

7.322 7.130 67 275.700 .439 7 39.092 .070 6 6.677 . 2

1.767 1.808691

18 173.828 .814 6 22

3.459 3.508 28 33

5.937 6.378 335 5501.814 1.988 97 101.698 .706 30 71.807 .888 107 167.189 .163 7 7

.3.545. .487

43 48 3

66 565

10 381.517 1.522 41 123

2246 2.2.339 45

339 881.475.872 43

2.483 .418 771.259 1

2.21650

.688 .658 15

.399 .416 75.140 .129

4.798 4.877 74

4.655 4.719 66.141 .158 8

13.456 12.799 3754.797 4.547 1083.376 3.205 92

.488 .482 a

.575 .548 44

.181 .177 13

.256 .254 211.771 1.716 412.012 1.890 48

12.508 12.110 627.705 .665 55885 .807

1..545 1.469 1051.770 1.685 781.641 1.506 764.348 4.482 1821.614 1.538 82

3.529 3.584 93.925 .904 21.449 .416 7

335

17722

23672468

3441

266

22244

1,295404275

165 44

90

15566

2,098202169

287319

223 663262

151

WHOLESALE PRICES

WHOLESALE PRICE INDEX-CONTINUEDRELATIVE IMPORTANCE, NUMBER OF ITEMS AND PRICE QUOTATIONS FOR MAJOR GROUPS AND SUBGROUPS

101

Grouping

Relative importancein total 1963 weights

Number of itemsand price quotations,

January 1971December 1970 December 1966 Items Price Quotations

12-3 Floor coverings12-4 Household appliances12-5 Home electronic equipment12-6 Other household durable goodsI:. Nonmetallic mineral products13-1 GlassI13-2 Concrete ingredients13-3 Concrete products13-4 Structural clay prods., excluding refractories13-5 Refractories13-6 Asphalt roofing13-7 Gypsum products13-8 Glass containers13-9 Other nonmetallic minerals14 Transportation equipment14-1 Motor vehicles and equipment14-2 Aircraft14-4 Railroad equipment15 Miscellaneous products15-1 Toys, sporting goods, small arms, etc.15-2 Tobacco products15-3 Notions15-4 Photographic equipment and supplies15-9 Other miscellaneous products

.336

.862

.5.37483

3.175.388.630.922

.115970

.120.090.305.392

7.4416.642

.490

.310

2356.515.863.101.30571.7

.385.89 1.459.529

3.040.364.612.882.168.139.127.106.273.369

72446.932

.312

2.498.513.802.102386

9281018

45444655377

82105

130 19

95

4035

134

73

40514

16968495

189

2647

1521321010

350103481069

120

Subgroup index not published.NOTE: Relative' importance represents the basic value

weight of an item or items multiplied by the relative ofprice change between the weight date and a later date, and

Prices for individual commodities reportedby 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 orderthat a change in the company-reporter sampleitself not affect the measure of percent change,the change is calculated for any 2 months fromidentical-company data. Thus, a new reportaffects the index no earlier than the secondmonth.

Classification

The classification system of the WPI followscommodity lines. Products are grouped by sim-ilarity of end-use or material composition,rather than by industry of origin. The WPIclassification does not match the Standard In-dustrial Classification (SIC), the StandardCommodity Classification, the United NationsStandard International Trade Classification(SITC), or any other standard classification.Historical continuity and the needs of indexusers have been important in developing theclassification. No single classification plan can

the result is expressed as a percentage of the total for allcommodities. The differences between the relative impor-tances as of December 1966, the date of last major weightchange, and that of December 1970 are the result of pricechanges only.

meet all of the requirements for wholesale pricestatistics, but the plan adopted should be flex-ible enough to facilitate regrouping of priceseries to make special grouping indexes. InJanuary 1971, the index was made up of 15major groups, 87 subgroups, 291 productclasses, 554 subproduct classes, and 2,598

To meet the needs of index users, a numberof special group indexes are calculated andpublished each month. Among these are indexesby stage of processing,8 indexes by durabilityof product, and indexes of construction ma-terials, in addition to about 22 other specialgroup 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 appropriateprices and weights are those of the WPI. The

See table for the major groups and subgroups included inthe WPI,

The broad stages of processing are: Crude materials forfurther processing; Intermediate materials, supplies, andcomponents; and Finished goods. Each of these is subdividedfurther.

ICC

102 BLS HANDBOOK

stage of processing indexes, however, regroupeach item priced in the WPI according to theamount of processing, manufacturing, or as-sembling it undergoes before entering themarket. A commodity may appear in severaldifferent categories in this scheme. Thus, 29percent of the fresh vegetables (by value-weight) was assigned to crude foodstuffs andfeedstuffs for further processing and 71 per-cent to consumer foods (as "finished" goods).The value weights are the same as those of theWPI and the allocations among the stages ofprocessing are from an inter-industry trans-action study made for the year 1958 by theOffice of Business Economics.


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 pricechanges are infrequent, the shuttle schedule ismailed quarterly, but monthly prices are re-quested. Generally, the price data used in theindex are obtained directly from the producingcompany, but some trade publications are usedwhen the publication generally is accepted asreliable by the Bureau and the industry. Forfish and most agricultural products, the pricesused are those collected and published by otherGovernment agencies.

Price reporting is initiated, wherever possi-ble, by a personal visit by a Bureau representa-tive to the prospective respondent. Pricing ofadditional products from established reportersoften is started by mail. In any event, a de-tailed report describing all of the price-makingcharacteristics of the commodity is preparedfor each new price series. This commodityprice information sheet (BLS 1810) is shownon pages 107 and 108. The form becomes a partof the permanent record for the series. Afterthe initial collection of prices, monthly infor-mation is collected by mail on a shuttle sched-ule. (BLS 473, shown on pages 109 and 110.)

OF METHODS

Weights

The price data are combined using weightsbased on value of shipments. The major sourcesof the value data are :

Bureau of Census _Census of ManufacturesCensus of Mineral Industries

Bureau of Mines __Various publications, e.g.,Minerals Yearbook

Department ofAgriculture __ ____ Various publications, e.g.,

Agricultural StatisticsBureau of

Fisheries Various publications, e.g.,Fisheries of the United States

In addition, many other sources of data, suchas trade associations, are used. Import dataare obtained from a report of the U.S. Depart-ment of Commerce, United States Imports forConsumption.

Sampling

The monthy index is based on a judgmentsample of commodities, a sample of specifica-tions (descriptions), and a sample of reporters.The sample of commodities is chosen after areview of the data of the industrial censusesand other statistics of value of transactions.Generally, the commodities chosen are those ofthe largest shipment values. Starting withJanuary 1967, expansion of Industry SectorPrice Index sample coverage has been a majorinfluence in selecting new products for theWPI. New items are not added until they havebecome established in the market.° They areadded, normally, in December of any year, andhave their first effect on the index in January.

Samples of specifications and of reportersare selected after consultation with trade as-sociations or other industry representativesand with staff of other government agencies.Individual commodity specifications are se-lected also on the basis of net dollar sales. Thatis, the "volunie seller" of the industry (not of

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, impartsa downward bias to the index. Also, many new products turnout to be of only transitory significance.

1( 3

WHOLESALE PRICES 103

the company) is preferred. The specificationdescribes 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 basedon the company's own most common practice.For some commodities, prices are quoted byproducers and sellers in terms of a single spe-cification taken as standard ; all other pricesare quoted as differentials from the standard.The lattet is true for some farm products suchas wheat and cotton. When no standard com-modity basis exists, the specification to bepriced is selected with the help of industryexperts.

The number of reporters is determined, tosome extent, by the variation of price move-ments among them and the degree of priceleadership, Whenever possible, a minimum ofthree companies is obtained, so that data forspecified commodities" can be published withoutdisclosure of information supplied by individ-ual companies. For commodities with morethan one major production area and a definiteregional pattern, a larger sample is selected.Among these commodities are waste materialsand building materials such as brick, cement,and stone.

A comparatively small list of properly se-lected commodities would produce a reliableindex, if only an All Commodities index weredesired. However, historically interest has beengreat in indexes for groups of commodities andfor individual commodities. To meet theseneeds, the Bureau has increased the sample inorder to provide more detailed indexes as wellas many special-purpose indexes.


Formula and Calculation

In concept, the Wholesale Price Index is cal-culated according to a modified Laspeyresformula:

1. I, [2; Q.P,/2; Q,,P] x. 100, where P. is the price ofa commodity in the comparison period and P1 is itsprice currently. Q. represents the quantity shippedduring the weight-base period.

An alternative formulation more closely ap-proximates the actual computation procedure :

2. I, = (Q,P,,) (Pi/Pd / Q.P,) x

In this form, the index is a weighted averageof price relatives for each item (Pi /Pa). Theexpression (Q,J30) represents the weights invalue form and the "P" and "Q" elements (bothof -which Originally relate to period "a" but areadjusted for price change to period "o") arenot derived separately. Each value weight in-cludes not only the value of items priced butalso the values of impriced items whose pricemovements are assumed to behave similarly.When new weights are introduced, the indexwith new weights is linked to the index con-structed with the earlier weights. The weightadjustment itself, therefore, affects only thelater calculations of average price change.When specifications or samples change, theitem relatives must be computed by linking(multiplying) the relatives for the separateperiods 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 1971.10 Ithad been based 1957-59=100 from January1962 through December 1970. Earlier baseswere 1947-49, 1926, and 1913. New items (ornew index groupings consisting primarily ofnew items) introduced into the index after1967 cannot be calculated on the 1967 base.Such indexes are published with separate basesrelated to the date of introduction.

Weights

The WPI weights represent the total net sell-ing value of commodities produced, processed,

to Conversion of indexes from the 1967=100 base to the1957-59 base may be accomplished by multiplying the1967=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 andPrice Indexes" may be. used for base conversion.


or imported in this country, including Alaskaand Hawaii, and flowing into primary markets.The values are f.o.b. production point and areexclusive of excise taxes. The value of inter-plant transfers, military products, and goodssold at retail directly from producing establish-ments also are excluded. Thus the definition ofthe weights conforms to ths; universe definition.

Each commodity price series is consideredrepresentative of a class of prices and is as-signed its own weight (the shipment value ofthe commodity) plus the weights of other re-lated commodities not directly priced but whoseprices are known or assumed to move sim-ilarly." The assignment of price movementsfor priced commodities to those for whichquotations are not obtained is referred to asimputation. For some commoditiessuch asships and some kinds of custom-made ma-chineryit is not possible to obtain directmeasures of price movement. The weights forsuch items are assigned to other commoditiesor groups of commodities for which prices areavailable. Usually, this assignment is made topriced commodities that have a similar manu-facturing process, on the assumption of similarprice movements. Price movements for attach-ments and parts for certain machinery oftenare imputed to the machine itself.

The Bureau's policy, i$ to revise the WPIweights periodically when data from the in-dustrial censuses become available." Theweights beginning in 1967 are based on the1963 industrial censuses, The next revisionwhich normally would follow the 1967 indus-trial censuses has been postponed because oflack of resources. Indexes for 1947 through1954 are based primarily on the 1947 censuses.In the January 1955 index, adjustments weremade to align the major group weight totalswith 1952-53 average shipment values as re-ported in the Annual Surveys of Manufactures.Weights based on the 1954 census shipmentvalues were introduced in kmualy 1958. From1961 through 1966, weights were based on 1958census values. In January 1967, new weightsfrom the 1963 industrial censuses were in-corporated with the comprehensive reclassifica-tion mentioned previously. Subsequent minorredistributions of the weights have been made

each January to account for additions anddeletions of commodities.

The Bureau publishes the relative impor-tance of each item in the WPI rather than theactual values used as weights. The relativeimportance of an item represents its basicvalue weight used in the index, including impu-tations, multiplied by the relative of pricechange from the weight date to a later date ;the result is expressed as a percent of thetotal for all commodities or for some indexgrouping."

Imputing Missing Prides

Whenever price data are not available fora particular month, it is necessary to estimatethe missing price for use in the calculation ofthe index. For commodities in the farm prod-ucts and processed foods groups, out of themarket seasonally, the price in off-season isimputed from the combined movement of therelated commodities for which prices are avail-able for the two periods being compared. Forother commodities, delinquent prices are heldunchanged 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 producerswho reprice, each month, steel of the samespecifications as used in structures on whichthey had been engaged at the time pricing forthe WPI was initiated. Elevators, normallysold including installation, are reported f.o.b.planti.e., excluding transportation and in-stallation costfor use in the WPI.


The monthly WPI is published first in a pressrelease, usually issued in the first week of the

I% Before the 1952 revision (calculated back to 1947), priceditems in the index were weighted only by their own marketvalues.

In general, the censuses are collected at 5-year intervals.la The use of relative importance data to construct indexes

for groups of products is discussed in the January 1970 issueof "Wholesale Prices and Price Indexes." Relative importancesas of December for all WPI items are published in eachJanuary issue.

WHOLESALE PRICES

month following the reference month. Indexesare shown for all groups and subgroups as wellas for All Commodities, Farm Products andProcessed Foods and Feeds combined, and In-dustrial Commodities. Analytic tables also areincluded which show monthly percent changesfor the preceding 12 months for major group-ings, and selected seasonally adjusted and un-adjusted changes for some stage of processingclassifications. A brief description and analysisof the causes of price movements are included.The monthly detailed report, issued some timeafter the press release, carries all data forwhich wholesale price indexes are published,including item indexes and all special groupindexes. Prices for many individual commodi-ties also are included. Each quarter this reportincludes a more comprehensive analysis thanthat given in the press release. Annual sum:maries appear in the monthly report as theybecome available. In addition, numerous his-torical tabulations at various levels of detailare available on request.

The monthly indexes are published as final.Beginning with data for January 1967, onlymajor corrections are made and published im-mediately at the time the error is discovered.Each year, after calculation of the Decemberindex, all corrections reported during the yearare made, and the indexes for all 12 monthsand the annual average are republished asrevised.

Selected seasonally adjusted indexes or per-cent changes are published in the press releaseand monthly detailed report. About 50 indexeswhich historically show significant and con-sistent seasonal movement are presented eachmonth seasonally adjusted and unadjusted. Theapplicable season adjustment factors are avail-able on request from the Bureau. These factorsare recalculated annually to include morerecent data, and the most recent set of factorsmay differ somewhat from those previously inuse.


The WPI is used by government and privateresearch agencies for many purposes, including

105

market analysis, escalation of long-term pur-chase and sales contracts, and formulation ofmonetary policies. It is used, as well, as anindicator of economic trends.

A 1961 survey of users of the WPI revealedthat 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 buyingprices. The survey revealed that over 10 billiondollars (in terms of unexpired value) in long-term contracts for purchase of material orlease of industrial property are escalated ac-cording to changes in the total index or itscomponents. Government agencies and privateresearch groups also use the component seriesin deflating value data in preparation of thegross national product estimates and in studiesof economic growth._

The index `also is used by buyers and sellersof commoditiespurchasing agents and salesmanagers. In most of these cases, it is not theAll Commodities index, but rather the groupindexes and the individual price series thatare employed. Buyers of commodities are ableto check both the amounts which they pay forgoods and the general movement of their pur-chase prices against the index. The use of theindex for checking absolute price levels islimited substantially, however. The Bureau'smain goal has been to measure the directionand amount of change, and only incidentally tomeasure actual selling prices.

The index, as a measure of general and speci-fic price trends, also is used widely in budgetmaking and review, both in government and inindustry; in planning the cost of plant expan-sion programs; in appraising invey tories; inestablishing replacement costs; etc. Compo-nents of the index also are used in LIFO(Last-In, First-Out) inventory accounting bysome organizations.

Although the WPI often is used to measurechange in purchasing power of the dollar, itshould not be used to measure changes in gen-eral purchasing power, prices at retail, securi-ties prices, etc. Comparisons between the levelof the WPI, the Consumer Price Index, andthe indexes of prices of farm products showrelative change from a base period, but com-

106


parisons of the index levels should not be usedas a measure of the actual margins betweenfarm prices and manufacturing or betweenmanufacturing and retail. Its commodity clas-sification structure should be borne in mindwhen using it to measure price changes forindustries, many of which make divers prod-ucts not classified as their "primary" products."

Again, as in other measures, the WPI hassome limitations even in the field for which itis conceptually designed. Segments of the in-dex are used as deflators of gross nationalproduct data, but gaps in WPI coverage leaveconsiderable areas for which deflators havenot been provided.

The WPI is based on a purposive, judgmentsample. The All Commodities Index can be as-sumed to be more reliable than a component"group index, in general. Also, it can be assumedthat the reliability of the index has increasedover time as the sample has expanded." Asthe economy has produced an increasing pro-

See Chapter 12 on Industry-Sector Indexes.The sample of priced items'doubled in 1952 to about 1,850

iteMs and has increased to about 2,500 since then.

107

portion of fabricated finished goods (whoseprice changes are relatively infrequent), overthe years, movement of the WPI has becomesomewhat smoother. Currently, new productsare added each year. -In earlier decades, therewere also major additions of large numbers ofnew items at one time, in commodity areaspreviously underrepresented. These sudden ex-pansions could hive made it appear that priceshad stabilized suddenly.

To the extent that quality improves (or de-teriorates) over the years, the index errs whenno adjustment is made. However, the Bureaumakes suitable adjustments whenever possible.Assuming quality improvement, the indexwould have an upward bias if direct compari-son were made between unimproved and im-proved articles. If, on the other hand, suchchanges were consistently made by linking, adownward bias would result. Since the Bureauhas not adopted either method exclusively, andin many instances tries to evaluate the changesbrought to its attention, the bias that may existis considered to be small. However, no measureof its magnitude is available.

BLS 1810

Commodity

WHOLESALE PRICES 107

THIS FORM WILL BE HELD IN CONFIDENCE Budget Bureau No. 410602


Washington, D.C. 20212

COMMODITY PRICE INFORMATION SHEET

Expiration date: 12/31/73

Code No.

Firm name

Plant or division Mfr. Other(Specify)

Address(Street) (City end State) (Zip cod.)

Information authotIzed by Title

Information furnished by Title

Mail schedule to Title

Address(Reporter)

(Street) (City and State) (Zip code)

L. COMMODITY DESCRIPTION (include style no.,'model no., lot no., grade, brand, etc)

2. PRICE HISTORY FOR COMMODITY DESCRIBED ABOVE

Price Remarks Data Price Remarks

3. CHECK OR FILL-IN PERTINENT INFORMATION ABOUT PRICES REPORTED ABOVE

A. Class of seller to Class of customer(mfr., importer, ate.)

B. Prices are; Actual transaction prices' ; List prices subject to discountsList prices less discounts ; Other El (specify)

(wholesaler, craw. ore.)

;

C. Unit quoted D. Size of order

E. Shipping terms (f: o. b.; frt. allowed; ect.)

F. Type of package used; Crate In ; Carton ; Bag. ; Other

G. Is refund allowed for returnable container? Yes El No If "Yes", explain

Actual selling prices to class of customer, for Oise of order, shipping terms, and discounts reported.

108

ffU


4. ENTER DISCOUNTS AND ALLOWANCES APPLICABLE TO REPORTED PRICES WHEN COMMODITY DESCRIBEDi

IS SOLD TO THE CLASS OF CUSTOMER SPECIFIED IN 3A. iilA. Trade discount 7. i

B. Quantity discount (based on size of order specified in 3D) 7.

FOR THE FOLLOWING, INDICATE DISCOUNT TERMS AND ESTIMATE THE PERCENT OF SALESAFFECTED

EstimatedIt_ms % of sales

C. Cash discount

D. Seasonal discounts

E. Cumulative volume discount

F. Rebates (monthly)

G. Other discounts, allowances, free deals, etc. (explain fully)

CIRCLE ALL DISCOUNTS OR ALLOWANCES ABOVE WHICHHAVE BEEN DEDUCTED IN ARRIVING AT PRICES REPORTED

5. LIST DUTIES OR EXCISE TAXES APPLICABLE TO REPORTED PRICES.

A. These are included in prices quoted 0 Not included

B. If tax is included, give example of how to calculate price excluding tax.

6. ENTER APPROXIMATE PERCENTAGE OF SALES TO EACH CLASS OF CUSTOMER.

Other mfr.(O.E.M. orassembler)

Distributor Jobber Wholesaler Retailer User Other(specify)

Total

Sales (approx.%) 100%

Remarks:

BLS Representative J Date

WHOLESALE PRICES

U.S. DEPARTMENT OF LABORBUREAU Of LABOR STATISTICS

wAsEtINGTON. 0 C. 20212

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 WholesalePrice Index which is the officially accepted indici.tor of primary market pricemovements. The index is widely tlied by industry. And government.

These voluntary reports, submitted by you and other businessmen, arethe major source of information used in preparing this index. The informationyou provide is strictly confidential and open to inspection only to swornemployees of the Bureau of Labor Statistics.

Please use the enclosed envelope, which requires no postage, forreturning thic. schedule. Your continued cooperation is greatly appreciated.

COMMISSIONER OF LABOR STATISTICS

IMPORTANT INSTRUCTIONS

In the boxes provided on the other side, please be sure to indicate all changes inCOMMODITY DESCRIPTION, BASIS OF QUOTATION, DISCOUNTS, ALLOWANCES, AND TAXES

that may have occurred since your lost report.Your cooperation in keeping all information current

is o great aid in computing o reliable, accurate Wholesale Price Index.

(Remarks)

109


INFORMATION FOR THE WHOLESALE PRICE INDEX

CO mmODITT DESCRIPTION (Noe sod.cote oil thane' )

Cods No

CHANGESGIN dote, nature, and

Intimool voles of choolps

2 SASIS Of QUOTATION Rhos* woks% off demons

URA

Oast of sense end onnorson

Sto of mole

Shimoda. awns

Olson (Smoot? 1

3. DISCOUNTS, ALLOWANCES, AND TAXES lollies,. Il Comma, ell , end loins opploceble toobey. basis of euototIon. This information is noodnd to arrive at the ACTUAL SELLING PRICE. (Nomtodicota s1l chonoto.)

911 NO

OvaMoc obasam

Toed* Ammon, New OR? Mil.M.0111

dostorets Iowa dationstdhow No oosionost once'

COO. 10*(01101

Somonal amass'

On oto scoot

Ottosor chesosi Nos* oat al Memo

Nuns saws f hems mclookds

Deft sod Nolen of chweio

Thete and ooote sf chomp

4. NIKE pootmATIoN Poe Itio commodity described in item I, pious Wee bele. the tortoni Mice fot the dole omitted, en the basis quoted onwant 2.

jNice os of Moo 9 1971

PniC MOO6 linoCi 11fl OP +onceof 4.1r/

.......innICIN

()Ave

.....-.PISMO 011 Or 110C4

..... o (If 4wlApr 13 1971 Oct 12 1971

I

May 11 1971 No 9 1971

June 15 1971 Dec 14 1971

July 13 1971 Jan 11. 1972

Aug 10 1971 Feb 15 1972

Sept 14 1971 Mo.1.1.ree....maree

14 1972....

L

PERMANENT OPTIC/ RECORD

KINDLY TIMMTHIS NORM PROMPTLY

WHOLESALE PRICES


Number1. 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, datarequirements, 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 StatisticsReview Committee of NBER, covering uses, concepts, collection, and publication, sampling,and other aspects of the Consumer Price Index, Wholesale Price Index, Index of PricesPaid by Farmers, and other price measures.

5. Searle, Allan D., "Weight Revisions in the Wholesale Price Index, 1890-1960," Monthly Laborreview, February 1962, pp. 176 -182.

History of weight changes and weighting concepts, from inception of the Wholesalele Index.

ILLS Congress, Joint Economic Committee. Government Price Statistics: Hearings: Subcom-:atee on Ec-momic Statistics, 87th Congress, 1st sess., Part 1, Jan. 24, 1961, 526 pp.; Part2, May 1 -5. 1961, 265 pp.

Part I presents the report, Price Statistics of the Federal Government, prepared by theNBER (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 madeby 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. 206 -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,

1954-56 (Bulletin 1214, 1957).Method of Calculating Special Group Indexes, (pp. 12-13), Calculating Relative Im-

portance Data (p. 14), Description of Indexes by Stage of Processing (Economic SectorIndexes) (pp. 15-22) ; A Possible Ell'ect on Weight Revisions (p. 7).

10. U.S. Department of Labor, Seasonal Adjustment Factors; Wholesale Price Index: SelectedSeries 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. 10-12, 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).

112

111



NumberJanuary 1961 Revision of the Weighting Structure, pp. 14-16.

14. 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 inJanuary 1967.

15. U.S. Department of Labor, Wholesale Prices and Price Indexes, January 1970.Describes derivation and use of relative importances (weights) and lists all WPI weights

for December 1969.16. U.S. Department of Labor, Wholesale Prices and Price Indexes, January 1971.

Introduces the new standard reference base, 1967=100, and describes conversion fromthe former base.

17. 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, andon transportation made in response to a Senate resolution of March 3, 1891.

18. Searle, Allan D., "Toward Comprehensive Measurement of Prices," -Monthly 'Lab.Or-Revisw,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.

19. Clorety, Joseph A., Jr., "Measuring Changes in Industrial Prices," Monthly Labor Review,November 1970, pp. 30-36.

Relates Stigler-Kindahl study (NBER, 1970) to BLS program.

113

THOMAS R. TIBBETTS

Chapter 12. Industry-Sector Indexes

Background

During recent years a growing need for com-prehensive measures of industrial prices, inaddition to the market oriented prices of theWholesale Price Index, has become increasinglyapparent. As a result, the Bureau initiated aprogram of industry-sector price indexes basedupon data collected for the WPI.

An industry or sector price index is es-sentially a composite index made up of priceseries that match the economic activity of adefined industry or economic sector. TheWholesale Price Index, on the other hand, iscompiled according to commodity rather thanindustry groupings. The number of sectors orindustries for which these series are compileddepends largely upon resources available foradditional pricing.

A set of industry-sector price indexes cover-ing the years 1947 through 1953 was preparedin the early 1950's as part of the Bureau'sproject on interindustry economics, These in-dexes, generally a regrouping of the WholesalePrice Indexes into the interindustry (input-output) classification structure, were designedfor revaluing bills of goods and industry out-puts. In 1959, another set of such indexes wascompiled for the Bureau of the Census in con-nection with that agency's construction of the1958 production index benchmark.' This sec-ond group of price indexes was used in deflat-

!None of these earlier indexes wa:-, published, but theywere made available to other government research agencies.See Chapter 31, section on Economic Growth Studies, forbackground on interindustry studies.

*The classification of establishments into industries, in thisprogram, follows the guidelines established by the Office ofManagement and Budget in its Standard Industrial Classifica-tion (SIC) system, as revised in 1957. Under this classificationsystem, related products or services are grouped together andgiven an industry code number (consisting of 4 digits). Everyestablishment is assigned to the industry in which its mostimportant products or services, in terms of values, are classi-fied. Many industries contain establishments which producesignificant quantities of goods and services that are classifiedin other industries. These goods usually are referred to as"secondary products." See appendix B.

3Government 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 Economicand Social Council; Problems and Methods in the Gatheringof Representative and Comparable Wholesale Price Series,E/CN. 3/264, 15 March 1960, Chapter II.

ing values of shipments in those census productclasses where physical production data werelacking or unsatisfactory. Again, these wereessentially indexes of commodity prices, classi-fied as primary to a given industry.2

The need for the Bureau to develop Industry-Sector Price Indexes became increasingly ap-parent in 1960 and 1961, when the PriceStatistics Review Committee of the NationalBureau of Economic Research recommended tothe Bureau of the Budget that the basic ob-jectives of an industrial price program shouldbe comprehensiveness, maximum detail in re-porting, and groupings most useful in economicanalysis,3 The committee stated ". . . It seemsdesirable that the subclassification should aimat fitting into the Standard Industrial Classi-fication."

In 1962, the Bureau of Labor Statistics inti-ated the development of industry and sectorprice indexes. Because of its scope, the pro-gram was viewed as a long-run program to beaccomplished in several stages. The first stagewas devoted to the study of conceptual and dataproblems with only a gradual expansion ofcommodity pricing.

The first indexes to be developed were out-put price indexes utilizing gross shipmentsweights. Priority was given to indexes for themanufacturing and mining divisions of theStandard Industrial Classification. The nextstage will be the development of output priceindexes for the trade and transportation andother nonmanufacturing sectors. Input priceindexes, i.e., indexes representing the price ofindustrial purchases, will come later. Eventu-ally the work will expand sufficiently to permitthe development of a general price index ofthe entire economy.

Description of the Survey

Concepts

An industry or sector price index is a com-posite index derived from several series ofprices that closely match the economic activityof a specified industry or industry sector. These


indexes may be either output or input priceindexes based upon either the products andser 'ices sold or the products and services pur-chased by an industry. An output price indexfor a given industry represents price indexesfor a sample of the prodvds produced by thatindustry, averaged together according to therelative importance of production of each sam-ple product to the industry. An input priceindex for an industry consists of an aggre-gation of price indexes for a sample of thecommodities and services purchased by theindustry, averaged together according to therelative magnitude of the purchases.

The Bureau's work has been directed firsttoward two sets of output indexes. One set isweighted by gross shipments of products"made in the industry" to be used for deflatingindustry shipments. The second set will beoutput price indexes of shipments classified onthe industrial basis but weighted by shipmentsof the product produced anywhere in the eco-nomy. A principal use of the second set is forinput-output analysis.'

Universe

Ultimately, the scope of the universe will bedefined in terms of the Office of Managementand Budget's Standard Industrial Classification(SIC) system as revised in 1967, which coversall domestic economic activity. This systemgroups together related products or servicesand assigns them industry and sector codes.Currently, however, the scope of pricing iseffectively restricted to the commodities' cover-age of the WPI because of the use of WPI pricedata.

If price indexes are to parallel industry out-put data, the indexes should cover the totaloutput of each industry including the value of

' See Chapter 31. section on Economic Growth Studies, fora discussion of the nature and uses of input-output data. Seealso W. Duane Evans and Marvin Hoftenberg, "The Inter-Industry Relations Study for 1947." The Review of Economicsand 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.

a The SIC provides no product codes."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 allclasses of customers, and the value of indus-trial services. They should include the value ofsales for export but exclude excise taxes andcosts of transporting finished goods to pur-chasers. This is consistent with the "totalactivity" coverage of statistical series on em-ployment and production.

Input price indexes of materials consumedin production should cover total materials in-puts of the industry, This would include im-ports for consumption, and also transportationand delivery costs.

Prices and Base PeriodThe prices used in the current Industry-

Sector Indexes are in general those used in theWholesale Price Index. In the Wholesale PriceIndex, primary market prices, f.o.b. productionpoint, are used. For the Industry-Sector In-dexes, pricing eventually should be extended toall classes of customers, including retail, foruse in the output indexes. Buyers' prices, in-cluding shipping costs, should be used for inputindexes and should represent the particularmix of products purchased by the buying in-dustry.

As of January 1971, the reference baseperiod for Federal Government indexes is1967=100.

Classification

The Standard Industrial Classification (SIC)of the Office of Management and Budget pro.vides the framework for the Industry-SectorIndex classification scheme. Within this frame-work, individual products are given a 7-digitcode by the Bureau of the Census.° The productindexes are aggregated to 5-digit productclasses and 4-digit industries. Industry indexescan be aggregated to 3- and 2-digit levels aswell. Four-digit industry indexes also can beaggregated to fit the sectoring plan of the latestInput-Output model .°

Sampling and Estimating Procedures

Sampling

Currently the Industry-Sector programlargely depends on price data already available

t°.

INDUSTRY-SECTOR INDEXES 115

in the Wholesale Price Index. However, ex-tension of industrial pricing is being gearedto the new program. A sampling plan has beenprepared which outlines industrial sectors thatshould have priority as new pricing is under-taken. As the result of this analysis, pricingmay be cut back in some sectors to permit ex-tension of pricing into inadequately coveredsectors.

Price data used in computing an industry-output price index should be representative ofthe output of the industry, that is, of the valuesof products made in plants classified in theindustry, but should exclude prices of productsprimary to the industry but made in plantsclassified in other industries. Even though theproducts are the same, the prices received bythe primary and secondary industries may bedifferent, sometimes because these industriessell to different types of users. As a roughguide to the adequacy of sampling, the im-mediate objective is to represent at least 56'percent by value of the commodities includedin each 5-digit Census product class' This per-centage will be slightly less at higher levelsof aggregation. This criterion will be adjustedto levels indicated by differences in variabilityof 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 outputindexes are 1963 value of shipments obtainedfrom the Census of Manufactures, the Censusof Mineral Industries, and data of the U.S.Department of Agriculture. Indexes for 1957through 1966 are weighted by 1958 values.Values include interplant transfer values,values for goods produced and consumed in thesame establishment, and the value of goodssold for export. Values of imported commodi-ties are not included, The difference in thescope of the weights, as compared with theWPI, stems from the objective in this systemto match price data with the scope of domesticindustry production.

Each priced product actually represents aclass of commodities and is assigned its own

weight plus the weights of other products notdirectly priced in the index but whose pricesare known or assumed to move similarly.Values for unpriced products which cannot beassigned to a specific priced commodity areimputed to the average movement of the prod-uct classes in which they fall'

For use in deflating industry shipments, the4-digit (SIC) Industry Indexes are derivedfrom 5-digit product class indexes weightedtogether by their shipments value for the par-ticular industry, i.e., the "made-in-the-indus-try" value.

Formula and Calculation

A modification of the Laspeyres fixed-weightformula is used. The underlying formula is :.14=E(Ps/P..Q.)/EP.Q. where Ps/P. is an indi-vidual product price index. P.Q., the valueweight (base year price times weight yearquantity), currently is the 1963 value of ship-ments adjusted for price change from the year1963 to 1967. In succeeding years, new weightswill be introduced whenever weights are re-vised for the comprehensive Wholesale PriceIndex.'

In actual practice the calculation may besomewhat more involved than indicated by thesimple formula above. For example, indexesused for deflating industry shipments valuesshould employ product weights based uponvalues of commodities made within the sameindustry. Since the Census data for such valuesare available only at the 5-digit (product-class) level, it is necesary first to constructproduct-class indexes based upon total output(wherever made) weights. Then, using 5 -digitmade-in-the-industry weights, the product-class indexes are combined to the 4-digit indus-try level.

This procedure is the same as that employed in the WPI.However, as the product classes are defined differently, anunpriced commodity may have a different price movementimputed to it in the Industry Index program than it has inthe WPI.

Much of the value used in weighting the WPI is derivedfrom the periodic Census of Manufactures and Mineral In-dustries which are currently collected on 5-year cycles. SeeChapter 11, Wholesale Prices.

.73



The published indexes for selected 5-digitproduct classes and 4-digit industries are an-nual averages for the period 1957 through1964, and cover 44 manufacturing and eightmineral industries. Monthly indexes are avail-able beginning January 1965 for the samelimited number of industries and products.Additional indexes are published as they be-come available. By January 1971, indexes werepublished for 102 four-digit industries andabout 350 five-digit product classes.


Price statistics organized along industriallines have particular relevance to studies ofeconomic growth, productivity, and other typesof industrial and economic analysis where theemphasis is on industrial structure as distinctfrom market or commodity-use classifications.

Whether an index meets a given specific needdepends largely upon its commodity coverageand its weighting structure. An important useof an output index weighted by gross shipmentvalues is to deflate value series in order to arriveat measures of output in constant dollars. Mostmeasures of output and productivity rely pri-marily upon physical quantity data for thevarious products of an industry, but in caseswhere quantity data are not available, deflatedvalues can be used if suitable price indexes areavailable, for use as deflators. Deflated valuedata also may serve as a check on productionindexes prepared from quantity data and unit-value weights. There are many sectors of theeconomy for which the analysis of industrialoutput is severely limited because appropriateprice indexes are not available.

Essentially the process of deflation providesa means of obtaining an estimate of quantitychange from available data on total dollar valueand a price index. If the dollar values them-seves are divided by the price index, the result-ing dollar values express the sales value interms of purchasing power of the dollar as ofthe base period of the index. Or an index ofdollar volume can be divided by the price indexto obtain a production index.°

The output indexes also may be used forcomparing movements of prices with other in-dustry-based statistical measures such as em-ployment, earnings, productivity, etc. Priceindexes consistent with total shipments weightswill be useful for deflating industry inputs. Forexample, the appropriate index for deflatingthe value of aluminum purchased by an indus-try would be the index whose components rep-resent shipments of aluminum to buyers in thisindustry rather than the aggregate output ofthe primary aluminum industry.

Input price indexes will be especially usefulto research departments in private industry aswell as to public agencies in making cost stud-ies. They should be consistent in coverage withBLS series on average hourly earnings, anotherimportant element of cost. For contract escala-tion, they will give index users a wider choiceof indexes. Input price indexes, however, arenot available yet.

There are a number of uses which combineoutput and input indexes. For example, grossoutput price indexes and materials input priceindexes can be used to yield a measure of valueadded in constant dollars.

The prices used in contructing the currentlypublished indexes are those regularly collectedon a monthly basis and used in the calculationof the comprehensive Wholesale Price Index.These prices generally are at the primarymarket level but a few are at other levels. Itmust be assumed that these price movementsare similar to the market level of sales repre-sented by the Census data. To include inter-plant transfer values and values of goodsproduced and consumed in the same industry, itis necessary to assume that price movementsof goods in commercial markets represent theprice changes of goods not sold in commercialmarkets.

Until additional pricing can be done, thesenew indexes will be limited by the coveragecommodity and class of customerof the com-prehensive Wholesale Price Index.

It can be shown that division of the value index by theLaspeyres (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 aquantity index of the Laspeyres type. See Chapter 29, Outputper Manhour Measures: Industries.

111

INDUSTRY-SECTOR INDEXES


Number1. 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 of 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 ReviewCommittee of NBER, covering uses, concepts, collection, and publication, sampling, andother aspects of the Consumer Price Inden, Wholesale Price Index, Index of Prices Paid byFarmers, and other Price measures.

6. U.S. Congress, Joint Economic Committee. Government Price Statistics, Hearings before theSubcommittee on Economic Statistics, Part I and Part II, Washington, D.C., 1961.

Part I represents the report, The Price Statistics of the Federal Government, preparedby 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

118

117

Chapter 13. Spot Market Prices

Background

As early as January 1934, at the request ofthe U.S. Department of the Treasury, the Bu-reau of Labor Statistics began the computationof a daily commodity price index, using quota-tions for sensitive comodities. It was releasedfirst to the general public in January 1940. In1952, in connection with the revision of all itsmajor price index series, the Bureau issued anew Daily Index of Spot Market Prices. Thenew index was not a continuation of the oldseries, but was based on a new sample of 22commodities and was calculated on a 1947 49base; in contrast, the old index was based on28 con.modities and was calculated with August1939 as base.

In January 1962, the 22-commodity index wasrecalculated on a 1957-59=100 base to cor-respond to the base period adopted for otherFederal Government general purpose indexes.In January 1971, the index was rebased againin 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.


The Spot Market Price Index is a measureof price movements of 22 sensitive basic com-modities whose markets are presumed to beamong the first to be influenced by changes ineconomic conditions. As such, it serves as oneearly indicator of impending changes in busi-ness activity.

The commodities used are in most caseseither raw materials or products close to theinitial production stage which, as a result ofdaily trading in fairly large volume of stan-dardized qualities, are particularly sensitive tofactors affecting current and future economicforces and conditions. Highly fabricated com-modities are not included for two reasons : (1)they embody relatively large fixed costs whichfact causes them to react less quickly to

changes in market conditions ; and (2) theyare less important as price determinants thanthe more basic commodities which are usedthroughout the producing economy.

A spot price is a price at which a commodityis selling for immediate delivery. In the absenceof a spot price, a bid or an asked price may beused. Some of the prices used are nominalprices in that they are not actual transactionprices. Often they are exchange pricesa pricefor a completely' standard commodity whicheliminates the effect of minor quality changeson actual transaction prices.' Trade publica-tions may use this type of price for commoditiessuch as cocoa beans, coffee, and wool tops. Theprice for print cloth is an average of spot priceand price for most distant forward contractbecause it was determined that a large part ofthe sales of print cloth are made on a contractbasis.

The 22 commodities are combined into an"All Commodities" grouping, with two majorsubdivisions : Raw Industrials, and Foodstuffs.Raw Industrials include burial), copper scrap,cotton, hides, lead scrap, print cloth, rosin,rubber, steel scrap, tallow, tin, wool tops, andzinc. Foodstuffs include butter, cocoa beans,corn, cottonseed oil, hogs, lard; steers, sugar,and wheat.

The items upon which the index is based areclassified further into four smaller groups:Metals, Textiles and Fibers, Livestock andProducts, and Fats and Oils. However, some ofthe 22 commodities do not fall into one of thesefour groupings. For example, sugar is not in-cluded in any special group. Furthermore, thegroupings are not mutually exclusive. Lard, forinstance, is in both the Livestock and ProductsIndex and in the Fats and Oils Index.


The prices used in the index are obtainedfrom trade publications or from other Govern-

Exchanges which issue spot prices have committees tomake a determination of the spot price for the standardcommodity.

119


ment agencies. Prices for cocoa beans, steers,sugar, wheat, burlap, copper scrap, cotton, leadscrap, print cloth (spot), rosin, rubber, steelscrap, wool tops, and zinc, are of the samespecification and source ac those used in thecomprehensive monthly Wholesale Price In-dex.2 Prices for butter, corn, hides, hogs, lard,tallow, and tin are either differently specifiedspot prices or from different markets.

Selection of Products

The criteria for the selection of commoditieswere (1) wide use for further processing(basic), (2) freely traded in an open market,(3) sensitive to changing conditions significantin those markets, and (4) sufficiently homo-geneous or standardized so that uniform andrepresentative price quotations can be obtainedover a period cf time.

Subject to these restrictions, efforts were madeto include representative sensitive commoditiesfrom as large a segment of the economy as pos-sible. Also, the influence of international mark-ets upon the economy was taken into account bythe inclusion of some key commodities (such ascrude rubber and tin) which are important ininternational trade. Both in the sample and inthe index structure, an attempt was made toprevent price movements of agricultural prod-ucts from dominating the movement of theindex.

Estimating ProceduresThe Spot Market Index is an unweighted

geometric mean 3 of the individual commodityprice relatives, i.e., of the ratios of the currentprices to the base period prices. The use of thegeometric mean has the advantage that theindex is not dominated by extreme price move-ments of individual commodities. Since ex-tremely large movements may be atypical, itwas deemed better to minimize their effects,even at the expense of losing the effect of largerepresentative changes. However, the fact that

*See Chapter 11, Wholesale Prices.n 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 (1.5X2X9=21. 301=3). The arithmetic mean,

4.2.

each of the commodities is unweighted in tht:index means that a price change for rosin, acomparatively unimportant commodity, has asmuch effect as an equal percentage movementin the price of a very important commoditysuch as wheat, cotton, or steel scrap.

The computation procedure involves obtain-ing for each commodity the ratio of its price inany given period to its price in the base periodand taking the 22nd root, of the product ofthese ratios. This product is then multipliedby 100 to obtain the index number for eachperiod. The calculation is made by means oflogarithms. The formula reduces to

Log ik=x Log Pk-1 log Po-I-4422

wherelk, = Index for a given dayPk = Price for a given dayPo = Average (geometric) price in base

period44 = Logarithmic constant which when di-

vided by 22 equals log of 100.Monthly average indexes are obtained ac-

cording to the previous procedure, except thatPk=the geometric 1 average of the, Tuesdayprices (daily prices prior to 1969) over themonth. In maintaining the index over time, itmay 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.


Tuesday spot market indexes and prices arepublished each week, on the Friday followingthe day of reference. A summary of weeklyindexes and the average for each month arepublished with the first weekly release of thefollowing month. Beginning with 1950, his-torical indexes are shown for Tuesday of eachweek together with monthly averages ; fromJuly 1946 through 1949 indexes are listed forTuesday 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.

1.20

1

tit

SPOT MARKET PRICES 121


A survey of users in 1964 showed that theIndex is frequently used as a general economicindicator, for gaging the direction of basicprices, for forecasting general price move-ments, and for current prices of specific com-modities. Other uses, frequently mentioned, arefor market research and for comparing pricetrends with the user's selling or buying prices.

The Tuesday Index of Spot Market Prices dif-fers from the Wholesale Price Index in methodof construction and weighting, as well as in thesample of items for which prices are included.While it is independent of the monthly compre-hensive index, changes in the Tuesday Indexor its components may foreshadow turns inWholesale Price Indexes. However, the Tues-day Index is not a good indicator of currentprice trends for the whole economy. For thispurpose, the comprehensive Wholesale PriceIndex should be used. The Tuesday SpotMarket Index is, by design, very sensitive toprice changes in basic commodities but, becauseof its unweighted structure, the magnitude ofchanges in any of the index groups cannot beused as a reliable measure of the general pricechange of all commodities within the groups.

For many of the 22 items, the commodity ex-change prices are based upon transactionswhich cover as little as 25 percent of the totalsold in all markets. In some cases, the priceis 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 marketprice, a "nominal" spbt price is set, From this,it is apparent that the exchange prices maynot always be representative of the large vol-ume of private transactions occurring outsidethe organized market. However, it is believedthat the reported exchange prices generally areused as the basis for private negotiations.

Composition of Grouping Indexes

Metals: Copper scrap, lead scrap, steel scrap,tin, and zinc.

Textiles and Fibers: Burlap, cotton, printcloth, and wool tops.

Livestock and Products: Hides, hogs, lard,steers, and tallow.

Fats and Oils: Butter, cottonseed oil, lard,and tallow.

Specifications for

CommodityBurlap

ButterCocoa beans..Copper scrap.

Corn _

CottonCottonseed

oil.Hides

Commodities Included in the Irides as ofMarch 1971

Specifications10 oz., 40", ex-dock or ex-

warehouse, duty paid, peryd.

Grade A, 92 score, per lb. _Accra, per lb.No. 1 heavy copper and wire,

refiners' buying price, car-load lots, delivered buyers'works, per lb.

No. 3 yellow, per bu.Middling, 11,46", per lb.Crude, valley, per lb.

MarketNew York.

Chicago.New York.New York.

Chicago.12 markets.Memphis.

Cow, light native, packer P0/53 Chicago.lbs., fleshed, packer to tan-ner, dealer, or exporter perlb., f.o.b. shipping point.

Hogs U.S. No. 2's and 3's, 200-220 Omaha.lbs., per 100 lb.

Lard Prime Steam, in tanks, per Chicago.lb,

Lead scrap Battery plates, smelters' buy- New York.ing price, East, carload lots;delivered buyers' works, perlb. -

Print cloth 48", 78x78 count, 4 yds./lb. New York.spot and nearby, per yd.

Print cloth 48", 78x78 count, 4 yds./lb., New York.most distant contract, peryd.

Rosin Gum, WG grade, carlots, per New York.100 lb.

Rubber Crude, natural, No. 1 Ribbed New York.Smoked Sheets, oer lb.

Steel scrap No. 1 heavy melting, (dealer), Chicago.consumers' buying price, in-cluding brokerage, de-livered, per gross ton.

Choice, 900-1100 lbs., per 100 Omaha.lb.Raw, 98, duty paid, per 100 New York.

lb,Fancy, bleachable, inedible, Chicago.

per lb.Grade A, spot delivery, per New York.

lb,Wheat ....___ No. 1 Dark Northern Spring, Minneapolis.

per bu,Wheat _ _ _ No. 1 Hard Winter Ord., per Kansas City.

bu.Wool tops _. Certificated spot price, nom- New York,

inal, per lb.Slab, Prime Western, for New York.

prompt delivery, delivered,(f.o.b. New York equiva-lent), per lb.

Steers .

Sugar ___

Tallow ____-_

121-LLOYD E. WIGREN

Wages and Industrial Relations

Chapter 14. Occupational Pay and Supplementary Benefits

Background

The Bureau of Labor Statistics, for manydecades, has conducted studies of wages by oc-cupation and industry, based upon employerrecords. The Bureau's first such study, growingout of a study by the U.S. Senate in 1891, re-sulted in a wage rate record extending backcontinuously to 1860. Systematic collection ofwage data by occupation and industry has con-tinued since the turn of the century ; changesin coverage has been dictated mainly by gov-ernment requirements. A large survey programundertaken for the War Industries Board in1919 produced occupational pay rates by indus-try and State, and (for some industries) bycity. Between 1934 and 1940, the selection ofindustries studied was determined largely byadministrative needs under the National Re-covery Act, Public Contracts Act, and the FairLabor 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 towar production. Implementation of wage sta-bilization policy during the war required alarge-scale program of occupational wage stud-ies by industry and locality. The emphasis ondata by locality has continued since 1945 withinthe framework of industry studies generallydesigned to also yield national and regionalestimates. In addition, the Bureau developedtwo new types of surveys.

Area wage surveys, initiated in the late1940's, were designed to meet the growing de-mand for pay data related to office clerical andmanual jobs that are common to a wide varietyof manufacturing and nonmanufacturing in-dustries within metropolitan areas. This surveyprogram was firmly established and tem-porarily expanded for use in the wage stabili-zation effort during the Korean emergency. Theneed for nationwide estimates of white-collarpay in private industry for use in appraising

the Federal white-collar salary structure re-sulted in a survey design that would producenational averages, based on an area sample.Data for individual areas studied also serve thewage administration needs for other govern-ment agencies.

Prior to 1960, studies in a very few profes-sions provided salary data. Beginning in thatyear, salary surveys have been made on anationwide 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 anddrafting jobs included in the area wage sur-veys, are utilized by the administrative agen-cies directly concerned with Federal paymatters.

Description of Surveys

Although differing in industrial, geographic,and occupational coverage, the three types ofsurveys described form an integrated programof occupational wage surveys based upon acommon set of administrative forms, manual ofprocedures, and common concepts and defini-tions. Employer cooperation in surveys is on avoluntary basis. Confidential individual estab-lishment data compiled by the Bureau's fieldeconomists are grouped in published reports ina manner that will avoid possible disclosure ofan establishment's rates. Establishments in-cluded in all surveys are classified by industryas defined in the 1967 edition of the StandardIndustrial Classification Manual prepared by

le U.S. Office of Management and Budget.'survey reports identify the minimum size ofestablishment (measured by total employment)studied. Definitions for Standard MetropolitanStatistical Areas are employed in all pro-grams.2

See appendix B.2 See appendix C.

122


Industry wage surveys provide data for occu-pations selected to provide representativenessof the range cif rates, methods of wagepayment, and of men's and women's work ac-tivities. Consideration also is given, in theirselection, 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 first-shift rates (or hours and earnings for incentiveworkers) for individual workers in the selectedoccupations, surveys in most industries alsoestablish the ws.ge frequency distribution for1-,1-d employment groups, i.e., production andtelated workers or nonsupervisory workers.Weekly work schedules; shift operations anddifferentials ; 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 studiesalso provide estimates of labor-managementagreement coverage, proportions employed un-der incentive pay plans, and the extent to whichestablishments provide a single rate or rangeof rates for individual job categories.

Fifty manufacturing and 20 nonmanufactur-ing industries, accounting for about 22.5 mil-lion employees, are surveyed on a regularlyrecurring basis. A majority are studied on a5-year cycle, but a number of comparativelylow-wage industries are on a 3-year cycle. Inaddition, special wage surveys also are under-taken at the request of other governmentagencies.

Nearly all of the manufacturing, utilities,and mining industries are studied on a nation-wide basis and estimates are provided also forregions and major areas of concentration. Sur-veys in trade, finance, and service industriesusually are limited to a score or more of metro-politan areas. Nationwide surveys generallydevelop separate estimates by size of establish-ment, size of community, labor-managementagreement coverage, and type of product orplant group.

Area wage surveys provide data for occupa-tions common to a wide variety of industries inthe communities surveyed. The 76 occupational

categories studied include 31 office clerical ; 15electronic data processing, drafting, and indus-trial nurses; and 30 maintenance, toolroom,powerplant, and custodial and material move-ment jobs. Thus, they provide representationof the range of duties and responsibilities as-sociated with white-collar, skilled maintenancetrades, and other "indirect" manual jobs.Weekly salaries reported for individuals inwhite-collar jobs relate to regular straight-timesalaries that are paid for standard workweeks.Average hourly earnings for maintenance andother manual jobs relate to first-shift hourlyrates.

Industry divisions included are (1) manu-facturing; (2) transportation, communication,and other public utilities; (3) wholesale trade ;(4) retail trade ; (5) finance, insurance, andreal estate ; and (6) selected service industries.Establishments employing fewer than 50 work-ers are excludedwith a minimum of 100applying to manufacturing; transportation,communication, and other public utilities ; andto retail trade in the dozen largest communities.

In addition to the all-industry averages anddistributions of workers by earnings classes,separate data are provided for manufacturingand nonmanufacturing in each area and,wherever possible, for individual industry di-visions in the nonmanufacturing sector. Amongthe 89 Standard Metropolitan Statistical Areasin this annual survey program as of 1971, sep-arate data are provided for transportation,communication, and other public utilities in 88areas; for retail trade in 25 areas ; for whole-sale trade and finance, insurance, and realestate in 17 areas ; and for the selected serviceindustries in 8 large areas. In 22 of the largerareas, wage data are presented separately forestablishments that have 500 workers or more.

Data on weekly work schedules ; paid holidayand vacation practices ; and health, insurance,and pension benefits are recorded separatelyfor nonsupervisory officeworkers and plantworkers (nonoffice). Shift operations and dif-ferentials are collected for plantworkers inmanufacturing. Data on minimum entrancerates for inexperienced officeworkers are col-lected in all industries. These items are studiedbiennially in all areas. This survey program

123

OCCUPATIONAL PAY AND

also has developed information on profit-shar-ing plans, characteristics or sick leave plans,wage payment systems, and other items relatedto employee compensation.

Special area wage surveys have been con-ducted annually since 1967 at the request of theEmployment Standards Administration for usein administering the Service Contract Act of1965. The surveys also meet the needs of thegeneral public and provide information onhourly earnings for 14 office occupations ; 22maintenance, toolroom, powerplant, and cus-todial and material movement jobs ; 7 laundryjobs; and 6 food service jobs.

The industrial scope includes manufactur-ing; transportation, communication, and otherpublic utilities; and wholesale trade industrydivisions ; and general merchandise stores;eating and drinking places; real estate, hotelsand other lodging places; engineering andarchitectural services; personal services ; andmiscellaneous business services. Data on inci-dence of paid holidays and vacation practices,and health insurance and pension benefits areprovided biennially.

The National Survey of Professional, Admin-istative, Technical, and Clerical Pay providesa fund of broadly based information on salarylevels and distributions in private employment.The 80 occupation-work levels studied wereselected from the following fields : Accounting,legal services, personnel management, engi-neering and chemistry, buying, clerical super-visory, drafting, and clerical. Definitions forthese occupations provide for classification ofemployees according to appropriate work levels(or classes). Although reflecting duties andresponsibilities in industry, the definitions weredesigned to be translatable to specific paygrades in the General Schedule applying toFederal Classification Act employees. This sur-vey, thus, provides information in a form suit-able for use in comparing the compensation ofsalaried employees in the Federal civil servicewith pay in private industry.

Average salaries, monthly and annual forall occupations and also on a weekly basis forclerical and drafting, relate to the standardsalaries that were paid for standard workschedules, i.e., to the straight-time salary cor-

tt

SUPPLEMENTARY BENEFITS 125

responding to the employee's normal workschedule, excluding overtime hours. Nationwidesalary distributions and averages are presentedfor men and women combined. Averages alsoare presented for establishments in metropoli-tan areas combined and for establishmentsemploying 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) engineeringand architectural services, and commerciallyoperated research, development, and testinglaboratories.

Limited to the Nation's metropolitan areasfor the years 1960 through 1964, the annualsurvey was expanded in 1965 to include non-metropolitan counties. The minimum establish-ment size included in the survey was raisedfrom 100 to 250 in 1961. In 1966, the minimumestablishment size was lowered to 100 in trans-portation, communication, and the other publicutilities ; wholesale trade; and the service in-dustries studied. The minimum was also low-ered to include establishments with 50 workersin the finance, insurance, and real estate in-dustries. Since the survey scope is subject tochange, users are directed to, the Scope andMethod of Survey. appendix in the reports fora description of current practice.3

Concepts. The Bureau's occupational wagesurveys summarize a highly specific wagemeasurethe rate of pay, excluding premiumpay 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, incentivepay plans, an earned rate is computed by di-viding straight-time earnings for a time periodby corresponding hours worked. Productionbonuses, commissions, and cost-of-living bo-nuses are counted as earnings. In general,bonuses that depend on factors other than theoutput of the individual worker or group ofworkers 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 particularsurNey being described.

k

424


year-end or Christmas bonuses, and cash dis-tributions under profitsharing plans.

Unless stated otherwise, rates do not includetips or allowances for the value of meals, room,uniform, etc. The earnings figures, thus, rep-resent cash wages (prior to deductions forsocial security, taxes, savings bonds, premiumpayments for group insurance, meals, room oruniforms) after the exclusion of premium payfor overtime, weekend, holiday, or late shiftwork.

Hours shown for salaried occupations relateto standard weekly hours for which the em-ployee receives his regular straight-time salary.

Occupational classifications are defined inadvance of the survey. Because of the emphasison interestablishment and interarea compar-ability of occupational content, the Bureau'sjob descriptions may differ significantly fromthose in use in individual establishments orthose prepared for other purposes. The jobdescriptions used for wage survey purposes aretypically brief and usually more generalizedthan those used for other purposes, The pri-mary objective of the descriptions is to identifythe essential elements of skill, difficulty, and

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 anestablishment. Work involves most of the following: In-terpreting written instructions and specifications; planningand laying out of work; using a variety of machinist's hand-tools and precision measuring instruments; setting up andoperating standard machine tools; shaping of metal parts toclose tolerances; making standard shop computations relatingto nimensions of work, tooling, feeds, and speeds of machin-ing; knowledge of the working properties of the commonmetals; selecting standard materials,' parts, and equipmentrequired for his work; and fitting and assembling parts intomechanical equipment. In general, the machinist's worknormally requires a rounded training in machine-shop prac-tice usually acquired through a formal apprenticeship orequivalent training and experience.

In general, workers are included in a classification if theduties as described are performed a major part of the timeand the remainder is spent on related duties requiring similaror 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 forclassification purposes is given to those elements of the jobwhich are most important in determining its level for paypurposes. Thus, a worker meets the basic concept of thestenographer classification if taking of dictation is a regularrequirement of the job even though a majority of time isspent on routine typing.

responsibility that establish the basic conceptof the job.'

Although work arrangements in any one es-tablishment may not correspond precisely tothose described, those workers meeting thebasic requirements established for the job areincluded .5

In applying these job descriptions, the Bu-reau's field representatives exclude workingsupervisors, apprentices, learners, beginners,trainees, handicapped workers, part-time ortemporary workers, and probationary workersunless 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 applicableto all nonsupervisory plant or office workers ifa majority of such workers are eligible or canexpect eventually to qualify for the practiceslisted. Data for health, insurance, and pensionplans are limited to those plans for which atleast 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 Governmentrepresentatives to obtain views and recom-mendations related to scope, timing, selection,and definitions of survey items, and types oftabulations. Particularly in planning surveys inspecific industries, these discussions impor-tantly supplement comments and suggestionsreceived from the regional offices at the con-clusion of the previous study. Reflecting its usein evaluation of Federal white-collar pay, thedesign of the National Survey of Professional,Administrative, Technical, and Clerical Paywas developed in conjunction with the Officeof Management and Budget and the CivilService Commission. Changes in the surveyscope, item coverage, and job definitions areinitiated by these agencies.

The industrial scope of each survey is identi-fied in terms of the classification system pro-vided in the Standard Industrial ClassificationManual. The scope may range from part of a

125

OCCUPATIONAL PAY AND SUPPLEMENTARY BENEFITS 127

4-digit code for an industry study to a uniformcombination of broad industry divisions andspecific industries for the area wage surveys orthe salary survey of professional, administra-tive, technical, and clerical jobs. The needs ofmajor users are a major consideration in de-signing the multi-purpose occupational studies.

The minimum size of establishment includedin a survey is set at a point where the possiblecontribution of the excluded establishments isregarded as negligible for most of the occupa-tions surveyed. Another practical reason forthe adoption of size limitations is the difficultyencountered in classifying workers in small es-tablishments where they do not perform thespecialized duties indicated in the job defini-tions.

Considerations in timing of industry surveysinclude date of expiration of major labor-management agreements, deferred wage ad-justments, seasonality of production (e.g.,garments), and interests of users. Whereverpossible, area wage surveys are timed to followmajor wage settlements as well as to meet theneeds of government agencies engaged in wageadministration as required by law.

The types of occupations studied and criteriaused in their selection were identified in the de-scription of the various types of surveys. Thejob list for each survey is selected to representa reasonably complete range of rates in thewage structure for the employment categoriesinvolved, i.e., production and related workersin a specific manufacturing industry or non-supervisory office, maintenance, material han-dling, and custodial workers in a metropolitanarea The established hierarchy of job rates tobe found within establishments and industriespermits the use of pay data for such key orbenchmark jobs for interpolating rates forother jobs. Technological developments or userinterests may dictate changes in the job listsand definitions. New definitions for jobs usuallyare pretested in a variety of establishmentsprior to their use in a full-scale survey.

Questionnaires. Two basic schedules are usedin obtaining data in all surveys. The first (BLS2751A) includes items relating to products orservices, employment, shift operations and dif-ferentials, work schedule, overtime premiums,

paid holidays and vacations, insurance andpension plans, union contract coverage, andother items applicable to the establishment. Thesecond (BLS 2753G) is used in recording oc-cupation, sex, method of wage payment, hours(where needed), and pay rate or earnings foreach worker studied. Supplementary forms areused to meet particular needs.

Collection. Bureau field economists collect databy personal visit to each of the sample estab-lishments. Job functions and factors in theestablishment are carefully compared withthose 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, onoccasion, observation of jobs within plants. Asatisfactory completion of job matching per-mits acceptance of company-prepared reportswhere this procedure is preferred by the re-spondent. Generally, however, the field eco-nomist secures wage or salary rates (or hoursand earnings, when needed') from payroll orother records and data on the selected employerpractices and supplementary benefits fromcompany officials, company booklets, and labor-management agreements.

Area wage surveys in all areas involve per-sonal visits every second year with partialcollection by mail in the intervening years.Establishments participating in the mail collec-tion receive a transcript of the job matchingand wage data obtained a year earlier by thefield economist, together with the job defini-tions. The up-dated returns are scrutinized andquestionable entries are checked with the re-spondent. Personal visits are made to establish-ments not responding to the mail request and tothose reporting unusual changes from year-earlier data.

The work. of all field economists is checkedfor quality of reporting, with particular atten-tion directed to accuracy in job matching. Therevisits are made by supervisory and senioreconomists. Systematic technical audits of thevalidity of survey definitions, made by staffwith specialized training, also are maintainedfor the technically complex nationwide white-collar salary survey.

AiY

sa

zs

128

BLS 2751A(Rev. January 1970)


U.S. DEPARTMENT OF LABORBureau c; Labor Statistics

WAGE SURVEYGENERAL ESTABLISHMENT INFORMATION

1. ESTABLISHMENT IDENTIFICATION

Budget Bureau No. 44-R033BApproval expires November 30, 1972

Your report will beheld in confidence

A. Survey Payroll Period19--

Payroll Period19--

ESTABLISHMENT NAME

STREET ADDRESS CITY

COUNTY, STATE AND ZIP CODE AREA CODE - TELEPHONE

NAME AND TITLE OF AUTHORIZING OFFICIAL

19...

NAME AND TITLE OF OFFICIAL SUPPLYING DATA

ISADDRESS OF OFFICE FROM WHICH DATA WAS OBTAINED, IF DIFFERENT FROM ABOVE

a obtained from this source)TELEPHONE

B. Central Office (Complete if clearance and/or data

NAME OF COMPANY NAME OF AUTHORIZING OFFICIAL

STREET ADDRESS CITY, STATE AND ZIP CODE

TITLE

2. CURRENT PRODUCTS OR SERVICES AND PROCESSES

A. PRODUCT OR SERVICEAPPROXIMATE

% ANNUALVALUE

PRODUCT OR SERVICEAPPROXIMATE

% ANNUALVALUE

5. SCOPE OF OPERATIONS

19 19

3. OFFICE USE ONLY

YEARSCHEDULE NO.

1-5

(DENT.

6.8

AREA

9.11

REGION

12

STATE

13-14

CITYSIZE

15

SIC CODE

16.19

EST.SIZE

20

UNION

21

WEIGHT

22.24

SPECIAL CHAR.ACTERISTICS

III25-27

121

28-30

19

19

OCCUPATIONAL PAY AND SUPPLEMENTARY BENEFITS

BLS 2751A Continued U.S. DEPARTMENT OF LABOR(Rev. Jonuory 1970) Bureau of Labor Statistics

129

SURVEY

ESTABLISHMENT

PAYROLL PERIOD SCHEDULE NO.

5. UNION

A. Are

B. Arc

C. With(Civ

sImmarI

........ ._....__

a majority of your production workers covered by union agreements)

a majority of your office workers covered by union agreements)

what unions does this establishment have contracts?e name and affiliation below.)

No Yes

31 on 1( 1 T32 On i [7 F--

D. What occupational groups(List groups below opposite

are covered by the contract?the appropriate union.)

Production Workers:

Office Workers:

6. ESTABLISHMENT EMPLOYMENT (APPROXIMATE)

A. What is the approximate total employment' in this establishment? 33.38

B. Bow many are nonsupervisory production (plant) workers? 39.43

Men 4448

Women 49.53

C. Nonsupervisory office workers? 5458

Men59.63

Women 64.68

D. Other employees (executive, professional, supervisory, etc.)?_ 69.73

E. 74.78

F.End of card-4.80

Includes salaried officers of corporations but does not include proprietors, members of unincorporated firms, pensioners,members of the armed forces carried on the payroll, or unpaid family workers.

G. Remarks

128

vr

130

BLS 275 30Roo, 63

Schedule No.


ht. Name

OCCUPATIONAL RATES Payroll period

OCCUPATION AND GRADEOccupational

codeIII

Sea(2)

Method ofpay(31

Number ofworkers

(41Hours

131

Salary, rote, arearnings

lel

LineNo.171 (6)

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

if:928


Sampling

The sampling design employed is almost al-ways highly stratified. Before the sample isselected, information on all known establish-ments that might possibly fall within the scopeof the survey is compiled from lists provided byregulatory governmental agencies (primarilyState unemployment insurance agencies), sup-plemented by data from trade directories, tradeassociations, labor unions, and other sources.

Establishments then are stratified as pre-cisely as available information permits. Eachgeographic-industry unit for which a separateanalysis is to be presented is sampled inde-pendently. Within these broad groupings, afiner stratification by product (or other perti-nent attributes) and size of establishment ismade. Stratification may be carried still furtherin 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 theoccupational structure of the various industrysegments differs widely.

The sample for each industry-area group is aprobability sample, each establishment havinga predetermined chance of selection. In order tosecure maximum accuracy at a fixed level ofcost (or a fixed level of accuracy at minimumcost), the sampling fraction used in the variousstrata' ranges downward from all large estab-lishments through progressively declining pro-portions of the establishments in each smallersize group, in accordance with the principlesof optimum allocation. Thus, each sampledstratum will be represented in the sample by anumber of establishments roughly propor-tionate to its share of the total employment.Though this procedure may appear at first toyield a sample biased by the over-representa-tion of large firms, the method of estimationemployed avoids the possibility of bias by theassignment of proper weights to the sampleestablishments.

In the event a sample establishment withinscope is unable to supply usable data, a sub-stitute is assigned in the same industry-loca-tion-size class. (Since no close relation existsbetween failure to participate in these surveys

and the items being studied, little bias is intro-duced by this procedure.) The overall non-response rate in published surveys averagesabout 5 percent, and in exceptional casesreaches 10 percent.

The size of the sample in a particular surveydepends 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 accuracyrequired. Estimates of variance based on datafrom previous surveys are used in determiningthe size of the sample needed.

As indicated earlier, area wage surveysare limited to selected metropolitan areas.These areas, however, form a sample of allsuch areas, and, when properly combined(weighted), yield estimates of the national andregional levels.

The sample of areas is based on the selectionof one area from a stratum of similar areas.The criteria of stratification are region, type ofindustrial activity as measured by percentageof 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 self-representing, i.e., each one forms a stratum byitself and is certain of inclusion in the areasample. The area sample contained about 80percent of all nonagricultural employment ofthe metropolitan area complex of the entirecountry in 1970.

In some strata it was impossible to give someareas their proper chance of selection, becauseof difficulties in making surveys in these areasprimarily due to predominance of single em-ployers. Then, too, some areas were chosen torepresent some strata because of interest in thearea of itself. These departures from a strictprobability design, covering only 14 percent ofthe total, are believed to be negligible in theireffect.

Estimating ProceduresEstimated average earnings (hourly, weekly,

monthly, or annual) for an industry or an oc-cupation are computed as the arithmetic mean

130


of the individual employees' earnings. Tl.ey arenot estimated by dividing total payrolls by thetotal time worked, since such information al-most never is available on an occupationalbasis.

All estimates are derived from the sampledata. The averages for occupations, as well asfor industries, are weighted averages of indi-vidual earnings and not computed on an es-tablishment basis. The proportion of employeesaffected by any fringe provision likewise isestimated from the sample; all workers in eachestablishment are considered to be covered bythe predominant benefit policy in effect, andthe entire employment of the establishment isclassified accordingly.

As mentioned previously, the use of a vari-able sampling ratio in different strata of thepopulation would result in biased estimates ifstraight addition of the data for the variousestablishments were made. Therefore, each es-tablishment is assigned a weight that is theinverse of the sampling rate for the stratumfrom which it was selectede.g., if a third ofthe establishments in one stratum are selected,each of the sampled establishments is given aweight of 3.

To illustrate the use of weights, suppose theuniverse were 7 establishments, from which asample 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 thesample. It therefore, is, given a weight of 2.Establishment B, on the other hand, was takenwith certainty (or a probability of 1) and isthus given a weight of 1. Establishment C wastaken from a group where a fourth of all plantswere used in the sample, and hence is given aweight of 4. The following calculations aremade in estimating average earnings for agiven occupation.

Workers in occupationin sample establish-

ments at specified rate

Average Estimates of totalhourly in stratum

Total earn-Establishment Weight number ings Workers earnings

2 40 $1.60 2x40 2x40x$1.601 30 1.70 1x30 1 x30x 1.70

20 1.95 1x20 1 x20 x 1.954 10 1.65 4x10 4 x10x 1.65

Estimated universe 170 3284.00

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 estimatethe proportion of employees in establishmentsgranting paid vacations of 2 weeks after 2years of service, for instance, the establish-ments are classified according to the length ofvacation granted after 2 years' service, estab-lishment weights are applied to employment,as in the previous example, and the proportionof the estimated employment in the 2-weekcategory, of the estimated total employmentthen is computed. Using the same three estab-lishments as in the previous example, this canbe illustrated as follows:

Actual total Weighted Vacationestablishment employ- provisions

Establishment Weight employment ment after 2 yearsA..____.. 2 100 200 1 week.B 1 500 500 2 weeks.C 4 75 300 1 week.Estimated universe 1,000

Thus, the estimated percentage of workers inestablishments granting 2 weeks' vacation

after 2 years of service is 1i00 or 50 percent.1,000

When a large establishment within surveyscope, for which no substitute exists, is unableto supply data, the deficiency is alleviated byincreasing the weight of the most nearly sim-ilar units. Should any segment be affected bya substantial amount of such noncooperation,the publication of materials will be diminishedby omitting separate presentation of sectorsseriously affected.

Where a sample of selected metropolitanareas is used to represent the totality of suchareas, a second stage of weighting is used toexpand the individual area totals to regionand/or national estimates. Since, as indicatedin the description of the sampling method, eacharea represents a stratum of similar areas, thetotal from each area are weighted to the esti-mated stratum totals by multiplying by theinverse of the chance of selection. This pro-cedure provides the ratio of nonagriculturalemployment in the stratum to that in the

11


sample area (one in the case of the large self-representing areas). Summing all such esti-mated stratum totals yields the earnings andemployment totals for the region and thecountry as a whole.


Where an industry survey is designed toyield estimates for selected States or areas,these are published separately as informationbecomes available from all sample firms in theState or area unit. Industry surveys limited toselected areas do not provide a basis for theexaminations of pay levels by size of com-munity, size of establishment, product, orlabor-management agreement coverage thatgenerally are included in bulletin reports onnationwide surveys. Regardless of geographicscope, industry survey reports record the inci-dence of incentive pay plans and, to the extentpossible, average pay levels separately for timeand incentive workers.

Individual bulletin reports on individual areawage surveys are supplemented by two sum-mary bulletins. The first compiles the resultsof individual area surveys made during a fiscalyear. The second contains information on oc-cupational earnings, employer practices, andsuplementary wage benefits for all metropoli-tan areas combined and by industry divisionwithin the four broad census regions.

Wage-rate indexes are constructed for broadoccupational categories, e.g., office clericalworkers, skilled maintenance workers, and un-skilled plant workers. These indexes are pub-lished annually, separately for all industriesand manufacturing, for each metropolitan areastudied, and for all metropolitan areas com-bined and by region. Area pay relatives for thethree occupational categories are publishedannually, permitting ready comparisons ofaverage pay levels among areas. Estimates oflabor-management agreement coverage are alsopresented annually. Occupational pay relation-ships within individual establishments aresummarized periodically.

Bulletin reports on the National Survey ofProfessional, Administrative, Technical, and

Clerical Pay present occupational averages anddistributions on an all-industry basis, nation-wide and separately for all metropolitan areascombined, and for establishments employing2,500 workers or more. Average pay levels forindustry divisions are shown as percentages ofthe all-industry averages. Year-to-year per-centage changes for occupation-work levels andtrend estimates for occupations are reported.

Industry and area wage survey reports areissued throughout tt e year as the surveys arecompleted. The bulletin on the National Surveyof Professional, Admistrative, Technical, andClerical Pay is made available in March.

Summaries of the data in the bulletins andspecial analyses appear also in the MonthlyLabor Review.


Occupational wage data developed in thesesurveys have a variety of uses. They are usedby Federal, State, and local agencies in wageand salary administration and in the formula-tion of public policy on wages, as in minimumwage legislation. They are of value to Federaland State mediation and conciliation servicesand to State unemployment compensation agen-cies in judging the suitability of job offers.Knowledge of levels and trends of pay rates byoccupation, industry, locality, and region is re-quired in the analysis of current economic de-velopments and in studies relating to wagedispersion and differentials.

Bureau data are used in connection withprivate wage or salary determinations by em-ployers or through the collective bargainingprocess. To the extent that wages are a factor,survey data also are considered by employersin the selection of location for new facilitiesand in cost estimating related to contract work.

Occupational wage survey programs are notdesigned to supply mechanical answers toquestions of pay policy. As suggested ,3arlier,limitations are imposed in the selection anddefinition of industries, of geographic units forwhich estimates are developed, of occupationsand associated items studied, and in determina-tion of periodicity and timing of particular

134 BLS HANDBOOK

surveys. Depending upon his needs, the usermay find it necessary to interpolate for occupa-tions or areas missing from the survey on thebasis of knowledge of pay relationships.

Because of interestablishment variation inthe proportion of workers in the jobs studiedand in the general level of pay, the surveyaverages do not necessarily reflect either theabsolute or relative relationships found in themajority of establishments. To illustrate, em-ployment in the specialized maintenance craftstends to be concentrated in the larger establish-ments, whereas employment in custodial andmaterial-movement jobs is distributed morewidely within an industry or area. Thus, to theextent that pay rates in the larger establish-ments vary from the average level, the skilldifferential measure based on the survey aver-ages will differ to some degree from thatobtainable within each of the larger estab-lishments.

The incidence of incentive methods of pay-ment may vary greatly among the occupationsand establishments studied. Since hourly aver-ages for incentive workers generally exceedthose for hourly-rated workers in the same job,averages for some incentive-paid jobs mayequal or exceed averages for jobs positionedhigher on a job evaluation basis but normallypaid on a time basis. Wherever possible, dataare shown separately for time workers andincentive workers in the industry surveys. In-centive plans (generally plant-wide in applica-tion) apply to only a very small proportionof the workers in the indirect plant jobs studiedin the area wage program.

Although year-to-year changes in averagesfor a job or job group primarily reflect generalwage and salary changes or merit increasesreceived by individuals, these averages alsomay be affected by changes in the labor forceresulting from labor turnover, labor force ex-pansions and reductions for other reasons, aswell as changes in the proportion of workersemployed in establishments with different paylevels. A labor force expansion might increasethe proportion of lower paid workers andthereby lower the average, or the closing ofa relatively high-paying establishment couldcause average earnings in the area to drop.

OF METHODS

Reliability of surveys. Results of the surveysgenerally will be subject to sampling error.This error will not be uniform, since, for mostoccupations, the dispersion of earnings amongestablishments and frequency of occurrence ofthe occupation differ. In general, the sample isdesigned so that the chances are 9 out of 10that the published average does not differ bymore than 5 percent from the average thatwould be obtained by enumeration of all estab-lishments in the universe. That error applies tothe smallest breakdown published. Hence, theerror present in broader groupings will besomewhat less.

The sampling error of the percentage ofworkers receiving any given supplementarybenefit differs with the size of the percentage.However, the error is such that rankings ofpredominant practices almost always will ap-pear in their true position. Small percentagesmay be subject to considerable error, but willalways remain in the same scale of magnitude.For instance, the proportion of employees inestablishments providing more that 4 weeks'paid vacation to long-service employees may begiven as 2 percent, when the true percentagefor all establishments might be only 1 percent.Such a sampling error, while considerable, doesnot affect the essential inference that the prac-tice is a rare one.

Estimates of the number of workers in agiven occupation are subject to considerablesampling error, due to the wide variationamong establishments in the proportion ofworkers found in individual occupations. (It isnot unusual to find these estimates subject tosampling error of as much as 20 percent.)Hence, the estimated number of workers canbe interpreted only as a rough measure of therelative importance of various occupations.The greatest degree of accuracy in these em-ployment counts is for those occupations foundprincipally in large establishments. This sam-pling error, however, does not materially affectthe accuracy of the average earnings shown forthe 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

133

j

it-

r.

0;

OCCUPATIONAL PAY AND

(i.e., the errors tend to be in the same direc-tion), the sampling error of the estimate (aver-age hourly earnings) is considerably smallerthan the sampling error of either total earningsor total employment.

Since completely current and accurate infor-mation regarding establishment products andthe creation of new establishments is not avail-able, the universe from which the sample isdrawn may be incomplete. Sample firms in-correctly classified are accounted for in theactual field work, and the universe estimatesare revised accordingly. Those firms whichshould have been included but were classifiederroneously in other industries cannot be ac-counted for.

SUPPLEMENTARY BENEFITS 135

Since some measure of subjective judgmententers into the classification of occupations andother characteristics, there is some reportingvariability in the results. A repetition of thesurvey in any establishment with different in-terviewers and respondents would undoubtedlyproduce slightly different results. However,when spread over a large number of establish-ments the differences, being random, wouldtend to balance out. Hence, analyses based on asmall number of respondents must be used withcare, even when all eligible establishments areincluded. No evidence of any consistent errorhas been uncovered.


Number1. Cohen, Samuel E., "Studies of Occupational Wages and Supplementary Berefits." 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 shortdiscussion 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. Liststhe type of survey and cycle for each of 70 industries studied separately, and identifies thearea sample as originally determined for the labor market survey program.

GEORGE L. STELLUTO

Chapter 15. Employee Earnings and Hours Frequency Distributions

Background

An extensive program of studies of the fre-quency distribution of employee earnings hasbeen maintained by the Bureau of Labor Sta-tistics since 1954. Although the need for such.data had long been recognized and some workhad been done in the area, the program didnot receive its full impetus until 1955 whenCongress, in an amendment to section 4 (d) ofthe Fair Labor Standards Act, directed theSecretary of Labor to include in his annualreport an evaluation and appraisal of minimumwages established by the act and his legislativerecommendations.' To meet these requirements,the program of employee earnings distributionsstudies was established. Since 1964, frequencydistributions of weekly hours of work havebeen a part of this program.

This program was initiated to provide knowl-edge about the internal structure of earningsand hours which an average does not reveal.For example, distribution studies show whetherearnings or hours are dispersed evenly or tendto be clustered around one or more points; theyshow the gap between the highest and lowestpaid workers; they reveal the proportions ofemployees working short or long hours, andsimilar information on individual earnings andhours. They serve to locate the employed"poor," i.e., those not sharing in the generalrise in living standards available from work.

Description

Studies of employee earnings and hours dis-tributions generally include within their scopeall nonsupervisory employees. No attempt ismade to classify employees by occupation, al-though at times information is collected sep-arately for some types of employees. The datacollected relate to straight-time hourly earn-

The Bureau's studies conducted in this area before 1954frequently related to selected occupations, or were limited torelatively narrowly defined industries.

'See technical references for a listing of publications con-taining full descriptions of both types or studies.

,See appendix B.See appendix C.

ings, excluding premium pay for overtimework and for work on weekends, holidays, andlate shifts. Incentive pay, production bonuses,and cost-of-living payments are included inearnings, but nonproduction bonuses (e.g.,Christmas bonuses) are not. Hours informa-tion is collected for hours worked during theweek and for hours spent on vacations, holi-days, or sick leave for which pay is received.Earnings and hours information is obtained foreach employee, making it possible to calculateaverage hourly earnings and weekly hours ofwork for the entire group as well as to showthe distribution of employees by earnings,hours, or both.

Two basic types of distributions studies areconductedindustry and area.2 Industry stud-ies may have broad coverage, such as manu-facturing or wholesale trade, or they may belimited to specific industries, such as motorcarriers. 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 whichthe industry is concentrated. Industries aredefined on the basis of descriptions in the cur-rent Standard Industrial Classification Manual,published by the Office of Management andBudget.8

Area studies are limited in geographic cover-age, perhaps to a Standard Metropolitan Sta-tistical Area,' to a county or a group ofcontiguous counties, or to a region or part ofa region. This type of study usually is con-ducted on a cross-industry basisthat is, earn-ings and hours data are collected for mostindustry 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 exceptionsmay be made.


Data generally are obtained from employerpayroll records, using one of two collectionmethods. Establishments may be solicited for

105


information by mail or by personal visit. Thoseestablishments contacted by mail receive aquestionnaire form and a note requesting co-operation in the survey and explaining thenature and purpose of the study. Typically, thequestionnaire requests information on the prod-uct or service of the establishment and numberof employees, as well as other establishmentinformation which may be pertinent to thesurvey. The respondent is requested to reporteach employee's earnings and weekly hoursworkedthe primary focus of the survey. Pro-vision is made for the reporting of earnings onan hourly basis or on a salary or incentivebasis. Iii the last two instances, the respondentreports the number of hours paid for duringthe salary or incentive period, thus permittingthe computation of average hourly earnings foremployees paid on other than an hourly basis.

Certain large establishments, or companiesfrom which data for several establishments arerequested, are visited personally by the Bu-reau's field economists who may prepare thedata for the survey from company records, orarrange with the company for the completionof the questionnaire form. In addition, a sam-ple of establishments which do not reply to themail-questionnaire is visited by Bureau fieldeconomists to obtain the desired information.

To limit errors caused by incorrect reportingby the respondent, questionnaires received bymail are reviewed for reasonableness and con-sistency. Where data are questionable, a letteris written to the respondent asking him to re-view the item to assure its accuracy.

Sampling

Earnings and hours distribution studies areconducted on the basis of a sample of all es-tablishments within the scope of the survey.The sample generally is derived from StateUnemployment Insurance (UI) listings whichshow reporting units with four employees ormore by location, number of employees, andindustry classification. In industries whereestablishments with fewer than 4 employeesare of numerical importance (e.g., retail trade)the UI lists may be augmented by sources

such as other government agencies, or tradedirectories.

The size of the sample depends on severalfactors, among which are the size of the uni-verse, the distribution of establishments bynumber of employees, the relative dispersionof earnings among establishments, the degreeof accuracy required, and the cost of obtainingthe data. Estimates of variance based on datafrom previous surveys also may be used in de-termining the appropriate size of the sample.

The sample usually is selected using a highlystratified probability sampling design. Estab-lishments are first grouped, or stratified, ac-cording to industry, geographic location, andemployment size. Establishments in specificindustries or areas for which earnings andhours data are to be presented separately aregrouped independently of establishments inother strata, and sampled separately.

In order to obtain maximum accuracy perunit cost, the number of establishments in thesample is distributed among the various stratain the most efficient manner, in accordance withthe principles of optimum allocation. A fractionof establishments in each stratum is includedin the sample, with the sampling fractiondiminishing as the employment size of thestrata decreases, so that the probability of in-clusion in the sample is greater for the largethan for the small establishment. Frequentlythe entire stratum containing the largest estab-lishments is included in the sample.

The following example shows a hypotheticaluniverse of 47 establishments in three employ-ment-size groups. By applying the appropriatesampling ratio to each size group (stratum),the number of sample establishments is de-termined.

Weight ofNumber of Number eachestablish- Sam- of sample sample

Employment inents piing establish- establish-size group in universe ratio punts meet

Under 25 . .. _ 32 1/4 8 425-49 12 1/3 4 350 and over 3 1/1 3 1

No assumption is made that the earnings andhours structures of establishments not respond-ing to the mail questionnaire are similar tothose of establishments which do respond.

EMPLOYEE EARNINGS AND HOURS

Therefore, a sample is taken of the non-respondents following a procedure similar tothe one just described. Establishments in thissubsample 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 sampleestablishments supply usable data to thesurvey.

Estimating Procedure

Although a greater proportion of large thanof small establishments is included in'the sam-ple, any possible bias which might result fromthis difference is avoided by means of the esti-mating procedure. Each establishment in thesample is assigned a weight which is the re-ciprocal of the sampling ratio in the stratumfrom which it was selected. That is, an estab-lishment selected from a stratum in which asampling ratio of 1 out of 4 is used is assigneda weight of 4, so that it represents itself andthree other establishments in the stratum (seeprevious example). Data for each establish-ment are multiplied by the weight assigned tothe 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 allnonrespondents in the stratum. It is assigneda new weight the product of the originalweight and the inverse of the subsampling frac-tion. Thus, if a third were subsampled of agroup originally sampled at the rate of 1 outof 2, the weight of 6 would be assigned. In thecase of an establishment included in the sam-ple with certainty, another establishment whichis similar to the nonrespondent would beweighted to represent it.

In industry surveys, estimated employmenttotals derived from the weighting process arefurther adjusted to the employment levels forthe payroll period studied, as reported in theBureau of Labor Statistics monthly establish-ment employment series.5 This adjustment is

139

necessary to reduce the hazards of sampling,and because the State UI listings, which con-stitute the universe or a large part thereof; areprepared prior to the time of the survey andthus do not account for establishments openedor closed between the compilation of the listsand the date of the survey.

Estimated average hourly earniz,vs orweekly hours of work are the arithmetic meanof weighted individual employees' earnings orhours. Generally, they are derived by totalingweighted individual hourly earnings or weeklyhours and dividing the sum by the weightednumber of employees in the group. However, inindustries Karh as retail trade, in which hourlyearnings frequently vary by weekly hours ofwork anti in which the length of the workweekcovers a broad range among employees, themost representative group average hourlyearnings figure is considered to be the quotientof total individual weekly earnings divided bytotal individual weekly hours worked.

The following example illustrates the morecommon method of estimating group averagehourly earnings. Referring to the samplingscheme described in the preceding example, as-sume that the eight sample establishments inthe under 25 size group had 30 employees, eachearning $1.50 an hour, 20 earning $2 and 15earning $3, and that the 12 establishments inthe 25-49 size group and the 3 in the 50- andover-size group had employees with earnings asshown. Making the calculations that follow,estimated group average hourly earnings areobtained.

,Weight Xemploy-ment=

Weightedemploy-rnent X

Hourlyearnings=

Weightedgrouphourly

earnings

30 120 $1.50 $180.00

4 20 80 2.00 160.00

15 60 3.00 180.00

50 150 1.50 225.00

3 40 120 2.50 300.00

30 90 3.00 270.00

150 150. 2.00 300.00

1 100 100 2.50 250.00

50. 50 3.00 150.00

920 2,015.00

Estimatedgroup weighted group hourly earning= $2,015.00

average $2.19Total weighted group employment 920

hourlys See chapter 2, "Employment, Hours, and Earnings". earnings

137


If the average is to be computed following thesecond procedure (such as would be done inretail trade studies), employment would be re-placed by weekly hours 1..n the example and theremainder of the calculations carried through.In this case, the ratio now shown in the ex-,*ample would be replaced by the ratio of totalweekly earnings to total weekly hours.


The Bureau generally issues a bulletin on theresults of each survey of employee earningsand hours distributions. The report contains adescription and analysis of the survey resultsas well as tabulations of data collected.

in industry ,studies, tabulations are pre-sented for the entire industry, and frequentlythey are also presented for important segmentsof the industry. For example, in a survey ofmanufacturing, 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-digitSIC codes), and perhaps for some individual in-dustries (4-digit SIC codes). In addition to na-tionwide presentation, data for all or some ofthe industry segments generally are shownseparately on a regional basis, and frequentlyon a metropolitan-nonmetropolitan area basis.

In area studies, which include most industrydivisions, data obtained are tabulated for allthe industries in the area combined, and thenseparately for manufacturing and nonmanu-facturing industries. Where sufficient data areavailable, tabulations are presented for selectedindustry groups as well.

Earnings are tabulated to show the distribu-tion of employees by intervals of 5- and 10-centaverage hourly earnings. Hours are tabulatedto show the distribution of employees byweekly hours of work. The total number ofemployees, their average hourly earnings andaverage weekly hours, of course, also areshown. Data also may be presented to showaverage weekly earnings and cross-tabulationsof average hourly earnings by weekly hoursof work.

The text accompanying the tabulations gives

a description of the area or industry studiedand provides a summary and analysis of thesurvey results. Important relationships and dif-ferences among areas and industries are high-lighted and, where data are available fromearlier surveys, period-to-period changes inearnings and hours are discussed. In addition,a detailed description of the scope and methodof survey is included in the report.


The design of these employee earnings andhours distributions studies makes them par-ticularly useful in the analysis of Federalminimum wage and maximum hours legislationin analyzing the effects of legislation, in con-sidering new legislation, and in formulatingwage and hours policy. The information is usedby the executive and legislative branches ofgovernment, organized labor, business, aca-demicians, etc. Special tabulations are preparedfor the U.S. Department of Labor's Employ-ment Standards Administration, for use in theSecretary of Labor's annual report to the Con-gress required under section 4(d) of the FairLabor Standards Act. In addition, the data areused in collective bargaining, wage setting,broad economic analysis of labor area and in-dustry wage structures, comparisons of wagelevels in various parts of the country; and toshow trends in employment, earings, and hours.

Employment estimates are subject to someerror, and in smaller groups this error may berelatively large. Therefore, these estimates fre-quently are provided to serve only as generalguides to the size of the labor force includedin the survey and as reference points fromwhich to measure the direction and generalmagnitude of employment changes.

The average earnings reported are straight-time, and any premium pay for overtime orlate shift work is not reflected. Similarly, dif-ferences in prevailing supplementary compen-sation practices ("fringe benefits") amongestablishments, industries, and areas are notconsidered.

138

r

I

BIS 30106

L

EMPLOYEE EARNINGS AND HOURS 141

Report Number

Keep this copy foryour compaq

DEPARTMENT OF LABORBUREAU OF LABOR STATISTICS

WASHIIIGTON. D.C. 20212

Office of Managementand Budget No, 44,S-71001iApproval expires t2 -31 -71

Your report will beheld in confidence

The data, except for Item I which relates to theentire company, should cover the establishment(s)In the location designated to the left.

SURVEY OF INDIVIDUAL HOURS AND EARNINGSOF NONSUPERVISORY EMPLOYEES

I. ANNUAL GROSS SALES OR RECEIPTS FOR THECOMPANY, ENTERPRISE, OR INSTITUTION

(I)(Check artrpriate box.)

(3)

1Under 5250.000 to 51,000,000

5250,000 51,000,000 or moreCheck the box which indicates the annual gross volume of sales or receipts (exclusive of excise taxes) from all related activities ofthe company, enterprise or institution. including receipts from the establishment(s) covered by this report. Use the last calendar orfiscal year.

2. ESTABLISHMENT INFORMATION:

Please enter the Information requested in the columns below for each separate establishment covered by this report. Each place ofbusiness, 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 of 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; movietheatre; etc,

(c) Employment: Report all fult.time, part-time, seasonal, and casual employees who received pay for any part of the payrolliWicTrirric uding May t2, t971. Include those paid for sick leave, holidays, vacations, etc. DO NOT INCLUDE proprietors,

members of unincorporated firms, or unpaid family workers.Total Enter total number of employees. Include all classes of employeesexecutive, administrative and professional, as wellas nonsupervisory employees.

Nonsupervisra- Enter number of workers below the supervisory level. Include employees such as inside salespersons, shippingand 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, andsupervisory personnel.

(d) Collective Bargaining: Are a majority of nonsupervisory employees covered by union-management agreements? (Check appro.priate column).

Annual Gross Sales or Receipts for the Establishment: Check the column which indicates the annual gross volume of sales orreceipts (exclusive of excise taxes).

(e)

(a)

Location( address

and city)

(b)

Type of A etivitY

(e)Employment

for payroll periodincluding

May t2, t971

(d)Unlonmanagement

Averment(Are majoritycorned?)

(e)Gross establishment

sales(werelaslresessalel5250,000 or MOM?)

(f)

For,,01-ZUseOn yTotal

Non.supervisory

Yes

(I)No

(2)Yes

(I)No

(2)

2

3

4

5

6

7

8

9

(t) How many hours per day and days per week are the majority of your nonsupervisory employees scheduled to work?

Hours per day F-1 Days per week

(g) After how many hours per day would your nonsupervisory employees receive overtime premium pay?Mter, hours per day or ED we do not pay daily overtime.

Check here if you want a copy of the Bureau's report on this survey, 0

Name and title of person furnishing data(Maas Type or Print) Area code, Phone No

IEnter hours to onedecimal. e.g.. 7.5

139


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 thelocations listed in item 2(a).

Nonsupervisory Employees are defined in Item 2(c) on page I 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 -rime Ermines 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, ifovertime Is paid at time and one-half report only two-thirds of this pay).These data are requested separately for each nonsupervlsory employee. However, one entry may be made if two or more em-ployees worked Identical hours and received identical straight-time pay (see example A below). To avoid correspondence, pleasedo not report aggregate hours and earnings for several employees.Detailed instructions and examples for reporting the necessary data in each column are listed below.

'

ALL NONSUPERVISORY EMPLOYEES,COMPLETE COLUMNS (2) and (3)

Column (2) Use a separate line for each employee and enter "I," unless two ormos3 employees worked the same number of hours and received Identical hourlyor salary mtes (see example A). Please check to see that the sum of the employeesIn this column equals the number of nonsupervisory employees reported In item2(c) on page I.

column (3) Enter the number of hours paid for during the week of. May 9 -IS,1971 regardless of the length of the payroll period. Include hours paid for sickleave, holidays, vacations, etc. Count the number of overtime hours actuallyworked. (Not hours "boosted" to reflect premium pay.)

EMPLOYEES FOR WHOM STRAIGHT-TIMEHOURLY EARNINGS ARE AVAILABLE,USE COLUMN (4)

Column j4) . Enter the average straight -time hourly earnings. Ftclude premiumpayments for overtime and for weekend, holiday, and Lit: shift work. (Youshould use this column to report earnings of employees even when paid on asalary basis if average straight-time hourly earnings are available). If the hourlyearnings exclude commissions or bonuses, also complete columns 7 and 8 (seeexample D).

EMPLOYEES PAID ON A SALARY BASIS(IF HOURLY EARNINGS ARE NOTAVAILABLE),USE COLUMNS (5) and (6)

Column (.5) - Enter the straight-time salary for the salary period (weekly, bi.weekly, semimonthly or monthly) which includes May 12, 1971. Includestraight -time pay for overtime but exclude overtime premium. Do not include"draws" against commission as salary.For employees pa id a commission or bonusin 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 forwhich earnings were reported in column 5 to permit accurate calculation of aver-age straight -time hourly earnings.

EMPLOYEES RECEIVING COMMISSIONAND/OR BONUS PAY.USE COLUMNS (7) and (8)

Column 17) - Enter for each employee the total commission and/or bonus pay,including "PM's," "Slims," or any special bonuses based on sales paid quarterlyor 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 notrepresentative of normal commission earnings, a longer period may be used. Ifemployees receive both commission and bonus payments for an identical periodof time, report the combined figure (see example D). If bonus payments cover aperiod longer than the commission period, add only the prorated amount of thebonus 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 bonusperiod. (The hours should refer to the total hours worked during the period andnot necessarily only to those hours during which commissions or bonuses wereearned.) For employees paid an hourly rate or salary in addition .o commissionsor bonuses, it is also necessary to complete column 4, or columns 5 and 6 (seeexamples 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 SI68.50 (exclusive of any premium pay) for88.0 hours worked during the salary period (% month).

C. One employee. who worked 32.5 hours during the survey week, was paid on a straight commission basis and received $361.20for 168 hours worked during the commission period (I month).

D. One employee, who worked 40.0 hours during the survey week, was paid a straight-time hourly rate of SI.95 and also received$35.00 in commissions and S7.50 in "PM's" for 173.6 hours worked during the commission period (I 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 I-month period, and 51 50,00 in bonuses earned during a 3-month period. Only 1/3 ofthe bonus ($50.00) is reported to that the bonus period corresponds to the commission period.

140

f:

t'

e.

EMPLOYEE EARNINGS AND HOURS 143

3. HOURS AND EARNINGS OF NONSUPERVISORY EMPLOYEES age 3

LineNo,

(I)

Complete these columnsfor all nonsupervisory

employees

Use thiscolumn if

hourly earningsare

available

Use these columns for nonsupervisory employees paidother than on an hourly NW..

Salary Commissions and/or Bonus

Numberof

employeen

(2)

Hours paid forduring the

week ofMay 9.

15, 1971(Nearest tenth)

(3)

Strightlimehourly earnings

(Nearest cent)

(4)

Stringhtimesalary for

salary periodincluding

May 12, 1971(Neatest cent)

(5)

Houn paid forduring

salary period

(Newest tenth)

(6)

Totalcommissions

and/orbonus pay

(Neatest cent)

(7)

Hours paid forduring

commissionperiod

(.Veateet tenth)

(a)

Illustrations of examples on page 2.

A. 2 36.8 S 2.20 S

B. 1 40.0 168.50 88.0

C. 1 32.5 361.20 168.0

D. 1 40,0 1.95 42.50 173.6

E. 1 37.5 75.00 37.5 152.00 162.0

DATA FOR EACH ESTABLISHMENT SHOULD BE REPORTED SEPARATELY AND THE ESTABLISHMENT IDENTIFIED.

01

02

03

04

05

06

07

08

09

10

II12

13

14

15

16

17

IR

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

141


3. HOURS AND EARNINGS OF NONSUPERVISORY EMPLOYEES pagt. 4

l.ineNo.

(1)

Complete these columnsfor all nonsopervisory

employees

Use thiscolumn if

hourly earningsare

available

Use these columns for nonsupervisory employees paidother than on an hourly basis.

Salary Commissions and/or Bonus

Numberof

(2)

Hours paid forduring theweek of

May 12-IS, 1971

(Nearest tenth)

(3)

Straight-timehourly earnings

(Nearest serif)

(4)

Straight-timesalary f Or

salary periodincluding

May 12, 1971(;rearett rent)

(3)

Hours paid forduring

salary period

(Nearest tenth)

(h)

Totalcommissions

and/orlitmus pay

(Nearest ern()

(7)

Hours paid (orduring

commissionperiod

(Nearest tenth;

(S)

01 S S S

02

03

04 .

05

06

07

08

09

I0

11

12

13

14

15

16

17 .

18

19L.

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42....

EMPLOYEE EARNINGS AND HOURS


145

Number1. 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.

ALVIN BAUMAN

Chapter le. Union Wage Rates

Background and Description of Survey

Annual studies of union wage rates andhours are conducted in four industries : build-ing construction, local transit, local trucking,and printing.' Union wage rates and hours arethose agreed on through collective bargainingbetween employers and trade unions; they aredefined as (1) the basic (minimum) wage rates(excluding holiday, vacation, or other benefitpayments regularly made or credited to theworker each pay period) and (2) the maximumnumber of hours per week at straight-timerates. 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 unionrecords in studies of occupational wages ispracticable in industries that are characterizedby 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 de?ineated, aE4 in localtransit.

The Bureau's annual union wage studiesbegan in 1907. Originally, information was ob-tained for 39 cities, but the number was ex-panded gradually until in 1948, 82 cities werecovered.' That number was reduced to 77 in1949 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 having100,000 inhabitants or more. The scope of theinformation for individual industries has alsobeen expanded. For example, 24 journeymencrafts and nine helper and laborer classifica-tions in the building trades are covered current-ly, in place of the 13 journeymen and 7 helperand laborer classifications in the initial studies.

The study of union wage rates and hoursin the building trades includes virtually alljourneymen and helper and laborer classifica-tions. Indexes and other data are shown foreach important trade as well as for all tradescombined.'

The trucking study embraces drivers andhelpers engaged in local trucking. Over-the-road drivers and local city drivers paid on amileage or commission basis are excluded. All

data, including indexes, are presented for thetwo classifications indicated.

Union wages and hours in the local-transitindustry are limited to operating employees.Data are shown separately for operators ofsurface cars and buses, and elevated and sub-way lines, except that indexes are shown onlyfor the industry as a whole.

In the printing industry, 12 book and jobtrades, 8 newspaper trades and 6 lithographytrades are studied, and for the newspapertrades, separate data are shown for day andnightwork. Indexes and other data are pre-sented separately, by type of printing, (exceptlithography) for each trade and for all tradescombined.


The union wage studies are designed to in-clude all local unions in the covered industriesin the selected cities. Periodic checks are madewith central labor unions, district councils, andother authoritative bodies to identify new localunions that should be included in the studies.

Information is collected by mail from localunions and when necessary from internationalunions 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 werecollected directly from local union officials(generally the secretaries or business agents)by Bureau representatives and entered onforms designed specifically for this purpose.

The coverage at various times also included barbers, line-men, longshoremen, and workers engaged in breweries,laundries, metal trades, millwork, restaurants, soft-drinkproduction, theaters, and bakeries. The Bureau plans toconduct the first in a series of biennial studies of union wagerates 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 broaderStandard Metropolitan Statistical Areas which are used inmost other Bureau surveys.

oIn addition to the annual studies in the building trades, aquarterly survey of 7 major construction trades is conductedin 104 cities. Estimated average hourly wage rates for alltrades combined and for each surveyed trade are presented,together with the estimated change during the quarter andthe year.

144


Information requested relates to July 1 forall industries. This date was adopted, afternumerous changes, because most new agree-ments in these industries have been negotiatedby that time each year. In order to maintainyear-to-year comparability, wage rate, hours,and membership data for the previous yearare transcribed onto the forms before they aresent out. Union officials are requested to checkthe previous year's data and revise any figureswhich may have been incorrectly reported, andto insert current data. Copies of union agree-ments also are requested from union officialsfor the purpose of (1) checking the data en-tered on the schedules with the terms of theagreements, and (2) building up the files ofunion agreements maintained by the Bureauof Labor Statistics.4 The reporting form usedfor the building trades survey is reproduced onpages 149-151.


The current series is designed to reflect unionwage rates and hours in all cities of 100,000inhabitants or more, excluding Honolulu. Allcities of 500,000 inhabitants or more are in-cluded, as are most cities in the 250,000 to500,000 group. The cities in the 100,000 to250,000 group selected for study are distributedwidely throughout the United States. Data forsome of the cities included in the study areweighted to compenbate for cities not surveyed.To provide appropriate representation in thecombination of data, each region is consideredseparately when city weights are assigned.6

Rates

An over-all average hourly rate is computedfor each of the industries included in the unionwage studies. In addition, averages are pre-sented by industry branch, trade, city, andregion in building construction and printing;and by city and region, in local transit andlocal trucking.

Average union rates are calculated byweighting each quotation for the current yearby the reported membership.° These averages

115

are levels designed to provide comparisonsamong trades and cities at a given time, Theydo not measure the trend of union rates, thefunction served by the index series.

Indexes

Chain indexes are calculated for each of thefour industries to portray the trend of unionwage rates and weekly hours. In calculatingthese indexes, the percent change in aggregatesis computed from quotations for all identicalclassifications in the industry for 2 successiveyears. To obtain the aggregates, the rates andhours for both the previous and current yearsare weighted by the membership in the par-ticular classification for the current year. Theindex for the current year is computed bymultiplying the index for the preceding yearby the ratio of the aggregate change. In the1970 study of building trades, the rate aggre-gate for all quotations increased 11.6 percentover the previous year. The July 1, 1970, indexof union hourly wage rates for all buildingtrades (128.8) is the result of multiplying theJuly 1, 1969, index (115.4) by the ratio of theaggregates (111.6). This method of index cal-culation minimizes the influence of year-to-yearchanges in membership.

Indexes of union hourly wage rates andweekly hours are computed for each classifica-tion as well as for all classifications combinedin the building construction and printing in-dustries. In the local trucking and local transitindustries, indexes are provided only for allclassifications combined. Irregular hours ofwork for operating employees in many of thecovered cities .prevent the computation of anindex for union weekly hours in the localtransit industry.

The base period for the indexes of unionwage rates and weekly hours is the 1967 aver-age. The series for the building trades andprinting industry date back to 1907, for local

4 See chapter 20. "Collective Bargaining Agreements.""The cities in the sample were not selected on a probability

basis. Continuity of city data has taken precedence oversampling procedure during periods of contraction and expan-sion in the program.

e Reported membership, as used in this study, as defined asmembers working or immediately available for work.

c. BLS 1150.1(Rev. 1971)

r.

f

UNION WAGE RATES


1

L J

Dear Sir:

149

Budget Bureau No. 44-R738.12.Approval expires 3-31-72.

The Bureau of Labor Statistics is now conducting its annual survey of union wage scales andhours in the building trades. The continued success of these surveys, begun in 1907, dependslargely upon your cooperation and we are asking that you furnish information requested on thisschedule form. The completed form should be signed and returned to the Bureau in the enclosedself-addressed envelope which requires no postage. We would appreciate your returning thecompleted form promptly.

Thank you very much for your cooperation.

Very truly yours,

Regional Director.

INSTRUCTIONS FOR COMPLETING THIS FORM

Part A: Answer general information questions in this part of schedule.

Part B: Check and revise, if necessary, information reported for July 1, 1970. If any of yourtrade classifications have been omitted, insert them and the negotiated scales in the appropriatecolumns. Should you have more than one negotiated scale for a trade or classification, reportseparate information for each scale. For each trade or_ occupation group shown in column (2),enter in column (5) the wage scale in effect on July 1, 1971, and in column(10), the number ofweekly hours before overtime rate is effective. The number of union members at each scaleshould be entered in column (15), and the number of apprentices, as of July 1, 1971, should beentered in column (17) for each trade.

Copy of Agreement: Please provide a copy of each of your agreements to assist us in our analy-sis of collective bargaining provisions. If you have only one copy avaiblale, we shall be gladto make a duplicate and return the original promptly. If requested, the agreement will be kept inconfidence and material used only for general analysis which will not reveal the name of eitherparty to the agreement.

F.

t;


UNION SCALES OF WAGES AND HOURS IN THE BUILDING TRADES

(Annual Survey)

PART A.GENERAL INFORMATION:

Please attach a copy of your agreement in effect on July 1, 1971, and answer the following questions:

1. When did your agreement go into effect?

When may it be reopened? When does it expire?

2. What is the maximum number of hours that can be worked each day before overtime rate is effective?3. Does your agreement provide for a health and insurance plan (life insurance, hospitalization, medical, surgical, and

other similar types of health and welfare programs) financedEntirely by employer? Yes No In part by employer? Yes O No

Amount of employer contribution per Hour Shift Week Month p(cents or percent)

4. Does your agreement provide for a pension plan financedEntirely by employer? Yes No In part by employer? Yes O No

Amount of employer contribution per Hour Shift O Week Month p(cents of percent)

5. Does your agreement specifically provide for a paid vacition, financed by employer payments(A) To a vacation fund Yes No

FUND PAYMENTS DEDUCTED FROM PAYROLL UPON WRITTEN, IMPLIED, OR CONTRACTURAL AUTHOR-IZATION OF WORKERS ARE NOT CONSIDERED EMPLOYER PAYMENTS

(B) To worker each pay period, as part of negotiated scale Yes No(C) To worker each pay period, in addition to negotiated scale Yes No

(D) Other(Explain)

If answer is Yes to any of above, indicate amount of employer payment per hour.(cents or percent)

6. Does your agreement provide for employer payments to other funds, such as holiday, educational, promotional, unem-ployment benefits? Yes 0 No If YES, list below each type of contribution separately and amount of employerpayments (cents or percent).

7. What is the total membership of your union? How many are journeymen, How many are

apprentices? flow many are helpers and laborers?

8. Between July 1, 1970, and July 1, 1971, how many apprentices completed their apprenticeship?

(If none enter zero.)9. Between July 1, 1970, and July 1, 1971, how many journeymen became unavailable for work because of death, perman-

ent disability, or retirement? (If none enter zero.)

Please sign your name here Title Dace , 1971.

Address City, state, and Zip Code

147

U. S

. DE

PA

RT

ME

NT

OF

LA

BO

RB

urea

u of

Lab

or S

tatis

tics

Uni

on S

cale

s of

Wag

es a

nd H

ours

in th

e B

uild

ing

Tra

des

(Ann

ual S

urve

y)

Part

B.

Scal

e an

d m

embe

rshi

p in

form

atio

n by

trad

e:

Plea

se e

nter

the

data

req

uest

ed b

elow

. Do

not u

se c

olum

ns m

arke

d 'C

ode.

'E

mpl

oyer

pay

men

ts to

insu

ranc

e, p

ensi

on, v

acat

ion,

or

othe

r fu

nds

shou

ld b

e ex

clud

ed f

rom

the

hour

ly s

cale

s be

low

, and

sho

wn

as c

ents

or

perc

ent o

f sc

ale

in th

e sp

aces

pro

-vi

ded

in p

art A

.Pa

ymen

ts s

peci

fica

lly d

esig

nate

d as

bei

ng in

the

nego

tiate

d w

age

scal

e fo

r an

y be

nefi

t sho

uld

be e

xclu

ded

from

hou

rly

scal

es r

epor

ted

belo

w e

ven

thou

gh th

ey a

re m

ade

orcr

edite

d to

the

wor

ker

each

pay

per

iod.

Mem

bers

hip

info

rmat

ion

will

be

kept

in c

onfi

denc

e an

d us

ed o

nly

to c

ompu

te a

vera

ge w

age

rate

s.

Sche

dule

No.

Page

of

Tra

de o

r oc

cupa

tion

Wag

e ta

le fo

r ea

ch tr

ade

or o

ccup

atio

nIn

effe

ct o

n Ju

ly I

Wee

kly

hour

s be

fore

ove

rtim

e ra

teIs

effe

ctiv

e on

Jul

y 1

Num

ber

of u

nion

mem

bers

wor

king

or

imm

edia

tely

ava

ilabl

e fo

r w

ork

at e

ach

tate

on

July

I

iliM

iN

umbe

r of

app

rent

ices

Inea

ch tr

ade

on J

uly

1

(1)

Cod

e

(2)

Titl

e(3

)19

69

(4)

1970

(5)

1971

(6)

Cod

e

(7)

Cod

e.18

)

1969

(9)

1970

(10)

197)

(11)

Cod

e

(12)

Cod

e((

3)19

69

(14)

1970

(15)

1971

(16)

1970

(17)

1971

(18)

Cod

e

F.& Qt

1-a


transit to 1929, and for local trucking to 1936.Although data for the latter two industrieswere collected for year.; before the dates ofthe index series, indexes were not constructedbecause of inadequacies in the available data.


The averages and indexes mentioned to-gether with other summary data are containedin the bulletins published annually for the in-dividual studiek Included among the informa-tion shown for individual trade classificationsis the proportion of union members havinghourly rates at different levels, as well as theproportion of union members having, since theprevious study, wage rate increases of specifiedamounts in terms of cents per hour and per-cent. The increase registered by the trade isshown also.

In addition, the union rates of wages andhours in effect on the date of the survey, asreported by union officials, for both the previ-ous and current years are published for eachclassification by city. These furnish a directcomparison of union rates between the 2 yearsfor each of the industries studied. The rates ofwages are indicated as hourly rates and thehours as the weekly hours of work before over-time rates are applicable. The current studiesalso present data on employer payments forinsurance (health and welfare) and pensionpayments; in addition employer payments forvacation and other funds (except those forapprenticeship) are shown for the buildingtrades. These payments are expressed in termsof cents per hour or as percent of rate.

7 Membership (used for weighting purposes) relates only toactive members in the city and contiguous suburban areas.It does not reflect the total jurisdiction of local unions, whichmay extend beyond these limits, and it does not necessarilyreflect metropolitan area rates.


The Bureau's union-wage series provide ameans of determining intercity wage differ-ences for comparable work, and the relation-ships between rates 'applicable to workers inoccupations requiring varying degrees of skill.The data are used in wage negotiations by bothmanagement and labor. The wage rates ofbuilding-trades workers are especially impor-tant in estimating construction costs, becauselabor expenditures constitute an importantelement in the total cost of building construc-tion. The index series derived from thesestudies provide barometers of year-to-yearchanges in rates of wages and hours in theindustries covered.

Average union rates provide comparisons ofwage rates among industries, trades, and citiesat a given time. Unlike the indexes, they arenot an accurate measurement of year-to-yearchanges because of fluctuations in membershipand other factors. Membership figures for thevarious trades or classifications do not remainconstant and changes may have a marked effecton average rates. For example, if organiza-tional drives in cities having relatively lowerrates of wages result in sharp increases inmembership, the movement of the rate levelsfor the affected trades as a whole is naturallyretarded. Conversely, increases in membershipin cities having high wage rates accelerate theupward movement of averages?

The union rates, are not necessarily theactual rates paid to all workers, and the unionhours are not necessarily the hours actuallyworked. Workers with above average exper-ience and skill may be employed at rates abovethe union wage rates, especially during pros-perous times when a tight job market createscompetitive bidding for the better workmen.During periods of depressed business activity,actual hours worked often are less than hoursspecified in the union agreement.

-THOMAS C. MOBLEY

119

Chapter 17. Current Wage Developments

Background

Since January 1948, the Bureau of LaborStatistics has issued a monthly report listinggeneral wage changes and changes in supple-mentary benefits agreed to in selected collectivebargaining situations, identifying the situa-tions by company and union name. The scopeof the listing has varied somewhat from timeto time, but since 1953, it has been limited toagreements affecting approximately 1,000 ormore production and related workers in manu-facturing and selected nonmanufacturing in-dustries.

Current Wage Developments was initiatedbecause of the rapid increase in wage rates andprices in the early post World War II period,the interest in determining the extent to whichsettlement patterns spread from industry toindustry; and the discontinuance of an indexof wage rates that had been initiated duringWorld War II. Interest in the listing was stim-ulated by the Korean emergency when theWage Stabilization Board needed data on theextent to which wages and benefits were beingchanged.

In 1949, and again in 1951 and 1952, sta-tistical summaries of wage changes were pre-pared to supplement the listing, but regularpreparation of a statistical summary began in1954. These quarterly statistical summariesshow the distribution of settlements and (since1955) of v+orkers by the size of the generalwage changes agreed to and the frequency withwhich various types of supplementary benefitswere 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, providesa much more detailed account of negotiated wage and benefitchanges than can be presented in a tabular summary. Whenavailable, information on changes for large groups of non-union workers, including professional, white-collar, and pro-duction employees, also is presented.

'Only changes in benefits that represent changes in costsare included.

Prior to 1988, the construction service trades and financeindustries are also excluded.

4 General wage changes are defined as changes affecting atleast one-tenth of the workers at any one time or all workersIn an occupation. Changes resulting from promotions, meritincreases, etc., are excluded.

facturing, but includes information on generalwage changes for nonunion and small unionsituations, as well as for large collective bar-gaining situations.' From 1959 through 1970,this summary also included information onchanges in supplementary benefits.

Description of Series

The summary of major collective bargainingsituations hereafter is referred to as the"major" series, and the summary that is basedon changes in wages in manufacturing firmsof all kinds is described as the "manufactur-ing" series.

The major series describes general wagechanges and changes in benefits 2 in all collec-tive bargaining settlements involving 1,000 ormore production and related workers in manu-facturing and 1,000 or more nonsupervisoryworkers in the nonmanufacturing sector, ex-cluding Governments

Supervisory or professional employees areexcluded. 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 areincluded if the agreement as a whole covers1,000 or more workers even though each indi-vidual plant employs fewer. Also included arecontracts with trade associations or withgroups of firms that bargain jointly with aunion or unions, even though the firms are notassociated formally and each has fewer thanthe minimum number of workers covered bythe series. Situations in which two or moreunions, together representing more than 1,000workers but individually accounting for fewer,negotiate essentially identical contracts withone firm or a group of firms, are tabulated asone bargaining unit.

The summary for manufacturing as a wholerepresents all establishments with four or moreemployees that adjust wages by means of gen-eral wage changes,' regardless of whether theworkers are represented by a union.

Wage change data are presented in cents per

150 153


hour and, since 1959, as a percentage of aver-age straight-time hourly earnings, adjusted toexclude premium pay for overtime work.

Two types of information are presented onwage changes: (1) information on changesthat go into effect within 12 months of the dateon which they are decided or negotiated, thusreflecting the economic climate at the time thechanges are decided upon ; and (2) all changeseffective during the period being summarizedwhether or not they were (a) decided uponduring that period, (b) were deferredthatis, decided upon earlieror (c) resulted fromoperation of cost-of-living clauses. In distribu-tions of workers by size of wage change, allworkers in an establishment or collective bar-gaining situation are distributed according tothe average wage increase in the establishmentor situation. The number of workers affectedby changes in supplementary benefits includesall production and related workers in the situa-tions where the benefit is changed, whether ornot all are affected immediately. For example,if a fourth week of vacation is added for work-ers with 20 years' service in an establishmentemploying 1,000 workers, a vacation changewould be recorded for 1,000 workers, eventhough only a relatively small proportion wouldbenefit from the change immediately.


The statistical summary of the major seriesis compiled from the summaries of collectivebargaining settlements presented in themonthly Current Wage Developments listingwhich, in turn, is derived primarily from sec-ondary sources, including general circulationnewspapers and periodicals, as well as union,management, and trade publications. Otherimportant sources of information are the fileof union contracts maintained by the BLS andthe U.S. Department of Labor's files of pensionand health and welfare agreements, maintainedby the Office of Labor-Management and Wel-fare-Pension Reports.° By the end of the year,the BLS contacts, almost entirely by mail,either management or labor representatives inany situation for which these other sources

have not yielded information on wage andbenefit changes during the year.

Information for nonunion and small union-ized firms is gathered quarterly (semiannuallyin 1965 and 1966) by, a questionnaire mailed toparticipating establishments. The informationon general wage changes is supplemented bythe contract file (unionized establishments)and from newspaper clippings, purchased froma commercial clipping service. At the end of theyear, BLS field representatives contact, pri-marily by telephone, a sample of firms that havefailed to respond to the mail questionnaire orthat have provided incomplete or unclear infor-mation.

.444mpling and Estimating Procedures

As indicated earlier, all bargaining situa-tions with 1,000 or more workers in manu-facturing and nonmanufacturing industries areincluded in the major series. It is believed thatthe current list of nearly 2,500 such situations,built up since Current Wage Developments wasstarted in 1948, is very nearly complete. Aftera bargaining situation is added to the universe,it is withdrawn only if it ceases to be within thescope of the survey (e.g., because of a changein business to one outside the scope of thesurvey, a change to nonunion from union, orbecause of an apparently permanent drop inemployment to substantially below 1,000).

The sample for manufacturing is derivedfrom State unemployment insurance listings(UI) which show reporting units with four ormore employees by location, number of em-ployees, and industry classification.° The sam-ple is a highly stratified probability design,with sampling ratios varying from 1 out of150 establishments with 4 to 19 employees toall of those with 1,000,or more employees.7

Information from contracts supplied on a confidential basisis used only in the statistical summaries, not for the monthlylisting.

a See appendix B, "Industrial Classification?' For a moredetailed description of Unemployment insurance data, see p.15, chapter 2.

In the ;case of a i:dvi companies with large numbers ofestablishments each with 'IMO workers or more, a sample ofplants is chosen.

Pr

r

f.

BLS 2675d

P

CURRENT WAGE DEVELOPMENTS



Wage Developments in Manufacturing, 1971

(Change If incorrect, include ZIP - code.)

155

Office of Management andBudget No. 44-R1135Approval expires 3/31/72

The Bureau of Labor Statisticswill hold all information furn-ished by the respondent instrict confidence.

Identification or location of establishmentfor which information is requested, if

.11. different from mailing address.

1. What was the major product (in terms of sales value) of this plant during 1970?

II. Please provide employment and payroll information for the payroll period including January 12, 1971.

A. All employees1. Number

B. All production and related workers2. Payroll 3. Man-hours1. Number

L

See page 4 forexplanation.

III. Do collective bargaining agreements cover a majority of your production and related workers?

If "No," please skip to section VII.If "Yes," please answer all questions except X.

Yes No 1::]

FOR UNION FIRMS ONLY

IV. Union and Agreement Identification:

A. With what union or unions do you have a collective bargaining agreement?

B. Are you a party to an agreement signed by an employer association with this union(s)? Yes 0 NoIf so, what is the association's name?

V. Agreement Expiration Date:

When does your collective bargaining agreement(s) expire or become subject to reopening on wages?

(Space has been provided for two entries since the date may change during the year.)

VI. New or Revised Agreement: Please mark appropriate boxes below to bring your report up to date for

During the quarter Jan. Mar. Apr.June JulySept. Oct.Dec.1971 1971 1971 1971

A. Did you negotiate a new or revisedcollective bargaining contract(s) for yourproduction and related workers?

B. Did you agree on an immediate ordeferred change in wages?

Yes No

ElYes No Yes No Yes No

152

PLEASE TURN TO NEXT PAGE.

156 BLS" HANDBOOK OF METHODS

2FOR UNION AND NONUNION FIRMS

VII. Wage-Rate Changes for Production and Related Workers, 1971:Please report any general wage-rate changes you have put into effect for your production and related workers in the previousqututer. (Food-processing companies should also report wage-rate changes affecting driver-salesmen.)

Include:

(1) AU changes affecting either (a) 10 percent or more of yourproduction and related workers at any one time, or (b) allworkers covered by a single agreement, even if the agree-ment applies to fewer than 10 percent of the workers.

(2) Any change in your pay scales even though no workers re-ceived imrilediate pay increases as a result of this change.

(3) Any cost-of-living escalator adjustments whether or notthey are part of your permanent rate structure.

(4) Increases decided on in earlier years but going into effectin 1971.

(5) Changes in hourly rates resulting from changes in hourswithout corresponding changes in weekly or daily pay.

A. Have you put into effect any such general wage changes during the quarter?

Exclude:

(1) Increases to individuals resulting from promotions,automatic increases with length of service, orprogression within an established rate range.

(2) Increases decided on in 1971 but scheduled to gointo effect in later years. (These should be listedin section VIII.)

(3) The cost of any changes in supplementarybenefits.

Jan.-Mar. Apr.-June July-Sept. Oct.-Dec.1971 1971 1971 1971

Yes

No

B. If your answer is "No", proceed to section VIII. If your answer is "Yes", indicate below the form of the wage change(s)and list in subsection C:

(1) Uniform cents per hour

(2) Uniform percentage change

(3) High:: cents per hour for skilled workers

(4) Other (specify in section XII. "Remarks")

C. If all workers did not receive the same amount (either the same number of cents or he same percentage) list changes foreach group on a separate line with the approximate number affected. For example, if there was a unifOrm across-the-board change plus added changes for some workers, list the uniform change first and show additional changes below. If acost-of-living escalator adjustment went into effect at the same time as another increase, list it separately.

In reporting information for incentive workers include, if possible, estimated effects of wage-rate changes on incentiveworkers' earnings, (Fr example, if base rates for incentive workers were raised 5 cents and this increased their hourlyearnings about 7 cents, report 7 cents.)

If any changes in scales were made that did not affect any workers immediately indicate the approximate number to be

Indicate whether change was given in percentage or cents terms.See page 4 for example ofreporting general wage changes.

Effec-tivedate

A pproxi-mate

numberreceiving

wageadjust-ments

Classes of production and relatedworkers or lobs affected

Hourlychange_

forworkers(:. o -)

Check if this was:

Deferred(effectivein 1971 but

decidedearlier)

Auto-merit

cost-of-living

escalator

t %

a %

4 %

4 %

C %

X53

1.

CURRENT WAGE DEVELOPMENTS

VIII. Future Wage-Rate Changes:

157

3

List any general wage-rate changes already decided upon for your production and related workers but scheduled to gointo effect in the future.

Effective date(month, date, and year)

Workers affected Amount of change forworkers whose rates

will be changedEstimatedNumber Classes or sobs

0

4

0 %

0 %

IX. Cost-of-Living Escalation:

Do you have a cost-of-living escalator policy whereby wages are changed automatically with specified changes in aprice index? Yes E] No El If so, please list the months in which wage changes will go into effect if the priceindex warrants

FOR NONUNION FIRMS ONLY

X. Policy Regarding Wage Changes:

Whether or not you changed wages during the first quarter of 1971, what is your policy regarding general wage changes?

1. Wages are normally changed only on an individual basis. El2. General wage changes are sometimes made. ED

XI. Do you want a copy of the Bureau's summary on this survey?

XII. Remarks:

Yes E] No ED

. Name and title of person furnishing data(Please Type or Print) Area code, Phone No.

c.

14

154


Explanations for Report on Wage Developmentsin Manufacturing, 1971

Section 11-A All employeestotal number on the payroll of the plant covered by this report who worked full-time or part-time orreceived pay for any part of the period reported. Include persons on paid vacations and sick leave. Exclude persons on leavewithout company pay the entire period as well as pensioners and members of the Armed Forces not working during the periodreported.

Section 11-B Production and related workers. Include working foremen and all nonsupervisory workers engaged in fabricating,processing, assembling, inspecting, receiving, storing, handling, packaging, warehousing, shipping, trucking, hauling, mainte-nance, repair, janitorial, watchmen services, product development, auxiliary production for plant's own use (e.g., power-plant),recordkeeping, and other services closely associated with these production operations.

Exclude employees engaged in executive, purchasing, finance, accounting, personnel, cafeteria, professional, and technicalactivities; sales, advertising; collection; installation, and servicing of products; route office functions; fattory supervisionabove theworking foremen level; and employees on yout payroll engaged in construction of majot.additions or alterations tothe plant.

NuMbor. Include both full-time and part-time production and related workers on your payrollwhether wage or salariedwhoworked during or received pay for any part of the payroll period reported. Include persons on paid sick leave, paid holidays,and paid vacations.

Payroll. Include pay earned during the payroll period by production and related workers reported in the preceding box.Payroll should be reported before deduction for old-age and unemployment insurance, group insurance, withholding tax, bonds,and union dues. Include cost-of-living allowances, pay for overtime holidays, vacations, and sick leave.

4,

Mon-hours. Include all hours worked, not scheduled hours, during the payroll period by the production and related workersreported in the first box plus hours paid for stand-by or reporting time and holidays, and man-hours equivalent to pay receivedby employee directly from your firm for sick leave and for holidays and vacations for this payroll period.

Section VIIC Each period report the general wage-rate changes you have put into effect for your production and related workerssince your previous report to us. (Note: Establishments in the food-processing industries should also report wage-rate changesfor driver-salesmen.) Entries of various types of wage changes can be illustrated as follows:

Effee-tivedate

Approxi-mate

numberreceiving

wageadjust-ments

Classes of production and relatedworkers or jobs affected

.

Hourly change forworkers(+ or -)

Check if this was:

Deferred(effectivein 1971

butdecidedearlier)

Auto-ma tic

cost-of-living

escalator

1/5 1,000 All + 8 4 %

1/5 950 All except common labor Avg. 3 4 + ii to 5 %

2/10 1,000 AR + 2 4 %

6/15 1,000 All4 8Avitg. 6

6/15 100 Maintenance + 9 add. it

8/5 400 Time + 6 4 % ..

8/5 600 Incentive + 7 4

10/1 1,000 All - 1 4 % X

12/5 125 Powerplant and maintenance + 10 it

CURRENT WAGE DEVELOPMENTS 159

The ratios are uniform for all industries. Sincedata are available from secondary sources forall unionized situations with at least 1,000production and related workers, data for allestablishments meeting this criteria also areincluded in the summary for manufacturing.The sample selected from the UI listings iscompared with this list of establishments forwhich information already is available; sincedata for these sample members are obtainedfrom secondary sources, these establishmentsare not sent questionnaires. Approximately6,000 establishments are left for the question-naire survey.

Although the sampling design yields asample in which large firms are relatively over-represented, this bias is overcome by the esti-mating procedure. Each establishment in thesample is assigned a weight which is the re-ciprocal of the sampling ratio in the stratumfrom which it was selected. An establishmentselected from a stratum from which 1 out of 4establishments is chosen is assigned a weightof 4, so that it represents itself and three otherestablishments. Information for each establish-ment is multiplied by the weight assigned to theestablishment. Thus, all establishments , re-gardless of size, are represented appropriatelyin the final estimates.

An establishment in the subsample of non-respondents followed up by a visit is weightedto represent all nonrespondents in the stratum.It is assigned a new weightthe product ofthe original weight and the inverse of thesubsampling fraction. Thus, 1 out of 3 non-respondent establishments subsampled from agroup originally sampled at the rate of 1 out of2 would be assigned a weight of 6. If an estab-lishment included in the sample with certaintyfails to respond, another similar establishmentwould be weighted to represent it.

To the estimates derived from the weightingof the sample questionnaire are added the datafrom secondary sourcesthe numbers of majorbargaining situations.

The totals thus obtained are further adjustedto reduce the hazards of sampling and to takeaccount of opening or closing of establishments

See chapter 2.

between compilation of the State unemploy-ment insurance listing from which the sampleis 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 industrygroups, as reported in the monthly employmentseries of the Bureau of Labor Statistics for theperiod covered by the summary of wagechanges." For example, if the estimate of prod-uction worker employment in an industrygroup derived from the sample is 100,000 butthe Bureau's estimate of employment in thatindustry group was 110,000 workers, each em-

110,000ployment count would be multiplied by

100,000or 1.1 The resulting industry group estimateswould be.added to provide the estimates for allmanufacturing.

The major series for manufacturing andnonmanufacturing combined is not adjusted inthis fashion, since it is presumed to be allinclusive.

A new sample of nonunion and small union-ized plants in manufacturing usually is selectedevery 3 years. Establishments with fewer thanfour workers are omitted because in manyStates they are not covered by these unemploy-ment compensation programs. After the initialcontact, establishments of any size that indi-cate that they have a policy of adjusting wageson an individual basis, rather than by meansof general wage changes, are omitted fromfurther survey.

Presentation and Analysis

Preliminary information on the "packagecost" and general wage changes resulting fromcollective bargaining settlements involving themajor situations is issued in press releasesabout 3 weeks after every quarter and the in-formation is also summarized in Current WageDevelopments (See Chapter 23 for a descrip-tion of the package cost program). Final infor-mation on wage and benefit changes is notavailable until the end of the year. It is pre-sented in Current Wage Developments and theMonthly Labor Review. Yearend summariesalso present information on total effective

156


wage changesthose negotiated in the periodplus deferred and cost-of-living changes re-sulting from settlements in prior years butgoing into effect in the current year. Informa-tion is presented for manufacturing industries,for nonmanufacturing industries, and for bothcombined.

Quarterly summaries of the manufacturingdata are published in Current Wage Develop-ments and the article covering the full year alsoappears in the Monthly Labor Review. The3-, 6-, and 9-month summaries, which are basedon preliminary data, stress wage changes re-sulting from settlements or management de-cisions made during the period, while theyearend article, which is based on final data,also analyzes trends in the size, frequency andtype of wage changes, and the prevalence andresults of wage escalation policies. Because itis based on data for both large and small union-ized and nonunionized establishments, the

manufacturing analysis can make many otheruseful comparisons of its components.


The data are used extensively by labor, man-agement, and the Federal Mediation and Con-ciliation Service in collective bargaining; byprivate institutions and universities in studiesof industries or groups of industries; and bylocal and Federal Government agencies in-terested in the current economic picture todetermine trends in wage and benefit changesas well as for wage, income, and gross nationalproduct forecasts.

Since the sample is relatively small, data arenot presented for individual industries.

11.57

-GEORGE RUBEN

Chapter 18. Employer Expenditures for Employee Compensation

Background

The measurement of employer expendituresfor employee compensation and the compositionof payroll hours was undertaken by the Bureauof Labor Statistics (BLS) to fill a large gapin the statistics of employee compensation andhours paid for.

Prior to World War II, compensation forAmerican labor consisted mainly of wages andsalaries 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 yearsof the Second World War, employers were en-couraged by the policies of the War LaborBoard to grant wage supplements instead ofwage increases, e.g., vacations, hospitalization.Shortly after the war, the NLRB ruled thatpension plans were within the purview of col-lectively bargained agreements.' Expendituresfor these and other compensation elements, inaddition to pay for working time, began tocomprise a substantial portion of the total com-pensation of labor.

As early as 1875 the American Express Com-pany instituted a private pension plan.2 In1929, a private study 2 indicated that therewere almost 400 such plans, and by 1968 therewere about 34,100 pension plans and 157,700private pension or welfare plaam in America.4

Paid vacations and holidays also have a rela-tively short history for most workers, Paidvacations were fairly well established forsalaried workers by the middle of the nine-teenth century. Industrial workers, however,first started to receive paid vacations aroundthe turn of the century, and not until afterWorld War I did the principle of paid vacationsbegin to assume importance in the developmentof labor policy; paid holidays generally werenot found in industry until World War II, al-though it had been customary for salariedworkers to receive pay for time not worked ondesignated holidays. By 1968, approximately 7percent of the production worker hours paidfor in manufacturing industries were leavehours, almost all of which were vacation andholiday hours.

The importance of recent changes in thestructure of compensation may be illustratedby examining those that occurred for manu-facturing production workers between 1959and I968. During that period pay for workingtime increased from $2.23 to $3.02 an hour orabout 35 percent. At the same time, employeeexpenditures for all other elements of compen-sation increased front 38 cents to 69 cents anhour or about 76 percent. Thus, the relativeimportance of pay for working time decreasedfrom 85.4 percent of total compensation in1959 to 81.8 percent in 1968.

The Bureau has for many years recognizedthe necessity of studying outlays for employeecompensation. Early attempts, however, werelimited to exploratory work on methodologyand the availability of data.5 By 1959, manyof the technical and conceptual problems hadbeen sufficiently resolved to permit the initia-tion of a regular program.

The first survey in the program, 1959 ex-penditures in manufacturing, was followed bya 1960 mining study ; a 1961 finance, insurance,and real estate survey, and another manufac-turing industry study in 1962. The 1963 studyof expenditures for salaried (white-collar)workers, which covered most nonagriculturalindustries in the private sector, represented thefirst shift in program emphasis from an indus-try to an economy-wide orientation. Since then,the program has been redesigned to cover allemployees in the private nonfarm sector andto cover all significant items of employee com-pensation.

Inland Steel vs. National Labor Relations Board, 170 Fed-eral Reports, Second Series 247 (1948), 251 (1949).

This was the first recorded private pension plan inAmerica.

;1Latimer, Murray Webb. Industrial Pension Systems in theUnited States and Canada, Industrial Relations Counselors,Inc., New York, 1932.

4 Total active plans covering 28 employees or more forwhich descriptions had been filed with the U.S. Departmentof Labor by April 1, 1968 under the provisions of the Welfareand Pension Plans Disclosure Act, as amended (P.L. 85-838 asamended by P.L. 87-420),

3 Problems in Measurement of Expenditures on SelectedItems of Supplementary Employee Remuneration, Manu-facturing Establishments, 1953 (BLS Bulletin 1188, 1958):Wage Structure: Basic Iron and Steel, January 1951 (BLSSeries 2, No. 81, 1951); Wage Structure: Electric and GasUtilities, September 1957 (BLS Report 135, 1958).

158161



The survey relates to employee compensationpractices, employer expenditures arising fromthese practices, and to all hours for which pay-ment is madehours worked, paid hours ofvacation, holiday, sick leave, and civic andpersonal leave.

The program is designed to provide databiennially for the entire private nonfarm sec-tor; separate information is given for manu-facturing and nonmanufacturing industries. Inthe intervening years a number of individualmanufacturing and nonmanufacturing indus-tries are studied. Both types of studies providedata for all employees and for office and non-office employees separately. Survey coverageextends to the 50 States and the District ofColumbia and separate data are provided forbroad economic regions.

The data relate to cash disbursements ofemployers and the hours-paid-for during acalendar year. Data for a lesser period of timedo not completely reflect the outlays made byemployers and the hours-paid-for. Paid leavetime, for example, usually is spread unequallythroughout the calendar year; similarly, ex-penditures for many of the legally requiredinsurance programs stop after a specifiedmaximum amount is earned by each workerduring the year. These practices result in widevariations between expenditures in the earlypart of the year and in the latter part of theyear.

Employee compensation is increasing contin-ually in complexity and magnitude. Practicesdiffer by industry group and new supplemen-tary practices are being added. Each survey isdesigned to show the individual characteristicsof compensation practices in particular indus-tries without impairing comparability withothers. This objective is accomplished by re-taining the same basic reference framework.

Expenditures

The expenditures studied are considered toconstitute the major elements of employeecompensation in American industry. The ex-penditures, and therefore their measurement,

159

fall into two broad groups determined by theway they are paid : payments made directly tothe workers and payments made to a thirdparty on behalf of the workers.

Direct payments include pay for workinghours; pay for hours not worked such as va-cations, holidays, sick leave, and civic andpersonal leave; premium pay for overtime,weekend, holiday, and shift work; nonproduc-tion bonuses; and severance pay. The total ofthese payments constitutes the gross payroll.

Indirect payments are those made by theemployer on behalf of the worker to funds,trustees, insurance companies or Governmentagencies which may make a payment to theworker at a later date or provide full or partialeconomic security against a future contingency(i.e. unemployment, retirement, medical ex-penses etc.). The programs from which theseexpenditures arise are either legally requiredor voluntary. Legally required programs in-clude social security, unemployment insurance,workmen's compensation, and State temporarydisability insurance. Voluntary plans studiedare life, accident, and health insurance; retire-ment plans ; vacation and holiday funds; rev-erence and supplemental unemployment benefitfunds; and savings and thrift plans.

Payroll Hours

The payroll hours studied are all hours forwhich the workers receive pay. These hoursconsist of plant or working hours, and vaca-tion, holiday, sick and other hours of paidleave. Although an hour normally is defined as60 minutes of elapsed time, a payroll hour doesnot necessarily consist of 60 minutes. For ex-ample, hours worked on a day that would other-wise have been a paid holiday are paid fortwiceonce as a paid holiday, and once asworking hours. Therefore, an 8-hour holidayworked for which 16 hours of payment wasmade is counted as 16 hourshalf of whichare holiday hours and half are plant-hours.Conversely, some hours of leave are paid for atless than the regular rate and only the equiva-lent hours are counted. Each overtime hourworked at premium rate is counted as 1 plant-hour.

EMPLOYER EXPENDITURES FOR EMPLOYEE COMPENSATION 163

Establishment Policies

Although the data on company policies areused primarily in the review and analysis ofthe expenditure and hours data collected in thesurvey, these policies also have significance intheir own right, and often signal changes insupplementary practices before actual expendi-tures are incurred. In addition, certain charac-teristics of American industry are measured ona national basis and the relationships amongthese characteristics studied. Examples are theactual distribution of workers by amount ofvacation and the degrees of unionization.


The data are obtained from establishmentrecords. Generally, no single record is sufficientand several record sources must be summarizedto arrive at annual totals. The data are enteredby the employer on preprinted forms in accord-ance with detailed instructions.

Not all companies keep records in the detailrequested and approximations in these casesmay be accepted. In general, two types of ap-proximations are used. First, if the establish-ment records are kept for a broader groupingof employees than are being studied, the pro-rated share for the workers included in thesurvey 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 keptbut the practice is observed. For example, theexpenditures for holiday pay may be approx-imated by multiplying the number of hourspaid for holiday leave by average straight-timehourly earnings. Errors occurring from the useof these approximations would have to be inthe same direction in substantially all the cases(overstatement or understatement of the actualvalues) 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 thelarge employers and to a sample of the estab-lishments that have not responded to a secondmailing of the questionnaire. A questionnaireform used in the expenditure study is repro-duced on pages 164-167.

Sampling Procedure

The surveys are conducted on the basis of ahighly stratified probability sample of estab-lishments selected by industry, location, andemployment siz.. The samples generally aredesigned to yield reliable data for an ,industrydivision at the national levels, in four broadeconomic regions, and for major industrygroups.

The lists of establishments from which thesamples are selected are those maintained bythe State agencies administering the employ-ment insurance laws. These lists show the em-ployment, industry classification, and locationof all establishments covered by those laws ineach State. Since some States do not coverestablishments with fewer than four employeesunder the unemployment insurance (UI) law,the samples exclude establishments in that sizegroup. (See method of estimation for treat-ment of the employment in such establish-ments.) Some establishments in particularindustries are exempted from the UI laws eventhough they employ more than four workers.The data used in sampling these establishmentsare obtained from lists compiled by regulatoryGovernment agencies, trade associations, andother sources.

Within each industry, the sample is selectedto yield the most accurate estimates possiblewith the resources availablethe principle ofoptimum allocation. This is done by includingin the sample a greater proportion of largeestablishments than of small. In general, anestablishment's chance of selection is roughlyproportionate to its employment size.

A subsample of establishments failing to re-ply to the mail inquires is selected to representall nonrespondents, following the same generalplan as is used in the original sample. Estab-lishments in this subsample are visited per-sonally, instead of being solicited again bymail.


Data for each sample establishment areweighted in accordance with the probability ofselection of that establishment. In the individ-

160

is


BIS 2868(Rev. '68)

U.S. DEPARTMENT OF LABORBUREAU OF LABOR STATISTIC!WASHNOTON, D.C. 20212

Budget Bureau Nc. 44-R1300Approval expires 10-31-71

j Location of unit for which data are requested.

Expenditures for Employee Compensation, 1970

Gentlemen:The Bureau of Labor Statistics is conducting an important survey of how muchcompanies spend for employee compensationfor wages and salaries and for legallyrequired and private programs which provide for employees' health and welfare.The Government has used such information from earlier surveys to aid in formu-lating economic policy and asks that you assist it by completing this form.Information from the survey will be of value to your company also, since it willenable you to compare your expenditures for employee compensation with those madeby industry in general. Keep a copy of your reportan extra form is enclosed.When the Bureau's report on the survey is issued, we will send you a copy andshow you how to compare data for your company with the national averages.Your report will be held in confidence and will be seen only by sworn employeesof the Bureau. Nothing will be released relating to individual companies.Please complete the form within 3 weeks and return it in the enclosed envelope.If you have questions about the information requested, or if you need assistancein completing the form, phone the Bureau collect at area code Z02,-L961-4019 or961-4030. Thank you for your cooperation.

Si ercly yours,

22°1,ZLGeoff y oreCommissioner

1. Cowpony official to contact If there are gnestIonn shout this report:

Name and title (Please print or type) Area code, phone no.

D. Unita covered by this repent

Is this questiornalre being completed for the unit(s) designated above?

0 Yes What Is the principal product, service, or activity of this unit?

0 No our records make it impossible to repot separately for the unit designated above. Units in addition to the onefor which data are requested are included in this report.

Please complete hem'1,11 at end of questionnaire to describe the units covered.

III. Average 1970 employment in milts covered by this report:

Please enter the average number of employees In each category during 1970. Include full- and part-time employees.Types of employees in each category are described below.

A. Office employees

B. Nonoffice employees

C. Total 1970 average employment

OFFICE EMPLOYEESInclude all employees in executive, administrative, and management positions, above the workingsupervisor level. Also include supervisory and nonsupervisory professional employees and their technical assidants; employeesengaged in office clerical operations; and all salespeisons whose sales activities are primarily performed outside of theertablishment (e.g., real estate salesmen, end door-to-door salesmen).

NONOFFICE EMPLOYEESInclude all employees, except office employees as defined above, in nonsupervisory, nonprofessionalpositions. Include employees engaged in fabricating, processing, or assembling; building or excavating; mining, drilling, orpumping; maintaining or repairing; shipping, receiving, handling, warehousing, packing, or trucking; retail sales; operatingor working on moving vehicles (buses, boats, etc.); Janitorial work; guard or watchman work; and similar activities.

TOTAL EMPLOYMENT Is the sum of office plus noragfropployees. Proprietor, members of unincorporated firms, andunpaid family workers are not considered to be 9mplat_ek.Vdre excluded from the survey.

EMPLOYER EXPENDITURES FOR EMPLOYEE COMPENSATION

2Instructions for Specific Items

Item IV.

A. Gross payrolltotal of wages, sala-ries, and other payments made during 1970before ani deductions. The amount shouldequal wages reported on Internal RevenueService Forms W-2 as subject to Federalwithholding taxes, or total remunerationreported on IRS Form 940, Employer'sAnnual Federal Unemployment Tax Return1971,, Schedule B, Item 1.

A-2. Pay for overtime and weekend andholiday workfor overtime, report thestraight-time pay for work beyond thenormal workweek in A-2a and premiumpay in A-2b. For example: If overtimeis paid at time and one-half, report two-thirds of this pay in A-2a and one-thirdin A-2b.

For work on weekends, holidays, orduring vacation periodsreport the regu-lar pay for work in A-2a, and the premiumpay for work on weekends, holidays, orduring vacation periods in A-2b. Paymentsmade in lieu of time off for holidays orvacations should be excluded from A-2 andincluded in A-4. For example: An em-ployee worked on a holiday. lie receivedhis iegular pay for working; plus one-halfhis regular pay as a premium for havingto work on a holiday; plus his regular holiday pay. Report his regular pay for work-ing in A-2a; the premium in A-2b; andthe regular holiday pay in A-4b.

A-3. Shift differentialstotal expendituresfor pay above regular day-shift rates forwork on late shifts. Include pay for hoursnot worked. For example, if late-shiftemployees work 71/2 hours a day but re-ceive pay for 8 hours report the total ofthe one-half hour payments.

A-4. Pay for leave--only payments thatwere part of the establishment's grosspayroll. Exclude payment s to funds,trustees, etc. (Report these in C-3.) Ifan employee received both "pay in lieu oftime off" and "pay for work" include only"pay in lieu of time off" here.

A-b. Nonproduction bonusestotal amountpaid for nonproduction bonuses includinglump-sum payments under profit-sharingplans, and other irregular or seasonalbonuses (such as attendance, Christmas,or yearend bonuses). Proceeds of profit-sharing plans which are paid into savingsand thrift funds or retirement plans shouldbe reported as expenditures for a privatewelfare plan (IV-C).

A-6. Severance paytotal of all payrnen:smade by the establishment to employeesbecause of temporary or permanent sev-erance of employment. Exclude paymentsto funds, and to pensioners under the pro-visions of pay-as-you-go pension plans.Report these in IV-C.

13, Legally required insurancenetliability incurred during 1970 under theprovisions of State and Federal laws foreach program. Exclude payments made in1970 for 1%9, but include payments madein 1971 for 1970. Include payments to gov-ernment agencies, insurance companies,and to employees under self-insured plans.Exclude payments made by or withheldfrom employees.Required Payments

Social Security; In 1970 the employer's payment was 4.8percent of the first 57.800 paid each employee, or a maxi-mum of $374.40 per employee. See Form 94IEmployer's,Quarterly Federal Tax Return which you filed In April, July,and October 1970 and January 1971. Report the sum of one-half the FICA Tax reported on line 5 for the 4 quarters.

Unemployment insurance--Federal: In 1970 the employer'spayment was 0,5 percent of the first $3,000 paid each em-ployee or a maximum of 515.00 per employee. See IRS Form940 for 1970, Schedule A, Item 16. Railroads Include totalpayments under Railroad Unemployment Insurance Act.State: In most States the payment was at varying rates onthe first $3,000 paid each employee. See IRS Form 940 for1970. Schedule A, Column 9.

C. Private welfare plansnet pay-ments (after deduction of refunds, rebates,and dividends) made during 1970 by theestablishment to funds (including union-management funds), trustees, insurancecompanies, and payments made under theprovisions of self-insured plans to employ-ees or their beneficiaries. Include pay-ments for current employees, employeeson layoff, retired employees and their de-pendents. Exclude employee contributionsand all administrative costs incurred bythe establishment. Also exclude paymentsmade by funds, trustees, and insurancecarriers to your employees or their ben-eficiaries.Life, accident, and health insurance: Life, accidental deathand dismemberment, sickness and accident, wage and salarycontinuance insurance, and death benefits; and hospitalization,surgical, medical, dental, optical, and drug plans. Excludeexpenditures for in-plant medical care and visiting nurses orphysicians.

Pension and retirement plans: Direct payments to pensionersunder a pay-as-you-go pension plan, payments under profit-sharing plans deferred until retirement, and payments for pastand current liabilities under funded plans.

Item V.

Total number of hours paid forall hoursworked at straight time, all hours workedat overtime, all paid leave hours, andhours equivalent to payments made by theestablishment directly to the workers forother hours not worked but paid for.B. Number of overtime hours all hoursfor work beyond the normal workday orworkweek for which straight-time or pre-mium payments were made' during 1970.C. Number of leave hoursreport hoursequivalent to the payments made and notthe actual time taken off. For example:3 hours paid for at two-thirds the regularrate should be reported as 2 hours. In-clude leave hours for which payment wasmade instead of time off.

162

165


3

General Instructions

If your records for an item combine data for office and nonoffice employees, please prorate the combined data betweenthe two employee groups in the most appropriate manner. If it is not possible to prorate the combined data, enterthe total figure under office and indicate "combined" under nonoffice. If your records combine data for several items,prorate the combined figure among the items to which it related in the most appropriate manner OR report the com-bined figure and clearly Indicate to which Items it relates. If it Is not possibe to make an estimate of expendituresOr hours for an item, please enter "not available" on the appropriate line. Please complete all items. If there wereno expenditures or hours for an item, enter "0."

For each item, enter totals for the year 1970

IV. Total ColopeneatIon in 1970:

A. Gross payroll

1. Straight-time pay

2. Pay for overtime and weekend and holiday worka. Straight-time pay for overtime and wesicand and

holiday workb. Premium pay for overtime and weekend and

holiday work

3. Shift differentials

4. Pay for leave

a. Vacations

b. Holidays

c. Sick leave

d. Civic and personal leave

S. Nonproduction bonuses

6. Severance pay

B. Employer Expenditures for Legally Required Insurance

1. Socia security or railroad retirement

2. Unemployment insurance

a. Federal or railroad unemployment

b. State

3. Workmen's compensation and payments under FederalEmployer's Liability Act

4. Other, e.g., State temporary disability insurance (specify):

C. Employer Expenditures for Private Welfare Plans

1. Life, accident, and health Insurance

2. Pension and retirement plans

3. Vacation and holiday funds

4. Severance pay and supplemental unemploymentbenefit funds

S. Savings and thrift plans

6. Other private welfare plans (specify):

V. Total number of hours paid for In 1970

A. Number of straight-time hours worked

B. Number of overtime hour worked

C. Number of leave hours paid for

1. Vacation

2. Holiday

3. Sick leave

4. Civic and personal leave

Office NonoffIce

,.1 *Errirs re " r S

d'r IN, Ur t ''' ,

the 4 4 'a!

, .

Hours Hours

EMPLOYER EXPENDITURES FOR EMPLOYEE COMPENSATION

4

VI. Establishment practices and policies:

A. Paid vacations. Report the number of employees who received vacation pay during 1970 directly from the establishmentaccording to the amount of pay.

EmployeesNumber of employees receiving

Novacation

Pay

UnderI week's

Pay

I and under2 weeks'

Pay

2 and under3 weeks'

Pay

3 and under4 weeks'

Pay

4 and underS weeks'

pay

5 weeks'pay orOrrOt

Office

Nonoffice

B. Paid holidays. Enter the number of days per employee. If more than one practice existed for an employee group, reportthat which applied to the greatest number in the group. If the greatest number of employees received no paid holidays,enter "0."

Full days Half days

Office employees

Nonoffice employees

C. Sick leave. Did the establishment have a practice or a policy (even though there May have beenno expenditure) for providing paid sick leavt!, for any

Office employeesNonoffice employees

D. Civic and personal leave. Did the establishment have a practice or a policy (even though theremay have been no expenditure) for providing paid civic leave (rn'litary, jury, witness, voting,etc.) or personal leave (such as for death in family) for any

Office employersNonoffice employees

E. Life, accident, and health Insurance. Did the establishment finance any of the following Insurance plans forLife Sickness and strident Hospitalization or medical

Office employees Yes [] No C:l Yes 0 No 0 Yes No 0Nonoffice employees Yes [) No En Yes 0 No 0 Yes rj No [..1

Did employees pay for part of any of these insurance plans (answer NO if payment was only for additionalbenefits or coverage for dependents)

Life Sickness and strident Hospitalization or medicalOffice employees Yes No Li Yea 0 No Ye C7] No CDNonoffice employees . Yes 0 No ED Yes Cl No El Yea [] No 0

F. Pension and retirement plans. Did the establishment finance such a plan for


Did employees pay for part of any of these plans (answer NO If payment was only for additionalbenefits)


C. Collective bargaining. Did union-management agreements cover a majority of the


H. Regular workweek, 1970. How many hours (e.g., 44, 40, 37.5, etc.) were normally worked eachweek by the majority of the


Hours per weekHours per week

VII. Units Included in report (If different from that requested In address box);

If this report relates to units In addition to the one designated at the top of page I, please provide the following Informationfor each unit included in the report.

LocationAverage 1970 employment

Office. Nonoffice,

Principal product,service, or activity

164

167


ual industry studies the selection is based onestablishment size strata. Thus, a reportingunit which is in a stratum at which the selectionprobability was set, for example, 1 out of 5establishments, will be given a weight of 5,representing itself and four other establish-ments in this same stratum. In the biennialstudies of the entire private nonfarm sector,the probability of selection is proportionate toestablishment employment size. Thus, a report-

, ing unit employing 1000 workers, in a samplewhere the employment size probability basewas set at 10,000, will be given a weight of 10,representing itself and other establishmentshaving an aggregate employment of 10,000workers. Under both procedures all establish-ments over a certain size are included.

The sample of nonrespondents for whichdata are collected by Bureau field representa-tives is weighted appropriately to represent ailnonrespondents.

In the event that usable data cannot be ob-tained from any unit visited in person, whetheramong the followup of nonrespondents oramong large units often selected in the samplewith certainty, its weight is assigned to unitsin the sample with the most similar industry-size-location characteristics.

All estimated totals derived from suchweighting procedures are adjusted further bythe level of total employment or other appro-priate measure for the survey year, based ondata from the Bureau's monthly establishmentemployment statistics program, in each of fourbroad economic regions. For instance, if thelevel of the aggregates, as derived from theweighting procedures, is 40,000,000 in an in-dustry-region class and the corresponding levelas shown by the employment statistics programis 44,000,000, the totals of the survey itemswould be multiplied by 1.1. The adjusted datarepresent all establishments; -including thosehaving fewer than four employees, in the in-dustries studied.°

Some improvisation is necessary in the con-struction of such annual benchmark totals. Themonthly employment series provides data foronly one pay period each month, and the esti-mate of annual totals is made by mnItiplying

by the average number of weeks in a year(52.14).

Information from other sources, wherein adetailed breakdown by State or region isshown, is used as a basis for prorating thecurrent employment (or hours) estimates intoregional aggregates. Such sources include theCensus of Manufactures and County BusinessPatterns (based on Social Security establish-ment data.) 7

Presentation

The expenditure data on,the-) individual ele-ments of compensation are combined to give ameasure of total employee compensation. Thekexpenditure data for each individual elementand for groups of elements are presented as apercent of total compensation, in cents per paidhour, and in cents per working hour. Thesemeasures are shown for all establishments, aswell as for only, those establishments that hadan actual expenditure for a particular practiceduring the reference year. Hours data forworking hours and for paid leave hours arepresented 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 allX 100

establishments

2. Expenditures as a percent of total compensa-tion for establishments reporting expendi-tures=Aggregate expenditures for the practice

Aggregate compensation inX 100

establishments reportingexpenditures 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 allestablishments=Aggregate expenditures for the practice

Aggregate paid hours minusaggregate paid leave hours

6 See chapter 2, "Employment, Hours and Earnings".1 U.S. Department of Commerce, Bureau of the Census,

County Business Patterns (various years).

1

J

EMPLOYER EXPENDITURES FOR EMPLOYEE COMPENSATION 169

3. Expenditures in cents per paid hour for es-tablishments reporting expenditures=Aggregate expenditures for the practiceAggregate paid hours in establishmentsreporting expenditures for the practice.

4. Expenditures in cents per paid working hourfor eemblishments reporting ex penditures=Aggregate expenditures for the practiceAggregate paid hours minus paidaggregate leave hours in establish-ments reporting expenditures for the

practice.

The distribution of workers by establishmentexpenditure ratios and rates is published, aswell 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 payrollor nonpayroll elements of compensation arepresented in summary by this characteristic.However, the analysis of the data is relatedto the benefit function of each element. Thus,for analytical purposes, elements of compen-sation that provide similar or interchangeablebenefits are grouped together. The followinggroups of compensation elements are studied:

1. Pay for working time; straight time pay, and`premiums for overtime, weekend, holiday, andshift work.

2. Pay for leave time; vacations, holidays, miscel-laneous leave of absence, and payments to vaca-tiond 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, andworkmen's compensation.

5. Payments for unemployment benefit programs;unemployment insurance, severance pay, andseverance pay funds and supplemental unem-ployment benefit funds.

6. Nonproduction bonuses.7. Savings and thrift plans.

Data are presented on the importance ofvarious types of paid hours relative to all paidhours. Information is also published on the

number of paid holidays and number of weeksof paid vacation received by workers.


Data from the surveys are used by employersin comparing their expenditure and hours prac-tices with the averages for their industry andwith 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 theformulation of public policy, in producing esti-mates of industry output per man-hour, and inmaking international comparisons. They alsoare used in deriving estimates of the amountand type of labor compensation and the natureof the hours for which compensation is re-ceived by workers.

As indicated earlier, the expenditures studiedcomprise the significant elements of employeecompensation in American industry. The ag-gregate of the expenditures studied representstotal employee compensation. It does not, how-ever, represent total labor cost which is a moreencompassing concept and includes factors suchas the cost of recruiting and training labor, theadministrative expenses incurred in admin-istering benefit programs, and many other ex-penditures resulting from the use of labor asa factor of production. Some of these expendi-tures may be important in particular estab-lishments.

The expenditures and hours data are subjectto both sampling and reporting errors, theprecise magnitude and direction of which arenot known. Nevertheless, the errors resultingfrom sampling generally are considered to fallwithin acceptable confidence ranges ; and re-porting errors, to have a material effect on theaccuracy of the results, would have to be in thesame direction in substantially all of the cases.The omission of establishments with fewerthan four employees from the samples mayresult in some bias, but it is very small sinceless than 1 percent of the workers in mostindustries are employed by such establish-ments.

166



Number1. 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, 1966).

A study of the availability of records, willingness and ability of industry to providedata, 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 Fi-

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 (1969) ; Banks (362) ; Research Laboratories (363); Confectionery (364); Struc-tural Steel (365); Hotels & Motels, (366) ; Laundries. Cleaning &, Dyeing Plants (367) ;Men's & Boys Shirts (368).

7. Employee Compensation in the Private, Nonfarm Economy, 1968 to be published in1971.

Each expenditure bulletin contains descriptive information on the detailed proceduresand techniques used in the study.

8. Bauman, Alvin, "Measuring Employee Compensation in U.S. Industry," Monthly La-bor Review, October 1970, pp. 17-23.

ALVIN BAUMAN

167

Chapter 19. Work Stoppages

Background

Work stoppage statistics are compiled bythe Bureau of Labor Statistics to provide aquantitative measure of the extent to whichdisputes between labor and management resultin strikes or lockouts and of the immediateeconomic disruption resulting from such stop-pages.' When considered along with generaleconomic measures, these statistics also serveat times as a broad indicator of the state ofindustrial unrest.

The first attempt by any Federal agency tocompile statistics on strikes was made in1880,2 when the Bureau of the Census sentquestionnaires to employers and workers in-volved in all disputes which were noted in thepublic press during the year. Information wasreceived on 762 situations. Some data wereobtained 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 wasundertaken in 1887, when the Bureau of Labor,then in the Department of the Interior, ex-amined files of newspapers, trade journals, andcommercial periodicals for references to strikesfor all years from 1881 to 1886. Staff membersvisited the areas where strikes were reportedand obtained detailed information about eachstrike from every available person or source.Studies utilizing basically the same proceduressubsequently were made in 1894, 1901, and1906. As a consequence of these efforts, datawere published for the 1881-1905 period onthe number of strikes and workers involved,with breakdowns by industry and State; thenumber of establishments involved ; and thepercentage of strikes involving labor organi-zations.

No Federal agency collected national infor-mation on stoppages occurring during the1906-13 period. In 1914, relying exclusivelyon printed sources, the Bureau of Labor Sta-tistics attempted to compile a record of allstrikes and lockouts during the year. In thefollowing year, the Bureau inaugurated amethod for the collection of strike and lockout

material which has been followed, with modifi-cations, since that time, Briefly stated, theprocedure was to send questionnaires to theparties involved in work stoppages, followingreceipt from the press and other sources ofnotices relating to these situations.

Improvements in the program in 1927, inparticular the procurement of data on thenumber of workers involved in all stoppagesand the computation of man-days of idleness,inaugurated the modern series of monthly andannual strike data .3

Description of the Series

The present series on work stoppages coversall strikes and lockouts known to the Bureau ofLabor Statistics and its cooperating agencies.It covers all that continue for 1 full day orshift or longer and involve six workers ormire. These limitations on size and duration,somewhat arbitrary but of long standing, arenecessary for reasons of efficiency, and, in part,because of the difficulty involved in defining,identifying, and securing information onstrikes that last a few hours or less.

The Bureau defines a strike as a temporarystoppage of work by a group of employees toexpress a grievance or enforce a demand. Alockout is defined as a temporary withholdingof work by an employer (or a group of employ-ers) to enforce terms of employment on agroup of employees. Since 1922, the Bureau hasmade no attempt to distinguish between strikesand lockouts in its statistics ; both types areincluded in the term "work stoppages" and,for the sake of convenience in writing, in theterm "strikes."

Throughout this chapter, the terms "work stoppage" and"strike" are used interchangeably: both terms, unless other-wise noted, also include lockouts. The definitions, terms, andclassifications used by the Bureau in compiling work stop-page data were adopted for statistical and research purposesand have no legal significance,

*On the State level, the Bureau of T.,a b o r Statistics of Mas-sachusetts, issued a report in 1880 on strikes in that Statefrom 1825. In 1881, the Bureau of Industrial Statistics aPennsylvania issued a report on strikes in that State from

-1835.*For additional information on the early history of the

work stoppage statistics program, see BLS Bulletin 851,Strikes in the United States, 1880 to 1938 (1938).

171


Although an employer-employee dispute isimplicit in these definitions, some inclusions inthe Bureau's series relate only indirectly to thisconcept. For example, jurisdictional strikes andrival union disputes between two unions ormore often have the employer on the sidelines.In a sympathy strike, the issue of the stoppagedoes not usually involve the immediate em-ployer. Moreover, protest strikes against theactions of, governmental agencies are not theresult of a dispute between an employer andhis employees.

All stoppages, whether or not authorized bythe union, legal or illegal, are counted. On theother hand, the Bureau's series excludes strikesof American seamen or other workers in for-eign ports and strikes of foreign crews inAmerican ports. Also excluded are so-calledslowdowns, where employees continue at workbut at deliberately reduced production speed,and those instances in which workers reportan hour or two late each day as a protestgesture or quit work several hours before clos-ing time to attend rallies or mass meetings.

The number of work stoppages occurringduring a given period provides a measure ofthe frequency of disputes; the severity andeffect of such actions are measured by thenumber of workers involved, duration, and theresultant 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 stoppagefor a common objective, their action is classedas a single strike.

The figure used for the number of workersinvolved in a strike or lockout is the maximumnumber actually made idle in the establis:unentor establishments directly involved. No dis-tinction is made in arriving at this figure be-tween the active participants in the strike, thenumber of union members or workers coveredby an agreement, and those sent home by theemployer when a stoppage by one group ordepartment prevents plant operation.

Man-days of idleness, like the number ofworkers involved, are based on the idleness atthe establishment or establishments directlyinvolved. Workers involved multiplied by work-

days lost equal total man-days idle. In instanceswhere the number of workers idle varies dur-ing the period of the stoppage, appropriateadjustments are made in this calculation. Al-lowance is made in these computations forholidays and days not normally worked.


The task of collecting strike data has twobasic elements: (1) to learn of work stoppageswhen 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 stoppagescurrently is obtained from various sources, in-cluding: (1) clippings from daily and weeklynewspapers throughout the country providedby commercial clipping services; (2) noticesreceived directly from the Federal Mediationand Conciliation Service; (3) a periodic com-pilation by the local offices of the State employ-ment security agencies, provided through theU.S. Training and Employment Service of theU.S. Department of Labor ; (4) a number ofother State agencies, such as State mediationboards and labor departments; (5) variousemployers and employer associations; (6) in-ternational unions and their publications; (7)firms under contract to the Atomic EnergyCommission; and (8) other Federal agenciesand commissions.

Aside from the clippings from newspapersand other publications, most of these sourceshave been developed over a period of years. Asa general rule, expansion in the Bureau's"lead" sources brings an increase in the num-ber of stoppages reported, but has little effecton the total number of workers and man-daysof idleness, because the added stoppages tendto be small.

After the receipt notices regarding theexistence of work stoppages, the Bureau mailsquestionnaires to the parties involved to securedirect information on each stoppage. Shoulda reply not be received within 3 weeks, a sec-ond questionnaire is mailed, and, in the case ofcontinued nonresponse, a mailogram or tele-gram may be sent, or an effort made to securethe necessary data by telephone. In some in-

.-,

WORK STOPPAGES

stances of nonresponse, field representatives ofthe Bureau secure the necessary data; inothers, cooperating State agencies may beasked to contact the parties.

The types of information sought by the Bu-reau through its questionnaire have changedover the years, partly in response to changingneeds. The primary function of these reportsis to compile statistics, not to keep records onthe strike activity of individual firms andunions. The separate questionnaires currentlyused for private and public sector disputes areshown on pages 174-176.4

Although strikes, by their very nature, areusually matters of public knowledge and ofreporting by newspapers and other publica-tions, the Bureau holds confidential the indi-vidual reports submitted by employers andunions, as well as supplementary data collectedthrough State or Federal agencies. The rulesof confidentiality observed here are similar tothose followed in other Bureau surveys.


Since the Bureau is able to obtain informa-tion on virtually all work stoppages within thescope of its definition, estimating is necessaryonly in the preparation of its monthly reportson the level of strike activity in the UnitedStates as a whole. The availability of reason-ably accurate data on the larger stoppages atthe time these estimates are prepared ap-proximately 4 weeks after the end of the monthof referenceassures approximate conformityto the final statistics which are based almostexclusively 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 alag between the occurrence and reporting of anumber of relatively small strikes, the numberof stoppages beginning during a given monthis estimated by increasing the number ofstrikes on which leads have been received by apercentage which is fixed for each calendarmonth. An estimate of the total number ofstoppages in effect during the month is ob-tained by supplementing the latter estimate by

173

a percentage of the stoppages in effect duringthe prior month.

In estimating the number of workers in-volved and total idleness, efforts are made toobtain as much preliminary information as pos-sible-' on the size and duration of individuallarge stoppagesthose of at least 500 workersor 5,000 man-days of idleness. To the knownfigures for these large stoppages is added theproduct of the estimated number of smallerstrikes and the average number of workers (orman-days) that previous experience indicatesfor such stoppages.

In its preliminary reports, as well as in itsfinal reports, the Bureau relates the man-daysof idleness to the total estimated working timeof all workers. The "total economy" measureof strike idleness, which was instituted in1967, includes government and agriculturalemployees as well as private nonfarm workersin its employment count as well as in thecomputation of idleness ratios. Before 1967, theBLS series excluded government and agricul-tural workers from employment totals, but ac-counted for time lost by these workers whileon strike. This reevaluation of methods hasimproved the calculations of idleness and madethe Bureau's measurement of work stoppageintensity national in scope.

Analysis and Interpretation

The data presented in the parties' reportsare analyzed and classified according to a num-ber of significant factors, briefly describedhere:

(1) Each strike is assigned an industrialclassification in accordance with the StandardIndustrial Classification Manual prepared bythe Bureau of the Budget.5 In those rases inwhich a stoppage affects workers in more thanone industry, one of two procedures may befollowed. If the stoppage is small, the strikeis classified in the industry in which it wasinitiated ; in large interindustry stoppages, a

A modified form of this questionnaire is used in the caseof 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 thestatus of the stoppage.

See appendix B.

170

174 WORK STOPPAGES

BLS 817(Rev. 1963)

WORK STOPPAGE REPORTYour report will be held

in confidencePlease answer all questions

NO

L

U. S. DEPARTMENT OF LABORBUREAU OF LABOR STATISTICS

WASHINGTON, D. C. 2 0212

7

_J

Form ApprovedBudget Bureau No. 44-R210

This request for information relates to:

I. Employer Name and mailing address

2. Establish.ment sinvolved instoppage

Number of establishments directly involved or in which workers observed picket lines

If more than one establishment, use reverse side; if one, enter below:a. Locationb. Industry

(Indicate major type of activity and also principal products or services. e.g., Mining-bituminous coal; Con-struction -highways and streets; Manufacturing -wooden upholstered furniture; Wholesale trade -plumbingsupplies; Transportation-motor freight.)

3. Unioninvolved

Name AFL

IDCIO

OtherLocal No, Address

4. Dates ofstoppage

Stoppage began on Settlement was reached on Settlement was ratified on(if applicable)

Employees returned towork on

5. Number ofworkersaffect ed

TOTAL workers idled at least one full shift or day workers

(IMPORTANT-Include all workers directly involved in the stoppage and workers made idle by lack of work in thtsame establishments or by observance of picket lines. If exact figures are not available, please provide estimate.

Did the number idle change significantly during the stoppage? Yoe: No(If "yes," please enter changes in number idle and dates of changes on reveree side of sheet.)

6. Normal workweek prior to stoppage days

7. Contract status (check one)Stoppage occurred

In negotiation of first contract or in obtaining union recognitionln renegotiating contract terms (expiration or reopening)

During term of contract (change in contract terms not involved)Other (specify)

8. MAJOR issues in dispute (list in order of importance)

9. Did the agreement to return to work include a procedure for handling any unsettled major issues involved in the stoppage (e.g.,by submitting issues to arbitration)? Yes No If "yes", indicate the issues and the procedure agreed upon.

10, Did a Federal, State, or local government agency, or a private mediator, mediate in this dispute or assist in arranging the returnto work? (Check more than one if applicable). Federal State Local -Private None

Please identify assisting government agency, if any

(Date) (Signature and title of person making report) (Company or organization)Use reverse side for any

clarifying remarks.

4


Supplementary Information for Items 2 and 5:

If the stoppage involved more than one establishment or if idleness varied from period to period duringthe stoppage, please use the following space to indicate the number idle in each establishment and thevariation in idleness at different dates. Include both workers directly concerned and those made idlebecause of dispute in the same establishment.

IF EXACT FIGURES ARE NOT AVAILABLE, PLEASE FURNISH ESTIMATES.

Establishment involved and location(City, County, State)

Industry orprincipalproduct

Approximate numberof workers idle

a full shift or more

Dates this numberwas idle

a full shift or more

A

REMARKS:

Or U. O. GOVERNMENT RINTING ORRICE: 1913930 1.7 14

1 72

176

BLS 3006

WORK STOPPAGES REPORTGovernment

No.

WORK STOPPAGES

U.S. DEPARTMENT OF LABORBUREAU Of LABOR STATISTICS

WAS:IINOTON, D.C. 20212

Budget Bureau No. 44-111397Approval expires 6/30/71

This request I. r infornsation relates rot

1. Government Department. Agency. or installation

Name: Address:

Facility where stoppage occurred (name if different from above):* Address:

Level (check one): 0 Federal 0 Municipality School district0 State 0 County 0 Other (specify) -

Function (check one):0 Administrative services Sanitation services Hospitals and health services

Welfare services . Education Transportation and allied facilities0 Law enforcement and conection rj Streets and highways 0 Other utilities0 Fire protection Parks, recreation, libraries, etc. 0 Other (specify)

2. Union or Association

Name: Local no. Affiliation:

0 AFL-C10Cl Other union0 Employee association

Did the organizationcall or support thework stoppage?Yes NoE) No information

Does this organizationhave official recogni-don?

Yes NoAddress:

,

3. Dates of Stoppage and Workweek

Stoppagebegan on:

Settlementreached on:

Settlementratified on:

Employees returnedto work on:

Scheduled '.orkweek prior tostoppage

Days Hours

4. Employees Affected

Total employees idled at least one full (IMPORTANT - Include all employees directly involved in the stoppage and employeesshift or day made idle by lack of work in the same facilities or by observance of picket lines. If

exact figures are not available, please provide estimate.)

Did the number idled change significantly during the stoppage? J Yes No(If "yes" please enter changes in number idle and dates of changes on reverse side of this form.)

Occupational classification (check one or more):Ei Teachers Policemen Other blue collar and manualEl Nurses 0 Firemen ICI Other (specify)

Other professional and technical employees 0 Sanitation men0 Clerical 00 Craftsmen (specify) .

5. Agreement Information

Stoppage occurred (check one): .-

0 In attempting to obtain . 0 During agreement term ID No formal agreement involvedrecognition (change in terms not involved) Other (specify)

El In negotiating first 0 In renegotiating agreement ---agreement (expiration or reopening)

Major issues in dispute in order of importance:

,..

Did employees return to work voluntarily, or under the terms of a count order or injunction?

Did agreement to return to work include a procedure for handling any unsettled major lames Involved its the stoppage (e.g., bysubmittal to arbitration or factfinding)?0 Yes No If yes, note issues and procedures agreed upon on reverse side of th's form.

Did a government agency, or private individual, or organization assist in arranging the return to wor;c?(Check one or more): Federal State Local Private 1M NonePlease identify government agency

Signature of person making repon: Title: Department or organization: Dater

* If more than one facility was involved, please enter information on reverse side of this form. Also use reverse side forclarifying remarks, particularly on nature of stoppage (mass sick leave, or resignations, etc).

4 144

WORK STOPPAGES

stoppage is recorded for each industry affected,and the approximate numbers of workers andidleness are allocated to each.

(2) The duration of each stoppage is takenas the number of calendar days from its be-ginning to end. For stoppages which begin ata definite time and are terminated by a formalagreement, no problem arises in the determina-tion of duration. However, some stoppages, fora variety of reasons, are never settled formally.These range from situations in which theworkers gradually return to their jobs withouta settlement to those in which the employerdecides to go out of business. In cases of theformer variety, the details of each individualsituation are studied before a stoppage isterminated for statistical purposes ; in the lat-ter instances, the stoppage is terminated withthe employer's announcement of his decisionto discontinue operations. On occasion, if actualsettlement is reached later, the statistical rec-ord of the stoppage is adjusted accordingly.

(3) Number of establishments involved. Thestandard definition of establishment is used.(See appendix B.) An establishment is a singleworkplace, for example a factory, mine, orstore. In a widespread strike of intercity busdrivers, truckdrivers, or railroad workers, theestablishment is regarded as the terminal outof which the employees work; in a strike ofseamen, the ship is the establishment ; and ina strike of dockworkers, the individual dock orloading place is regarded as the place of work.

(4) Geographical classification of -stoppagesfollowed State and city boundary lines, through1951. Beginning in 1952, the compilation ofdata by Standard Metropolitan StatisticalAreas superseded city boundary lines.° Ininterarea stoppages, a stoppage is recorded ineach area affected, and workers and man-daysof idleness are allocated proportionately.

(5) The issues in dispute in most strikes aremany and varied, and do not always lend them-selves readily to immediate and exact classifica-tion. Stoppages are classified by major issueinto the following broad groupings: (a) wages,hours, and supplementary benefits ; (b) unionorganization and security; (c) job security ;(d) plant administration ; and (e) inter- or

f.

177

intra-union matters. Each of these groups issub-divided into more specific categories.'

(6) Stoppages are classified by tbF, con-tractual relationship existing between theparties involved. The following four situationsapply : (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 majorclassification. For this purpose, the union is theorganization whose contract was involved orwhich has taken active leadership in the stop-page. Disputes involving more than one unionare classified as jurisdictional or rival uniondisputes or as involving cooperating unions. Ifunorganized workers strike, a separate classi-fication is used. For publication purposes,union information is presented by major affilia-tion of the union, i.e., AFLCIO, 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 aresettled involves classification into the follOwing_categories: (a) those ending with a formalsettlement; (b) those terminating without aformal settlement, and those in which work isresumed with either the old or new workers;and (c) those concluded by the employer'sdecision to go out of business.

(10) A classification also is made of themanner in which unsettled issues are to beresolved in those situations where strikes areterminated with the understanding that suchmatters will be handled after the resumptionof normal operations.

Presentation

Publications in the area of work stoppages.include monthly preliminary estimates, annual

"See appendix C.7When a major change in the classification of issues was

introduced in 1861, the Bureau included in its annual reportfor that year a method of linking the new classifications withthe old.

174

ti


reports, and special reports which are issuedirregularly.

Monthly preliminary estimates are issuedapproximately 30 days after the end of themonth of reference in the form of press re-leases. Such reports presently are prepared forthe 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 theclose of the calendar year.

Selected final tabulations of strike activityare presented in a release which usually is is-sued in June. An annual bulletin,8 containingdetailed information on the characteristics ofwork stoppages during the prior year, is pub-lished each winter.

Special reports containing historical workstoppage data by industry and area, or ananalysis of a particular aspect of strikes, areissued irregularly. This latter category alsomay include Bureau reports of a nonstatisticalnature, including chronologies of "nationalemergency" disputes arising under the termsof the Taft-Hartley Act.


The use of strike statistics as an indicator ofindustrial unrest has been the traditional rea-son for their compilation in the United Statesand in other industrialized countries.9 Whetherthey serve this elementary purpose today isopen to question.'° In any event, some qualifica-tions must be taken into consideration in thisuse. The willingness of ,workers to strike as aprotest against existing conditions may be en-couraged or deterred by outside influences, suchas the employment situation, the state of thebusiness cycle, and possible political or publicreaction. Within the plant, the strength of theunion or of employer opposition may influenceboth the willingness of the workers to start astrike or to extend its duration.

Whether as a measure of industrial unrest orthe state of labor-management relations, strikestatistics are necessary for Federal, State, andmunicipal government agencies, particularlythose concerned with labor affairs. Unions, em-ployers, and employer associations use strikedata to assess their own experiences, and busi-

175

ness and civic organizations are concerned withtheir community promotional possibilities.Schools, particularly those teaching courses inindustrial relations, and industrial relationscounselors also find strike statistics useful.Finally, the press is interested, since strikesand information about them are newsworthy.

Although it is virtually certain that the Bu-reau is able to locate, and obtain informationon, the larger work stoppages, some smallstrikes undoubtedly escape notice each year.While these omissions do affect data on numberof strikes, the statistics on workers and man-days of idleness are virtually complete. As hasbeen noted, the addition of new sources of in-formation has not materially changed theselatter figures, but these new sources have actedto lessen slightly the degree of comparabilityin the number of strikes reported from periodto period. It follows that the narrower theclassification of strike data, the greater is thechance of a significant omission. For example,while the figures for individual States may betaken as reasonably complete, the figures for aspecific industry group within a State may beappreciably affected by the ommission of onestrike.

Secondary idleness is not measured ; that is,the figures do not cover those employees madeidle in other establishments or industries asa result of material or service shortages result-ing from a work stoppage. At times, the idle-ness of employees directly involved in a strikemay be considerably less than the idleness ofother workers brought about indirectly. Nosatisfactory method, however, has been foundto measure or estimate such indirect effectsadequately.

The Bureau does not attempt to measure thecost of strikes in terms of the amount of pro-duction and wages lost. The calculation of cost

0Since 1949, these bulletins have been titled Analysis ofWork Stoppages (year).

0For a discussion of the methods used in compiling strikestatistics in other countries, see Methods of Compiling Statis-tics of Industrial Disputes, Geneva, International LabourOffice, Studies and Reports, Series N, No. 9 (1926). See alsoA. 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 publishedannually in the Year Book of Labour Statistics,Geneva, In-ternational Labour Office.

10 See article by Joseph W. Bloch, "The Strike and Macon.tent," Monthly Labor Review, June 1963, pp. 645-651.

it WORK STOPPAGES 179

involves many complex and interrelated factorsfor which information is not readily available,including such matters as production schedulesbefore 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."The problem is magnified beyond statisticalcontrol if secondary costs are to be accountedfor.

" A framework for the measurement of the cost of strikesis developed by Neil W. Chamberlain and Jane MetzgerSchilling in The Impact of Strikes: Their Social and EconomicCosts, New York, Harper & Bros. (1954).


Number1. 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 analyzingwork stoppages in the United States.

4. Strikes in the United States, 1880 to 1936 (Bulletin 661, 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.

EDWARD D. ONANIAN

Chapter 20. Collective Bargaining Agreements

Background

Collective bargaining agreements and relateddocuments setting forth the provisions of em-ployee-benefit plans provide a valuable sourceof information on industry wage practices,supplementary benefits, job and union security,the timing of wage negotiations, the nature ofplant operations and working conditions, andmany of the day-to-day aspects of employer-employee and union-management relationships.

The Bureau of Labor Statistics has utilizedthese basic industrial relations materials forpublic and Government informational purposesin three major ways, by : (1) maintaining a fileof current agreements open to public inspectionand inquiry; (2) preparing reports-which re-produce representative agreement provisionsor the variety of provisions relating to similarproblems, or digests of selected identifiedagreements or benefit plans ; and (3) by pre-paring studies measuring the prevalence andcharacteristics of specific types of agreementand benefit plan provisions or of other aspectsof collective bargaining such as multiemployerbargaining.

The development of industrial relations prac-tices that are now widely prevalent is reflectedin the Bureau's studies over the years. The Bu-reau's interest in the collection and analysis ofunion agreements dates back almost 70 years.'Systematic efforts to collect agreements beganin 1912. The first of a number of BLS bulletinsdevoted entirely to the subject of collectivebargaining agreements appeared in 1925. Alarge number of reports and bulletins on a widevariety of industrial relations subjects havesince been published.

The Bureau's responsibility in the field ofagreement collection and analysis receivedadditional sanction and guidance in the LaborManagement Relations (Taft-Hartley) Act,1947, section 211, which reads as follows:

Sec. 211. (a) For the guidance and information ofinterested representatives of employers, employees, andthe general public, the Bureau of Labor Statistics ofthe Department of Labor shall maintain a file of copiesof all available collective bargaining agreements andother available agreements and actions thereundersettling or adjusting labor disputes. Such file shall be

open to inspection under appropriate conditions pre-scribed by the Secretary of Labor, except that nospecific information submitted in confidence shall bedisclosed.

(b) The Bureau of Labor Statistics in the Depart-ment of Labor is authorized to furnish upon request ofthe [Federal Mediation and Conciliation] Service, or kemployers, employees, or their representatives, allavailable data and factual information which may aidin the settlement of any labor dispute, except that nospecific information submitted in confidence shall bedisclosed.

Concepts and Scope

Although the substance of collective bargain-ing rests partly upon a foundation of unwrittenindustry, company, and union practices, andupon various legal requirements, the basic unitin agreement collection and analysis is the writ-ten agreement itself. The agreement may covera single plant, a number of plants of .a multi-plant company, or a number of companies, insome cases over a thousand, bound togetherformally or informally in an association forcollective bargaining purposes. It may expressconditions of employment in simple terms, andleave much of the administrative details andother matters to the day-to-day relationshipsbetween the parties, or it may attempt to coverall details and, thus, leave as little as possible tolater bargaining or haggling. Agreements varyin size from a few sheets to over 300 pages ofa pocket-sized booklet. They reflect the diver-sity of employment conditions among Indus-tries and companies and of the scope of theissues over which bargaining takes place, aswell as differences iii the degree of precisionsought 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 isestimated at about 20 million. The Bureaupresently maintains a file of approximately

A bulletin of the Department of Labor, Number 42, Sep-tember 1902, included this note: "It is the purpose of thisDepartment to publish from time to time important agree-ments between large bodies of employers and employees withregard to wages. hours of labor, etc. The Department wouldbe pleased to receive copies of such agreements whenevermade." (p. 1057)

Between 1888 and 1903, the Bureau of Labor (now theBureau of Labor Statistics) had independent status as aDepartment of Labor, under the direction of a commissioner.

(1.81

177


6,500 current agreements in the private sector,covering about 9.8 million workers? All in-dustries are represented in the file except rail-roads and airlines. Since railroads and airlinesare required to submit copies of agreementsto the National Mediation Board, the Bureaudoes not attempt to collect these agreements.

As a result of the recent growth of collectivebargaining in the public sector, the file hasbeen expanded to include a variety of docu-ments covering employees of the Federal Gov-ernment, States, counties, cities, and specialjurisdictions. These documents range fromtraditional collective bargaining agreementsthrough memoranda of agreement to executiveorders, administrative promulgations, and leg-islative actions, which are clearly the resultof bilateral negotiations. The public file isgrowing daily, presently exceeds 1,000 docu-ments at all government levels, and involvesmore than 1 million public employees.

The Bureau's quantitative analysis of agree-ment provisions covers virtually the entirerange of issues dealt with in collective bar-gaining.2 The basic assumption underlying suchanalysis is that the variety of subjects can bedefined, classified, and counted..

In its general analysis of agreements, as dis-tinct from special industry studies, the Bureauis concerned with these major objectives: (1)the presentation of data by industry group andfor manufacturing and nonmanufacturing as awhole, (2) the presentation of data by regionor union affiliation if the subject requires-'it,(3) a realistic measure of representativenessin the agreements studied, and (4) the study' ofpractices which are (as yet) relatively uncom-mon in collective bargaining agreements. Foragreements covering public employees, dataare presented by the type of government ac-tivity (police department, veterans administra-tion, etc.) and for cities, by population size.

Methods of Collection and Analysis

Collection of Agreements

The selection of agreements for the file iscurrently based on two guides: To enlarge tothe fullest the opportunities for public and

governmental use of the file,' and to provide adiversified collection of agreements for specialreports, which the Bureau occasionally is calledupon to prepare. The extent to which theseobjectives are fulfilled is obviously affected bythe size of the file. A third guidethat of con-structing a file which truly represents all publicand private agreements and thus provides afirmer basis for sound generalizations on allagreementshas long been a goal of theBureau.

The maintenance of a current file of agree-ments is a continuous undertaking because oftwo factors : (1) The typical agreement has afixed -duration, after which it is no longerconsidered current (unless notice of renewalwithout change has been received), and (2)submission of agreements to the Bureau isvoluntary on the part of employers or unions.To allow for the ratification and the printingof new agreements, requests for copies aremailed about 2 or 3 months after the expira-tion date indicated in the previous agreementor upon other notice of contract change. As inother phases of the Bureau's work, the volun-tary cooperation of employers and unions is ofutmost importance. Any restrictions imposedby respondents on the public use of agreementsare observed scrupulously by the Bureau.

To facilitate the use of the file in accordancewith the types of requests customarily made,each agreement received is coded for a series ofidentifying features, which include union, loca-tion, number of workers covered, industry, andeffective and expiration dates. For public em-ployee documents, the level of government andthe government activity also are coded.

Agreement Analysis

The Bureau's utilization of the'private sectoragreements it collects has moved through dif-

During the early postwar period, the number of agree-ments on file exceeded 12,000.

n See Program Note by Joseph W. Bloch, "Union ContractsA New Series of Studies," Monthly Labor Review, October1984. pp. 11844185.

The agreements file is located in the Washington Office ofthe Bureau's Division of Industrial Relations. Agreementssubmitted to the Bureau with a stipulated limitation on publicuse are not available for inspection. Requests for informa-tion concerning specific agreements or agreement clauses areaccommodated, depending upon the nature of the request,within the limits of staff resources.

COLLECTIVE BARGAINING AGREEMENTS 183

ferent stages over the years, in pace with, orcontrolled by, the increasing prevalence andmaturity of collective bargaining. During theearly years, significant agreements were re-produced in their entirety. As collective bar-gaining spread, and the size and representativecharacter of the Bureau's file increased, atten-tion was directed towards reproducing andanalyzing the variety of agreement clauses re-lating to similar subjects, culled from a largenumber of agreements. Although illustrativeclauses continue to be utilized in most of theBureau's agreement studies, major emphasisduring recent years has been devoted to meas-uring the prevalence and characteristics ofparticular provisions and of types and levelsof .benefits. The Bureau's Bulletin 1425 seriesrepresents its most comprehensive efforts todate. The Bureau also has undertaken abroader, more extensive rather than intensiveanalysis, in which it searches the agreement tomeasure the prevalence, but not the detailedcharacteristics of more than 100 different col-lective bargaining provisions. In these kindsof analysis, problems relating to techniques ofcoding and analysis come to the fore.

In 1948 and 1949, when the Bureau's fileconsisted of more than 12,000 agreements, itwas decided that a sample of 3,000 agreementswould be feasible. The selection of specificagreements was based on a number of factors,including industry, agreement coverage, loca-tion, union representation, and bargainingpractices. Limited data upon which to base arepresentative selection of agreements werecompensated for, at least hi part, by extensiveexperience with collective bargaining practiceson the part of the sample selectors.

During subsequent years, however, availablestaff resources were not sufficient to deal with asample of this size. The feasibility of recon-stituting a sample of 1,500 to 2,000 agreements,which had become the maximum work load, andof assuring appropriate safeguards againstdeterioration, was rejected as being beyond theresources of the staff and the available data.The most advantageous alternative was to lbasethe private industry agreement studiei on allagreements covering 1,000 workers or moreand, thus, to avoid sampling. The Bureau's file

already contained almost all of these; the Bu-reau's monthly report, Current Wage Develop-thenis, was a ready source of information onthose that were not included. The total numberof workers covered by agreements of this size(now between 1,700 and 1,900) is about 7.7million and represents a very substantialworker coverage in agreement studies. Thenumber of establishments covered is notknown.5

A key analysis list containing all private in-dustry agreements covering 1,000 or moreworkers, although perhaps not the ideal cover-age, has definite advantages: (1) It achievesmaximum worker cove: age in the studies fora given investment of staff resources ; (2) itprovides a simple, objective measure of thecoverage of the studies ; (3) it permits thepresentation of various combinations andbreakdowns of the data without the necessityof complicated weighting schemes and withoutthe bias resulting from the lack of properweighting; (4) it is safeguarded against obso-lescence, since the Bureau is best able to keepabreast of changes in agreements of this size;and (5) it has a significant meaning to usersof these studies.

The distinguishing feature of agreementanalysis is that it deals mainly with 'legal lan-guage, which requires interpretation, ratherthan with numbers or other universal, sharplydefined attributes. The process of analysis con-sists of interpreting provisions, reducing themto numbers (codes), aggregating the numbers,and converting the aggregates back to types ofprovisions and prevalence. Data are presentedin terms of number of agreements and numberof workers covered. Since agreement provision'son the whole are notable more for their varietyof expression and details than for their uni-formity, the, process of analysis becomes asimplification process by which some of theoriginal content and variety is lost. Under suchcircumstances, the planning of studies acquires

6 The distinction between size of agreement (employees cov-ered). and size of establishment is important. A *ubstantialproportion of these agreements are association negotiated andcover a large number of small establishments. Two agree-ments, for example, involving the United Mine Workers .(IND.), cover most of the anthracite and bituminous coalmines in the country. Some association agreements in NewYork State cover more than a thousand firms.

e9


a special importance if significant differencesare not to be buried.

The keystone of agreement analysis study isobviously the interpretation of the agreementand the assignment of the predetermined codenumber. For some subjects, an agreement mustbe read in its entirety; for others, only a por-tion. Long and legal provisions must be reducedto their essentials. Since the interpretation ofagreement provisions is often a troublesomematter for the parties themselves (as reflectedin the widespread adoption of provisions forarbitration), misinterpretations undoubtedlyoccur. These are kept to a minimum. by a staffexperienced in agreement analysis and by con-tinuous efforts to assure consistency of in-terpretations.


The studies of agreement provisions are ofpractical use to public and private employersand unions engaged in collective bargaining, toarbitrators and factfinding boards, to admin-istrators of company wage and industrialrelations programs, and to legislators and Gov-ernment officials. Persons not directly involvedin collective bargaining or in related admin-istrative activities (e.g., teachers and studentsof labor problems, writers for newspapers andtrade and technical journals, and foreign ob-servers) find value in the broader aspects ofemployer-employee relationships revealed inthese studies.

The limitations of these studies of agreementprovisions and employee-benefit plans are de-termined, in large part, by their application.For studies of paid holiday provisions or othersupplementary benefits, the fact that thesestudies cover only the area of collective bar-gaining may constitute a limitation on .gen-eralizations applying to all workers but notnecessarily on their uses in collective bargain-ing or in wage and employee administration.On the other hand, these studies do not showlocality practice, which may reduce their use-fulness for some collective bargaining purposesbut not for broad generalizations relating toworkers under collective bargaining.

Additional limitations of agreement pro-vision studies are inherent in the selection ofagreements for studythe exclusion of rail-road and airline agreements and of agreementscovering fewer than 1,000 workersand in thetechnique of analysis, as iindicated previously.Limitations are also connected with the par-ticular subjects studied, which are-pointed outin each study. A funCiamental limitation whichmust be borne in mind in connection with certainstudies (e.g., grievance procedure) is that theyrelate to written policy rather than actualpractice. Practices which are not provided for

In the agreements but are, instead, matters ofcompany policy going beyond the agreement,traditional industry policy, informal acceptanceby management and unions, or arbitrators' de-cisions, can be neither detected nor measuredin agreement analysis.

4

f

..;

COLLECTIVE BARGAINING AGREEMENTS


Number1. Bloch, Joseph W. Union ContractsA New Series of Studies. Monthly Labor Review, October

1964, pp. 1184-1185.1

2. U.S. Department of Labor, Bureau of Labor Statistics, Major Collective Bargaining Agree-ments (Bulletin 1425 Series).

1425-1, 1964: Grievance Procedures1425-2, 1965: Severance Pay and Layoff Benefit Plans1425-3, 1965: Supplemental Unemployment Benefit Plans and Wage-Employment

Guarantees1425-4, 1966: Deferred Wage Increase and Escalator Clauses1425-5, 1966: ManagemenN Rights and Union-Management Cooperation1425-6, 1966: Arbitration Procedures1425-7, 1969: Training and Retraining Provisions1425-8, 1969: Subcontracting1425-9, 1969: Paid Vacation and Holiday Provisions1425-10, 1969: Plant Movement, Transfer, and Relocation Allowances1425-11, 1970: Seniority in Promotion and Transfer Provisions1425.12, 1970: Administration of Negotiated P9nsion, Health, and Insurance Plans

3. Bulletin 1353, Major Union Contracts in the tiv,i0d States, 1961 (1962).'4. Bulletin 1686, Characteristics of Agreements Overby, '5,00 Workers or More (1970).

JOSEPH W. BLOCH

185

Chapter 21. Union Membership

Background

The Bureau of Labor Statistics' interest inunion membership as a significant social andeconomic indicator is of long standing. Thefirst BLS publication listing membership fig-ures for national and international labor unionsin 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 anextensive revision in 1936 (BLS Bulletin 618).The Handbooks, however, devoted little spaceto overall membership figures or trends; theirmain emphasis was on the origins and earlyhistory of particular unions, their governmentand structure, trade jurisdiction, and types ofbeneficial activities.

In 1939, on a modest scale, the Bureau beganto publish an annual, trade union directory, butit was not until 1948 (Bulletin 937) that thelisting of union officers, headquarters' ad-dresses, etc., was supplemented with an entryfor each union on membership and number oflocal unions, and a compilation of total mem-bership. In subsequent years the informationsought from national and international unionshas expanded considerably. Data on womenmembers first appeared in the 1953 Directory,and separate tabulations for areas outside theUnited States (Canada, Puerto Rico, etc.), inthe 1955 edition. Since 1957, each Directo'ryhas carried information on the number andproportion of members who are white-collarworkers and on those in major industry groups.State figures were introduced on a limited basisin 1959, when AFLCIO central bodies wereasked to furnish estimates on the number en:-rolled by Federation affiliates. In its 1965 Di-rectory, the Bureau showed State figures asreported by national and international unions.Many of the items referred to have been re-fined since they were first introduced, and theaccumulated information now permits analysis

i For a membership serve/ of these unions see UnaffiliatedIntrastate and Single-Employer Unions, 1967 (BLS Bulletin:840, 1989).

2 The names of all reporting unions appear in the Depart-ment's Register of Reporting Labor Organizations, last issuedin 1988.

of trends in total membership and several ofits components.

Data Sources and Collection

For part of the Bureau's biennial Directoryof National and International Labor Unions inthe United States, unions meeting the criterianoted are asked to report the average numberof dues-paying members or the number ofmembers in good standing for the 2 most recentyears, including members outside the UnitedStates ; the proportion of white-collar andwomen members ; breakdowns by major in-dustry and by State, and other membershipinformation. (See copy of questionnaire on pp.188-191.) To be included in the Directory, aunion must be an affiliate of the AFLCIO or,in the case of unaffiliated unions, a party tocollective bargaining agreements with differentemployers in more than one State. Among theunions in the latter category are such un-,affiliated unions as the Teamsters, the.. MineWorkers, 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 unionswhose activities are confined to a single localityor to a single employer.1 Although the Bureausends questionnaires to all multilocal unionswho have submitted reports to the U.S. Depart-ment of Labor's Office of Labor-Managementand Welfare-Pension Reports, as required bythe Labor-Management Reporting and Dis-closuere Act,2 possibly some small unaffiliatedunions, 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 questionnaireitems, estimates are derived from other sourceson, file, notably union periodicals, conventionproceedings, financial statements, and collec-tive bargaining agreements. No sampling pro-cedures are used ; the data are based on the

fi 6/187

182

YA


t.

-

BLS 2441

U. S. DEPARTMENT OF LABORBureau of Labor StatisticsWashington, D. C. 20212

DIRECTORY OF NATIONAL AND INTERNATIONALLABOR UNIONS IN THE UNITED STATES, 1971

I. Union and Officer Identification:

Beigo Bums. 110. 44.770027Apfm...1 Empire, INK. 21, 1071

Telephone numberand ens cedes

1. Affiliation(Check appropriate hos) III AFL-CIO IN None III Other(specify)

2. President: III Mr. Mrs. N Miss Year first elected to thisoffice.

3. Secretary - Treasurer: III Mr. 0 Mrs. III Miss Year first elected to thisoffice:

4. Person in charge of organizing activities: NI Mr. III Mrs. Miss

Tide:Mailing address if different from headquarters (Street, City, State, Zip Code)

5. Research Director: Mr. 0 Mr.. C:3 Miss

Mailing address if different from headquarter. (Street, City, State, Zip Code):

6. Education Directot: Mr. III Mr.. II Miss

Mailing address if different from headquarters (Street, City, State, Zip Code)

;"

7. Person in charge of social insurance (health, insurance, pension, etc.) at.dt-itier:

III Mr. III Mrs. III Miss Title.

Mailing address if different from headquarters (Street, City, State, Zip Cade)

8. Person in charge of legal aclvities: III Mr. III Mrs. III Miss

Tide:Mailing address if different from headquarters (Street, City, State,, Zip Cade)

9. Person in charge of legislative activities: Ill Me. III Mrs. go Miss

Title;Mailing address if different from headquarters (Street, City, State, Zip Code)

10. Person in charge of public relations activities: III Mr. 0 Mrs. III Miss

TitleMailing addreu if different from headquarters (Street, City, State, Zip Code)

UNION MEMBERSHIP

? -

II. Conventions and Publicotions:

I. Frequency of conventions: ED Anounl Semi-annual Biennial Other (specify)

2. Next convention:Month Day Year (City, State)

3. Name of official publication(s) How often published Editor'

III. 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 1970.11 complete returns for 1970 are not available,'use 9 or 10 month average:

1969 members 1970 members

2. If your records do not permit an annual average dues-paying membership count, please indicate the number of members ingood standing or those carried on your rolls:

1969 members 1970 members

3. Indicate if retired members are included in 1 or 2 above: Yes NoIf yes, indicate number of retired members included: members

V. Classification of 1970 membership:

1. Indicate approximate percentage of membership who are women:(if none, enter zero)

2. Indicate approximate percentage of membership in the following "white-collar" categories:

Professional and/or technical %

Clerical*Sales

;Total "white-collar"-(if none, enter zero)

3. Industry composition of union membership. Indicate the approximate percentage of all union members working inestablishments in each of the following industry groups:

Manufacturing:Ordnance and accessoriesFood and kindred products (incl. beverages)Tobacco manufacturesTextile mill productsApparel and other finished products made from fabrics and similar material.Lumber and wood products, except furnitureFurniture and fixturesPaper and allied productsPrinting, publishing and allied industriesChemicals and allied productsPetroleum refining and related industriesRubber and misc. plastics products'Leather find leather productsStone, clay, glass, and concrete productsPrimary metals industries,Fabricated metal products; except ordnance, machinery and transportation equipmentMachinery, except electricalElectrical machinery, equipment and suppliesTransportation equipmentProfessional, scientific and controlling instruments; photographic and optical goods, watches

and clocksMiscellaneous manufacturing industries

184

189


V. Classification of 1970 membership: (Continued)

3. Indicate industry composition of union membership

Nonmanufacturing: :

Mining and (including crude petroleum and natural gas production) %quarryingContract construction (building and special trade)Transportation services (including railroads, air, bus, truck and water transportation, and allied

services)

%

%

Telephone and telegraph %

Electric, gas, and sanitary services (including water) %

Rimiest& and retail trade%

Finance, iqsuraoce and real esiateService industries (including hotels, laundries and other personal services, repair services,

motion pictures, amusements and related services, hospitals, educational institutions, non-membership organizations) %profit

Agriculture and fishing %Nonmanufacturing (classification not available) %

Government:FederalState and local

%

%Total 100 %

4. State distribution of union membership. Indicate the approximate number or percentage of membersin each of the 50 States.

tl.lsbamaiaainryleandDistrict

OklahomaMaryland District Oregon

Arizona of Columbia PennsylvaniaArkansas Massachusetts Rhode IslandCalifornia Michigan South CarolinaColorado - Minnesota South DakotaConnecticut Mississippi TennesseeDelaware Missouri TexasFlorida MontanaGeorgia Nebraska VermontHawaii Nevada VirginiaIdaho New Hampshire WashingtonIllinois New Jersey West VirginiaIndiana New Mexico ----- Wisconsin----lowa ---- New Yodc WyomingKansas North Carolina No. percentageNo erKentuckyLouisiana

North Dakota accounted for inOhio the U.S.

5. For any area outside the United Stow% please indicate the number of dues-paying members and thenumber of local unions in existenie as of the end of 1970 or any other appropriate current period:

Number of local unionsLocation Approximate number of union members

Canada

Puerto Rico

Canal Zone

Other (specify)

UNION MEMBERSHIP

VI. Tern of Officers:

1: Are officers required to retire at a specified age?

National officers: Yes No;. If yes, age

Local officers: Yes No; If yes. age

2. Are the number of terms an officer may serve limited?

National officers: Yes No; If yes, number of term.

Local officers: Yes No; If yea, number of terms

VII. Collective Bargaining Agreements:

1. Indicate the number of basic collective bargaining agreements with employers. Exclude supplemental and pension,health and insurance agreements:

Total agreements.United States agrtsments.

2. (a) Indicate number of differwit employers covered by collective bargainingagreements employers

United States only- employers

(b) If more than one employer, are the cmploycra located in at least twoStates? Yes 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 thaw

Expired in 1970 agreements.Will expire in 1971 agreements.

S. Indicate the number of agreements in the United States that provid'e wage reopener in:

1970 agreements.

1971 agreements.

VIII. Union Staff:

Indicate the number of full-time employees on the payroll of the national in the United States. Exclude electedofficials and employees on the payrolls of local unions and intermediate bodies:

Managers and administrators

Professionals

Clerical and secretarial

Organizers and representatives

Others (speCify)

A '

May we have your commentstreerding the present Directory and proposals for changes in future editions?

Name of person reporting Title Date

191


entire universe of national and internationalunions, as defined.

Presentation

The data for each union and summaries arepresented in the Directory and in articles in theMonthly Labor Review. Wherever appropriate,separate data are presented for AFLCIOaffiliates and unaffiliated unions and cover ques-tionnaire items such as total membership forthe last 2 years, members in and outside theUnited States, size of unions, women andwhite-collar members, and industry distribu-tion. Gains and losses over the past decade areanalyzed, and major unions are ranked accord-ing to size at particular points in time. Of con-siderable interest to users of. the Directory arethe changes of union membership related tochanges in the total labor force and to em-ployees in nonagricultural establishments. Databy industry- and on women and white-collarmembers permit rough approkimations on the

- extent of union penetration compared withearlier periods.

In addition to statistical summaries, theDirectory also contains appendixes which list,for each union, membership in areas outsidethe United States, number of *Omen, propor-tion of white-collar members, and proportionof members in major industry groups.


The Bureau's membership figures are pub-lished on a regular basis, and serve as theprincipal indicator of gains and losses for par-ticular unions and for the labor movement asa whole, particularly as a measure of inroadsor declines in industry and occupational groups.Comparisons of labor relations policies andtheir economic effects can be made betweenorganized and unorganized sectors. They areused by agencies of the Federal Government,State and locario.irernments, by managementpersonnel, union officials, students of the labormovement and economic affairs, and the gen-eral public. The Directory also is distributedwidely abroad, notably to international organi-zations and labor unions.

Difficulties in measuring union membershiparise from (1) the variety of concepts andpractices among unions as to the definition andreporting of membership, (2) the availabilityat union headquarters of the various data re-quested, and (3) the willingness of the unionsto make those data available to outsiders.

In an attempt to achieve uniform reportingpractices, the Bureau asks unions to report onthe annual average number of dues-payingmembers. Although a worker when joining aunion assumes an obligation to pay dues, uni-form reporting practices do not result fromapplying this criterion alone. Unions defineeligibility for membership in a variety of waysand payment of dues is only one of severalcriteria. Some unions set less than full duesrequirements or waive the payment for work-ers who are unemployed or on strike. Suchexonerated Workers, however, usually remainmembers in good standing, with the samerights as full dues-paying members. Similarqualifications may apply to members who areapprentices, retired, or in the Armed Forces.3-

In an attempt to determine union practicesin reporting membership, the Bureau repeat-edly has requested unions to indicate whetherthey include or' from membership re-ports five specified groups: the unemployed;those involved on work stoppages ; those in theArmed FoiNes; apprentices; and the retired.Moreover, unions were asked to furnish an

3 In its 1963 Dim:tory, the Bureau took a closer look at therelationship between dues submitted to the internationalunion, the so-called per capita tax, and reported membershiptotals for particular unions. Per capita tax receipts weredivided 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 forreasons which the unions explained in detail. Some unionsinclude a large number of seasonal employees whose duespayments are limited to several months during the year.Other unions questioned the use of a-computed membershipfigure where for large groups of workers (sick, unemployed.those promoted out of the bargaining unit, etc.) only apartial per capita 'tax is paid, at times less than one-tenth ofthe amount required4of other members. A few unions indi-cated categories of members completely dues exempt (e.g.,life members, 50-year members, etc.). Financial obligationsalso frequently are waived for workers recently organizedand for local unions in economic difficulties. In addition,several unions set dues on a sliding scale proportionate tothe income of members, a method which rules out computa-tions of this sort. Thus, on examination it became clear thatcomputing membership by dividing the tax rate into taxreceipts could not be used as a uniform yardstick applicablefor all unions.

1E7

UNION MEMBERSHIP 193

estimated or actual figure on the number ofMembers in "excluded" categories. Ideally, ifall unions could furnish such data, it would bepossible to compute the total number of work-ers who, at least in some way, are still attachedto unions. The responses, however, have fallenfar short of this goal. Only a small number ofunions reported, in whole or in part, the prac-tices they followed. An even smaller number ofunions were able to furnish figures on the num-ber of workers involved?

Thus, no uniform answer exists to the de-ceptively simple question : Who is a unionmember? The answer varies from union tounion, as determined by its own policies andpractices.

Although financial statements may be of

At various times, suggestions have be* made on the useof alternative sources for membership ii,,ormation, notablythe use of "voting strength" of international unions at AFL-.CIO conventions, based upon average paid membership tothe Federation. This, however, is not a reliable measure.Some unilns, for reasons of prestige, tend to overpay, whileothers, as a matter of convenience. pay an arbitrary amount.Moreover, this method could not be used for unions outsidethe AFLCIO.

some help in arriving at membership approxi-mations, they cannot be used to obtain data onvarious components of union membership, suchas the proportion of white-collar and womenmembers, those under contract in particularindustries, etc. For such data, the Bureau mustrely entirely on the cooperation of nationalunions. These, however, are not always able tofurnish the information for the simple reasonthat it is not compiled at union headquarters.Many national unions are decentralized or-ganizations' and as a rule, do not seek reportson these items from their local unions. Also,some unions in possession of such informationmay be reluctant to make it public.

Data presently compiled are submitted byofficial3 of national and international unions.PerhapS more accurate data, and certainlyMore detailed data, could be obtained by con-tacting local unions, as is done in Canada, butthe task of soliciting responses from more than70,000 organizations is beyond the present re-sources of the Bureau.

Technical. ReferencesNumber1. 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. Includei a discussion of sources and methods of measurement.2. U.S. Department of Labor, Bureau of Labor Statistics. "Membership of Labor Unions, 1897-

1950." Handbook of Labor Statistics, 1950 ed., (Bulletin 1016, 1951), pp. 137-139.3 Directories of Labor Unions in the United States; (Bulletins 937, 1948; 980, 1950; and

1127, 1953).4. _ Directories of National and International Labor Unions in the United States (Bulletins

1185,1.955; 1222, 1957; 1267, 1960; 1320, 1962; 1395, 1964; 1493, 1966; 1596, 1968; and1665, 1970). In various articles in Monthly Labor Review. (Excerpted)

In addition to membership data, the Directories also include a brief description of themethods used, and a copy of the questionnaire.

5. Paschell, William, "Limitations of Union Membership Data," Monthly Labor Review, Novem-ber 1955, pp. 1265-1269.

6. Wolman, Leo.-Ebb and Flow in Trade Unionis»t, New York, National Bureau of EconomicResearch (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 dataon the extent of organization by industry.

7. 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.

188

HARRY P. COHANY (

Chapter 22. Annual Earnings and Employment Patternsof Private Nonagricultural Workers

Background

The purchasing power of workers dependslargely on their annual earnings which aredetermined by the interaction of variables suchas straight-time rates of pay, number of hoursworked, and the hours worked at premiumrates which depend on other variables like occu-pation, union status, industry, and area. More-over, some workers move into and out of thelabor force during the year. In addition, asubstantial portion work for more than oneemployer in the same industry and for one ormore employers in different industries. Con-sequently, occupational wages or hourly andweekly earnings data cannot be extrapolated toannual earnings estimates with any degree ofprecision.

The Bureau's program of annual earningsand employment patterns studies, initiated inthe late 1960's, is designed to fill the gap inour knowledge of annual wage and salaryearnings from private nonagricultural employ-ment. The first study' for 1964, was limited towage and salary earnings covered by socialsecurity; the second,2 for 1965, included dataon wage and salary earnings covered undereither the Social Security Act or the RailroadRetirement Act. Subsequent studies, like thatfor 1965, will include information on earningscovered under either social security or railroadretirement. Bulletins providing 2 years of dataare expected to be issued biennially. Specialanalytical studies will be published intermit-tently in the. Monthly Labor Review.

Description of the Series

The series covers earnings and employmentpatterns in the private, nonagricultural sectorwhich, broadly defined, includes individuals,who work for wages and salaries in employ-ment covered by the Social Security Act andthe Railroad Retirement Act.3 Excluded areearnings in agriculture, self-employment, andin government units other than those that par

ticipate in social security and function likeprivate firms, such as hospitals and schools.

The studies provide distributions of medianand mean earnings for the private nonagri-cultural sector a§ a whole, for each industry di-vision, for each major industry group at thetwo-digit SIC level, and for selected industrygroups at the three-digit SIC level of industryclassification. The data, though available onlyfor white-collar and blue-collar workers com-bined, are unique, because unlike annual earn-ings data from other sources, they permit ananalysis of the distribution of wage and salaryearnings and employment patterns by industryand quarters of employment.

The studies focus separately on earnings inthe industry in which workers had greaterearnings than in any other, and earnings in allindustries. Some of these data and some -ofthe employment patterns data, separately andwith earnings data, also are presented forselected demographic characteristics.

Earnings and employment pat rns ofworkers who had covered wage an alaryeach quarter of the year are emphasize par-ticularly. Information about earnings a d em-ployment patterns of four quarter w9 kers isthe closest to data for workers fOf attachedto the private sector work):Irer that can beobtained from the source materials. However,some workers who have earnings in each quar-ter are attached to the work force to only alimited extent.

Source of Data

The data are developed by the, Bureau ofLabor Statistics from statistical information

Annual Earnings, and Employment Patterns, Private Non-.'agricultural Employment, 1964 (BLS Report 330 1969).

Annual' Earnings and Employment Patterns of PrivateNonapricultural,EMployees, 1985 (BLS Bulletin 1675, 1970).

'`discuthion of the interrelationship of, and joint} coverage under, the rallniad retirement and social security

systems, sce U.S. Department of Health, Education, and Wel-fare, Simla] Security Administration, Social Sectoity Hand-book, 4th ed. (1969), and U.S. Railroad Retirement Board,Handbook on Railroad Retirement and Unemployment Insur-

- once Systems (1969),

1199


obtained from the Social Security Administra-tion and the Railroad Retirement Board. Topreserve the 'confidentiality of the records, thedata are provided to the Bureau of LaborStatistics without identification of individualsor employers. However, to combine data fromvarious employments and to facilitate 'statis-)tiscal processing, each individual and employeris assigned a permanent control number, differ-ent from his social security or employer identi-fication number.

Each individual in the sample providesdemographic information (race, sex, and yearof birth) when he applies for a social securitynumber. Each employer under social securityfrom whom the individual receive any wagesor salaries during a calendar quarter reportsthe 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 afterthe worker has reached his taxable earningslimit in that employment situation. Employersreport wages paid to the maximum annuallimit under social security and to the maximummonthly limit under the railroad retirementsystem. Employer reports also indicate theindustry and. except for employment coveredby the Railroad Retirement Act (RRA), thearea in which the wages or salaries wereearned. Employers subject to the RRA, alsoprovide information about the occupationalcategory the worker was employed in.

Sample Design, and Sampling andNonsampling Variability

The sample, which includes 1 percent of allsocial security numbers, was selected on thebasis of a multistage systematic cluster sam-pling procedure. Social security numbers areused as the individual's identification numberin both the social security and railroad retire-ment systems. An individual selected for thesample remains in it permanently.'

Since estimates in this study are based on asample, they may differ from census figures.Mores. ar, the sample data are not adjusted tobenchmark levels established by complete

counts. In addition, the data are subject tononsampling variability due to errors in re-porting and classification and other possibleerror sources, that would be present in a com-plete enumeration as well as in a sample. Asa result, ratios established from the sampledata are considered to be reasonable estimatesof those existing within the population as awhole. Nevertheless, particular care should beexercised in interpreting medians and percentsbased on relatively small numbers of cases aswell as small differences between figures.' Ab-solutes are subjct to large error which mayvary from year to year.


To estimate total wages of individuals, theSocial Security Administration determines thequarter in which the taxable limit is reached("limit quarter"). Wages in-the prior quarterequal to or greater than the "limit quarter"wages are substituted for those in the "limitquarter" and in all subsequent quarters. Limitquarter earnings, however, are used to estimateearnings in the limit and subsequent quarters iflimit quarter earnings were higher than earn-ings in previous quarters. The summation ofthe quarterly wages after these substitutionsthen becomes the estimated annual total. Whenthe taxable limit is reached in the first quarter,the Social Security Administration imputes anestimated total.

Employers covered by the Railroad Retire-ment Act are required to provide informationabout the monthly earnings of each employee upto the maximum creditable limit subject toRailroad. Retirement Act taxes. Hence, evenearnings reported at the maximum level foreach month aggregated to annual totals may be

4For.a detailed discussion of the sampling procedure, re-porting criteria, and coverage under the social security andrailroad retirement systems, see U.S. Department of Healthi;Education, and Welfare, Social Security Administration,Workers Under Social Security, 1960 (1968) and Social Se-curity Handbook, op. cit.; also see Handbook on RailroadRetirement and Unemployment Insurance Systems, op. cit.

z, For an indication of the order of magnitude of the sam-pling errors and a fuller discussion of the sampling andnon sampling variability to which the series is subject, seeWorkers Under Social Security, 1960, op. cit.

120

ANNUAL EARNINGS & EMPLOYMENT PATTERNS

substantially below the worker's total earnings.The Railroad Retirement. Board, however, col-lects information from employer records aboutthe total annual earnings of a sample ofworkers covered'by the act. Factors for raisingcreditable compensation under the Railroad Re-tirement Act to total railroad earnings arederived by comparing the total earnings datafor individuals collected in the special studywith the aggregated monthly earnings data forthe same individuals. The incremental factorsfor workers in the same broad occupationalcategories then are averaged. The resultingfactors, developed by the Railroad RetirementBoard, are applied by the Bureau of LaborStatistics to the credited monthly earnings ofeach individual by taking into account hisoccupational category.°


Distributions of workers by annual earnings,and mean and median earnings are presentedfor all major earners and for those with fourquarters of earnings. In addition, data areprovided showing the number of workers whohad wage and salary earnings in the variousindustries and the proportion of these who hadgreater (major) eArnings in that industry thanin any other industry. Data on major earnersby industry showing The proportion with earn-ings in each. quarter, their demographic char-acteristics, what proportion had earningswithin a single industry and the number ofemployers whdm they worked for also arepresented. Some earnings and employment pat-terns data are presented for broad geographicareas.

Terms used in the series and the methodsused to classify workers by industry and regionof major earnings and the industrial classifi-cation scheme used in classifying nonpolicygovernmental 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, inemployment covered under the Social SecurityAct or the Railroad Retirement Act. Such pay-ments may be cash, cash equivalents, or other

OF PRIVATE NONAGRICULTURAL WORKERS 197

media such as goods, clothing, board, orlodging.

Self-employment earnings, payment for workin employment excluded from the coverage ofthe acts, and payment for work in agriculture,in covered governmental units engaged inpublic administration and for military servicehave been excluded from this study. Most pay-ments by employers to or on behalf of em-ployees, or for employees and their dependentsfor retirement, death, sickness or accidentaldisability, or medical and hospitalization ex-pense under provisions of a plan or systemmeeting certain general criteria, and employerpayments to a trust fund, such as a pensiontrust, exempt from tax under the InternalRevenue Code, are not counted as wages in thisseries.?Workers with some earnings in the industry.Each individual who earned $1 or more in anindustry during the year is counted in eachindustry in which he had any earnings. Aworker who had some earnings in each of fivethree-digit industries, as defined in the Stand-ard Industrial Classification Manual 8 for ex-ample, is counted for each of these industriesas well as in each two-digit industry and ineach division of which the three-digit industresare a part. Because a worker is counted in eachthree-digit industry, each twodigit industryand in each division in which he had $1 in cov-ered wage and salary earnings or more, theaggregate count at each level is greater thanthe total number of workers at each broaderindustry level (two-digit, division, private non-agricultural economy).

"Although the Railroad Retirement Board collects annualearnings data for a 1-percent sample of workers, the samplingcriteria differ from those used by the Social Security Ad-ministration. To permit the combination and integration ofdata from the two systems, the Railroad Retirement Boardprovides the Bureau of Labor Statistics with data for asample of workers selected according to the sample designestablished by the Social Security Administration.

7 Under certain circumstances tips are counted as wages forsocial security purposes and thus are included in the datafile used in this study series. Payments to workers from taxexempt trust funds are not considered wages (except forwages paid by the fund to its employees) and thus areexcluded. Employer payments to trust funds which are nottax exempt also are excluded. Depending on their natureand purpose, payments to workers from these fends maybe counted for social security purposes and thus are inlludedin the data file.Such would be the case if the payment wasa bonus or vacation payment. For a ,more detailed discussionof covered wages, see the Social, Spettrity Handbook, op. cit.

h Issued by the U.S. Office of Management and Budget.

191


Major earners and ustry of major earn-ings. A major earner in this study, is a workerwho earns more of his annual wages andsalaries in a specific industry than in any otherindustry. This test to determine the worker'sindustry of major earnings is based on aplurality earnings concept and is applied sepa-rately at each industry level.

Data for major earners are included in onlyone industry at each level in industry classifi-cation (i.e., at the three-digit, the two-digit,and at the divisional levels). Thus, the numberof workers who received the major proportionof their earnings in each industry, at any levelof industry classification,. is unduplicated.Hence, data for any level of classification atwhich all component industries are shown addto the total for the private nonagriculturalsector. However, data at the three-digit level(if all three-digit industries were shown)might not add to the detail at the two-digitlevel, and data at the two-digit level might notadd to the divisional level. Except for totals atthe private nonagricultural sector, this non-additive relationship exists because someworkers shift employment several times duringthe year and bring about. .;patterns like thatshown in the following tabulation of a hypo-thetical worker's employment and quarterlyearnings experience.

Private non-agricultural

Total(anyquar-ter)

Jan-,uuryto

March

Aprilto

June

Julyto -

Septem-ber

Octtoober

Decem-ber

economy ___ $530 $80 $130 $150 $170

Division A 150 60 902-digit group,

A-1 150 60 903-digit group,

A-11 150 50 90

Division B 180 60 50 702-digit group,

B-1 180 60 50 703-digit group,

B-11 130 10 50 703-digit group,

B-12 50 50

Division C 200 20 20 60 1002-digit group,

C-1 100 20 20 603-digit group,

C-11 100 20 20 602-digit group,

C-2 100 1003-digit group,

C-21 100 100

In this tabulation, the worker had greaterearnings at the three-digit level in industryA-11 than in any other three-digit industry; atthe two-digit industry he had greater earningsin industry 13-1 than in either A-1, C-1, orC-2; and at the divisional level he earned morein division C than in either A or B. Therefore,applying the plurality earnings test, thisworker is assigned to industry A-11 at thethree-digit level, B-1 at the two-digit level, andto C at the division level.°

Quarters of Work 10

In this series, because workers are classifiedboth by quarters worked 10 in their major in-dustry and in all wage and salary employment,quarters of work of the hypothetical workerwhose employment and earnings were illus-trated prevjously are as follows:

Majorindustry

All wage andsalary employment

Private nonagricultural 4 4Division r' 4 42-digit group, B-1 3 43-digit group, A-11 _ 2 4

Employer. An employer in this study is anindividual, partnership, or corporation recog-nized under the law as a separate entity meet-ing certain criteria." However, a firm whichseparately incorporates at each of its locationsMay be considered a separate employer at eachlocation. Thus, a worker transferred from oneto another location that is incorporated sepa-rately may have more than one employer inthe same year even though- he continued towork for the same firm.Industrial classification. Employment andearnings data based on the Social Security Ad-

*An unpublished tabulation, prepared by the Social SecurityAdministration, indicated the industry of major earnings atthe three-digit level for about 12 percent of all coveredworkers was part of a two-digit industry different from theworkers two-digit industry of major earnings.

10 "Quarters of work" and "quarters of coverage" are ncasynonymous. A quarter of coverage based on covered non-agricultural employment is a calendar quarter in which theindividual was paid $50 or more: a quarter of coverage alsomay be credited on other bases such as agricultural employ-ment or self-employment.

nFor details, see the Social Security Handbook, op. cit. Inaddition, because some workers work for more than oneemployer during the same week, data showing number ofemployers should be used with caution.

ANNUAL EARNINGS & EMPLOYMENT PATTERNS OF PRIVATE NONAGRICULTURAL WORKERS 199

ministration's (SSA) data file are classifiedaccording to the Administration's industrialclassification system. This system differsslightly from that published in the StandardIndustrial Classification Manual (SIC), andused in most statistical series in the assignmentof industry codes to governmental units. Moststatistical series classify governmental unitsinto SIC Division IGovernment, 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, allemployment 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 theact, however, were treated as service industryemployment, because the units were classifiedby SSA into SIC 82, educational services."'

Employment and earnings data based on theRailroad Retirement Board's file are classifiedinto the following industries as defined in theSIC Manual: Railroads, SIC 401; sleeping carcompanies, SIC 402; express companies, SIC404; rental of rail cars companies, SIC 474;and other companies performing services rail-road transportation and certain railway labororganizations, SIC 861 and 863. In each casethe assignment is based on the industrial classi-fication of the worker's last employer, who wascovered under the Railroad Retirement Act.Data for workers who had earnings in morethan one industry, all of which were covered

12 For detailed information, see U.S. Department of Health,Education, and Welfare, Social Security Administration, Com-parison of Social Security Administration and Standard In-dustrial Classification Systems, 1963 (undated) and the guideprepared by the Administration in 1968, entitled "IndustrialCodes in the Social Security Administration ContinuousWork History Sample (CWHS), Data for 1957 through 1908."

1, For all industries, except water transportation, employ-ment outside the 50 states or the District of Columbia is onlya small fraction, if any, of the total number of the industry'smajor earners.

',Employment under the Railroad Retirement Act maces upnearly all the major earners in the railroad industry (SIC 40)and a substantial proportion of those at the all transportationlevel, but only a very small fraction of the total employmentin other industries or at the total private nonagriculturallevel. Thus, the convention adopted has a serious effect onthe regional employment distributions and regional earningspicture at the all transportation level but little effect onother industries or at the total private nonagricultural level.

under the social security systeM or one of whichwas covered under the railroad retirementsystem, are classified and presented separatelyand in combination.

Single and multi-industry workersat eachlevel of industry classification the employmentexperience of each sample member was ex-amined to see if all of his earnings were inone or more than one industry. Those withearnings in more than one industry were classi-fied as multi-industry workers. This conceptualapproach may be seen in the case of a workerwho was employed, as illustrated below, by anemployer in each of two three-digit IndNstrieswithin the same two-digit industry.

Industry level Number of employers

Division A 2

2-digit group, AV 2

3-digit group, A-.11 1

3-digit group, A-12 . 1

Region of major earnings. The region in whichworkers had their major earnings is deter-mined, by a plurality earnings test similar tothat described in detail in "Major earners andindustry of major earnings." The region inwhich he had greater earnings than any otheris his region of major earnings.

In a few cases, earnings in the industry andregion of major earnings may not coincide.All data for major earnings are classified firstby industry and then by region. A worker whoearned 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 30percent in industry C also in the South wouldbe classified as a major earner in industry Aand as having had his major earnings in theSouth.

In this series, workers employed under theSocial Security Act or the Railroad RetirementAct are divided into five regions : Four coverthe 50 States and the District of Columbia ; thefifth includes all employment in U.S. territories,on foreign soil, or aboard ocean-going vessels."

Data in this series do not indicate wherewages and salaries covered by the Railroad Re-tirement Act were earned." Therefore, a con-vention was adopted ascribing all such em-ployment to the North Central region where

193


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 exceptNegroes. This convention, which is differentfrom that used in most statistical series, wasadopted to minimize the effect on those groupsfor whom the sample was not sufficiently largeto present data separately and to maximize theanalytical usefulness of the data.


The data provide an insight into the answerto the question : "How well do private non-agricultural workers do in their industry ofmajor earnings and to what extent do theysupplement these earnings by employment inother private sector industries." Data are usedin collective bargaining; in formulating publicpolicy 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 racesand sexes.

The data, however, have several substantiallimitations. Some workers whose annual earn-ings are included in the series also haveearnings in self-employment, agriculture, oremployment in governmental units excludedfrom the series. As a result, these workersappear to have low annual earnings. Probablymost of these workers were attached to theemployed work force only a very limited extent.The unavailability of data on hours or weeksworked or paid for, or occupational group(other than in railroading) seriously limits theanalytic potential of the series. The S-5 yearslag between the reference period and the dateof publication introduces another limitingfactor. These delays result primarily from re-porting requirements established by the law,the administrative requirements of the collect-ing agencies, and the processing required toreduce the mass of micro data into statisticalsummaries. Nevertheless, since employment

patterns usually change very slowly and rela-tive earnings distributions generally are quitecomparable from year to year even though thelevel of earnings moves upward, the relation-ships shown are indicators of the currentsituation.

Nothwithstanding the limitationsi data fromthis series, unlike those from other sources,f5permit an analysis of the distribution of wageand salary earnings and employment patternsof workers by demographic characteristics, in-dustry, and quarters of employment. Thus, theyare uniquely useful to all concerned with theannual wage and salary income of individualsand the employment patterns of those in theprivate nonagricultural work force.

15 Dissimilarities in concept or method between the BLSannual earnings and employment pattern series and otherseries may result in important differences in sampling andnonsampling variability between series. Therefore, cautionmust be exercised in using data from the BLS annual earn-ings series in conjunction with data from other statisticalseries. 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 notedbelow.

The Social Security Administration (SSA) publishes someannual earnings data by industry. Their most recent reportWorkers 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 industryattachment concept used by the SSA, however, is differentfrom that used in this study. Further, the Social Securitydata do not include earnings in employment covered by theRailroad Retirement Act:-..

The Railroad Retirement Board (RRB) annually publishesa "research and statistics note" which provides informationabout the total railroad earnings of railroad employees. TheRRB data, however, exclude earnings in employment notcovered by the Railroad Retirement Act.

Some annual earnings data at the all-industry level byoccupational group are published by the Bureau of theCensus, U.S. Department of Commerce, in "Consumer In-come," Series P-60 of the Current Population Reports. Thispublication also provides a distribution of wage and salaryearnings, 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-digitSIC) groups. The study, based on a' household survey, doesnot provide distributions of wage and salary earnings byindustry group and has different concepts of industry attach-ment from thov.: used in this study.

The Office of Business Economics of the U.S. Departmentof Commerce also publishes estimates of the average annualearnings of "full-time employees" in its Survey of CurrentBusiness; these estimates do not reflect the effect on averageearnings of workers who work less than a full year.

-ARNOLD STRASSER

Chapter 23. Measuring Collective Bargaining Settlements

Background

The Bureau's program of measuring theeffects of collective bargaining settlements onhourly labor compensation is a reflection oftwo developments : One, the growing impor-tance of fringe benefits as a proportion ofemployee compensation, and two, increasedconcern about the effects of collectively bar-gained wage and benefit changes on the pricelevel.

Whereas in earlier years the economic termsof negotiated settlements could be equatedlargely with agreed-upon changes in wagerates, today, possible changes in a host of paysupplements must be consideredsuch as vari-ous forms of premium pay, paid leave, cashbonuses, and employer contributions to fundsproviding pension or health and welfare bene-fits. Although sieztie.ht-time pay for workinghours is still the major element of compensa-tion, supplements are now a significant portion,accounting for about a fifth of total employeroutlays for worker compensation.1

Moreover. growing, concern during the 1960'sover the extent to which increased labor costs

1See Alvin Bauman, "Measuring Employee Compensationin U.S. Industry," Monthly Labor Review, October 1970, p.23. It is difficult both to measure the growth of supplementsover the years and to quantify their current importance.The national income accounts provide one pertinent sourceof data. They show that supplements to wages and salariesrose as a percentage of total employee compensation from1 percent in 1929 to 10 percent in 1970. The National Incomeand Product Accounts of the United States, 1929-1965: Sta-tistical. Tables, U.S. Department of Commerce, Office of Busi-ness Economics (1986), p. 14; Survey of Current Business,January 1971, p. 10. These figures, however, do not revealthe relative importance of all supplements in either year,since manysuch vs premium payments, leave payments, andcash bonusesappear as parts of wages and salaries. Thefigures therefore are not comparable to those found in sur-veys of employer expenditures for supplementary compen-sation.

2Publication of first-year changes is a recent innovation.Before 1970, the Bureau published two measures of changeover the life of the contract, the so-called equal timing andthe time-weighted measures. The former assumed equalspacing of changes during the term of the contract; the lattertook account of the actual effective dates of wage and benefitchanges. The time-weighted measure has been discontinued,because it appeared to be of significance primarily forthe analysis of individual settlements rather than for overallseries of the type produced by the Bureau. Moreover, drop-ping of the time-weighted measure and introduction of aseries on first-year changes provides parallel statistics bothon wage-rate changes alone and on wages and benefitscombined.

may contribute to inflation has heightened in-terest in the size of collective bargainingsettlements.

Responding to these influences, the BLS be-gan estimating the cost of wage and benefit(i.e., "package") changes in a limited numberof key settlements in 1964. The work was ex-panded the following year and, since 1966, theBureau has attempted to determine the priceof all settlements affecting 5,000 workers ormore in the private nonfarm sector. In addi-tion, a separate series has been developed forthe construction industry, covering settlementsfor 1,000 workers or more.

Description of Series

At present, the Bureau publishes two sets ofdata on wage-benefit decisions. One shows theannual rates of increases in settlementsreached in a given time period and scheduled togo into effect at any time during the term ofthe agreements. The other is limited to thechanges set for the first 12 months of theagreements.2 Published data summarize settle-ments reached during individual quarters of ayear, during full years, and during the first 6and 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 bya given action are entered at the average forthe bargaining unit. The sums of the individualsettlements are averagedboth means andmedians are presentedeach settlement beingweighted by the number of workers affected.However, the pricing of individual settlementsis not disclosed.

Averages for full years are available sep-arately for manufacturing and nonmanufactur-ing industries. Otherwise, no industry detail ispublished, except for the separate constructionindustry series.

As indicated in the preceding paragraphs,these series relate to the pricing of decisions,i.e., they measure the effect of changes agreedon in a given period although, considering thegeneral practices of negotiating multiyear

201


collective bargaining agreements containingprovisions for annual (and sometimes morefrequent) improvements, the changes may beintroduced only at a subsequent date. Work hasrecently been completed on the development ofdata on the wage and benefit changes actuallyplaced in effect in specified periods, whetheras a result of current bargains, changes agreedupon earlier ,but with deferred effective dates,or the operation of cost-of-living wage es-calator clauses.


The terms of the settlements to be priced areobtained primarily from secondary sources,such as general circulation newspapers andperiodicals and union, management, and tradepublications. Collective bargaining agreementsand documents on pension and health and wel-fere plans also are consulted. When thesesources are inadequate, direct requests for in-formation are made to the companies andunions involved.

Large quantities of statistical data, as wellas the settlement terms, are required. Theseare needed both to determine existing em-ployer outlays and to assay the effect on theseexpenditures of agreed wage and benefitchanges. Efforts are made to use existing data.However, when these prove inadequate, theparties are requested to furnish data. Such re-quests, it must be emphasized, are made toreceive specific information from which theBureau can price settlements ; the requests arenot made to receive the parties' own evalua-tions of the terms of their settlements.

One of the major sources of information oncurrent hourly earnings is the establishmentinformation which BLS obtains through itsmonthly employment, payroll, and hours sur-vey. Information on current outlays for paysupplements may be available from BLS sur-veys of expenditures for such benefits. Perti-nent information for estimating expendituresfor some items may be found in industry wagesurveys, e.g., extent of late-shift work andoccupational employment distributions. Annual

iSG

financial reports filed with the Department ofLabor under- the Welfare and Pension PlanDisclosure Act provide useful material. Not allthe sources tapped are governmental; for in-stance, information on workmen's compensa-tion insurance rates is reported by the NationalCouncil on Compensation Insurance.


As was noted earlier, the Bureau attempts tocover all settlements for 5,000 workers or more(1,000 or more in construction) in the privatenonfarm sector. Substantially all such settle-ments come to the attention of the Bureau andare included in its series, it is believed.

Discussion of procedures for pricing individ-ual settlements 3 may be centered around threequestions: (1) What items in a collective bar-gaining settlement are to be priced? (2) Howare 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-managementagreement, not merely the wage and benefitprovisions, may affect directly or indirectly anemployer's costs. For example, one of the so-called "noneconomic" terms of an agreementsenioritymay have a limited influence on coststhrougn its effect on employee efficiency.' How-ever, such an item is essentially not meas-urable.

Consequently, the BLS program is confinedto the wage-benefit component of collectivebargaining agreements, i.e., to the effect ofsettlements on employer outlays for employeecompensation. This component, clearly, is ofmajor significance in its own right. Includedare items such as changes in wage rates ; modi-

3Production of statistical series merely entails groupingand averaging the pricing of individual settlements. Publisheddistributions show the number of workers under knownsettlements for which data were insufficient to permit pricing.

Almost inevitably a tendency has developed to regard allcontract changes as serving to increase employer payments.Although this undoubtedly is the common result, somechanges, even when proposed by unions, may lower costs.Union-management cooperation schemes have at times pro-vided examples.

1

MEASURING COLLECTIVE BARGAINING SETTLEMENTS

fications in premium pay, bonuses, paid leave,and severance pay; and adjustments in em-ployer payments for pension, for health andwelfare, and for supplemental unemploymentbenefits, excluding the costs of administeringthese benefits. Also included are changes informal contract provisions specifying paidtime for clothes change, washup, and lunchperiods.

Excluded are informal modifications of un-written rest-period practices ; items related to,but not normally considered part of, compensa-tionsuch as per diem payments, moving ex-pense reimbursements, and payments for safetyclothing; and provision of facilities or servicessuch as parking lots and health units, the costsof which often are charged to capital ratherthan labor accounts.

Determination of costs. Since a value is placedon settlements at the time they are, reached,the costs attributed to them obviously are esti-mates of outlays to be made in the future; theycannot be taken from employers' accountingrecords. The estimates are made on the assump-tion that conditions existing at the time thecontract is negotiated will not change. For ex-ample, estimators assume that methods offinancing pensions will not change, and thatexpenditures for insurance will not change ex-cept as a result of altered benefit provisions ormodified participation because of changes incompany contributions. They also assume thatthe composition of the labor force will notchange.

In this regard, except for any guaranteedincreases, which are treated as deferred adjust-ments, possible wage-rate changes as a resultof cost-of-living escalator clauses are excludedbecause of difficulties in predicting movementsof the Consumer Price Index for 3 yearsthetime span of many collective bargaining agree-ments. Thus, the Bureau prices the wage and

a The series on wage and benefit changes actually placedinto effect includes escalator adjustments, since the issue ofprediction of CPI movements does not arise.

The terms "roll-up" and "bulge," among others, also aleused to express the same idea.

7 It varies from this ratio to the extent that there are limitson earnings that are subject to social security taxes.

203

benefit changes that would go into effect if theprice level were to remain stable.6

Nevertheless, package estimates do attemptto measure the costs associated with actualcharacteristics of the work force affected bythe 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 involvedrecognize that the choice in incorporatingalternative benefit changes into contracts isaffected by their costs, which, in turn, areaffected by the character of the work force.For example, an extra week of vacation after15 years of service will cost very little whenonly 10 percent of the workers have that muchservice, but will cost about 1 percent ofstraight-time hourly earnings when half of theworkers have been employed for 15 years ormore.

As a rule, indirect effects of settlements areignored : factors such as possible extension ofsettlement terms to nonunion workers in thesame firm or to members of other bargainingunits. Similarly, the cost of providing length-ened vacations is measured by the wages andsalaries paid for the additional time off; costsof hiring vacation replacements, if necessary,are not considered. Moreover, effects on unitlabor costs, which involve consideration of em-ployee efficiency as well as employer payments,are disregarded.

However, "creep" 6 is accounted for. Creepreflects the fact that an increase in wage rateswill have a secondary effect on employer coststhrough its effect on outlays for benefits. A20-cent-an-hour wage increase will affect notonly straight-time wage rates but also supple-mentary Myments governed by wage ratessuch as ovArtime premiums, leave payments,pension benefits related to salary level, andsocial security payroll taxes (if earnings are notat or above the statutory maximum tax base).Creep is taken into account by raising eachwage increase provided by the new contract bya creep, or loading, factor., This factor is es-sentially the ratio of current hourly- expendi-tures on benefits that vary with wage levels tocurrent straight-time hourly earnings.?

157


Many items in a collective bargaining agree-ment are priced without difficulty. This isparticularly true when settlement terms areexpressed as cents-per-hour adjustments, e.g.,a 20-cent-an-hour general wage increases or a5-cent increase in employer contributions to ahealth and welfare fund. These stipulatedcents-per-hour figures are utilized as the costsof the settlement provisions.° Percentage wageadjustments are converted to cents-per-hourfigures on the basis of current average straight-time hourly earnings. Although less direct, thecost of an additional holiday is estimated ade-quately by prorating 8 hours' average pay (ifthe normal workday is 8 hours) over the num-ber of annual working hours per employee.The cost of an additional week of vacation for25-year employees is estimated similarly, butone must know the number of employees withthe required seniority.

Other settlement terms are more difficult toprice. For example, the cost of an unfundedseverance pay plan depends on the frequencyof layoffs as well as on plan provisions.Estimates of such frequencies are at besthazardous. Pension improvement costs areparticularly difficult to estimate because of theconsiderable discretion employers often havein funding their obligations." The general ap-proach followed by the Bureau is to assumethat a given pension benefit increase will raiseexisting expenditures for current service pro-portionately. Since employer contributions forpensions frequently vary widely from yearto year, outlays in several past years areexamined to develop a measure of current pay-ments.

Under the BLS framework, estimates con-cerning most provisions are of actual cashoutlays to be made by employers. However, inthe case of improved paid leave provisions;' achange may entail time off for workers, butnot additional cash payments by the employer.

Where appropriate, the effect of a general wage increaseon incentive earnings is included in the measurement of thewage increase.

9 Since estimates are on a cents-per-hour-worked, ratherthan 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-vide given levels of pension benefits, rather than to expendspecified amounts of money.

However, payment per hour worked will riseand this change is taken as the cost effect ofthe settlement provision.

In case of a reduction in the basic workweek,the increase in hourly rates needed to maintainweekly pay is the major item BLS prices. Tosome extent, a reduced basic workweek may beaccompanied by additional overtime work.However, unless this overtime is provided spe-cifically in the agreement, it is ignored in thecost estimate.

IncreasesIn hourly pay rates are not theonly cost effects considered in this instance.Even if there is no change in total employeroutlays for particular pay supplements but thecontract provides for reduced hours, the out-lays for them per hour worked will rise andaffect the cost of settlement.

Expressing the costs. The total cost of a givensettlement is obtained by adding up the cents-per-hour-worked costs of each of the individualwage or benefit changes. This sum is thenexpressed as. a percent of pay, as this ratiofacilitates inter-company comparisons by elimi-nating influences of payroll size and wage level.Furthermore, since economic studies generallyemphasize relative rates of change in statisticalseries, percent-of-pay costs can be integratedinto broad economic analyses.

Expression of package costs as a percent ofpay requires estimation of an appropriate base(the denominator of the ratio) as well as thecost of the settlement terms (the numerator).The base used by the Bureau consists of cur-rent outlays per hour worked for all items ofemployee compensation, as defined, plus em-ployer expenditures for legally-required socialinsurance. The latter is part of employee com-pensation, although not subject to changethrough collective bargaining.

Since collective bargaining agreements gen-erally are for 2-year periods or longer, BLSexpresses the total percent increase over thecontract term at an annual rate to permit com-parison among agreements for differing timespans as well as to facilitate the use of the datain conjunction with other statistical series.These annual rates of increase take into ac-count the compounding of successive changes.

MEASURING COLLECTIVE BARGAINING SETTLEMENTS 205

In addition, the Bureau computes first-yearchanges, i.e., the total change scheduled forthe first 12 months of the agreement, expressedas a percent of current hourly compensation.As a general rule, the first-year increase islarger than the average annual increase overthe full term of the agreement ; contracts com-monly are "front-loaded."

Contracts are considered to run from theireffective dates to their termination dates. How-ever, where wage reopening clauses are found,the reopening date is taken as the terminationdate and any agreement under the reopeningclause is treated as a new settlement.

Pricing of a collective bargaining settlementis illustrated on the sample worksheet. Thisexample 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). Alsoassumed is a creep factor of 20 percent, 2,000annual working hours per employee, and a3-year agreement effective January 1, 1971,providing the immediate and deferred wageand benefit improvements shown on the work-sheet. The settlement provides a 7.1 percentfirst-yerr package and a 6.5 percent annualrate of increase over the 3-year term. Theworksheet also shows the wage and benefitgains scheduled for each of the 3 calendaryears (1971, 72, and 73), from which materialthe series on changes actually placed into effectis developed," and the wage-rate changes apartfrom benefit improvements. The latter data arecomputed without reference to creep and relatewage gains to average hourly earnings ratherthan average hourly compensation.

Prehentation and Analysis

Press releases covering wage changes andwage and benefit changes in major collectivebargaining settlements are issued toward theend of the month following the close of each

quarter. These releases contain preliminarydata for the first 3, 6, and 9 months of a yearand for the full year. This material also ap-pears in Current Wage Developments (CWD).Final quarterly material (both for individualquarters and the cumulative quarterly materialappearing in the press releases) is included inthe CWD article for the full year. An annualsummary also appears in the Monthly LaborReview.

Summary data for recent individual qudrtersand 4-quarter periods are shown monthly inCWD together with other statistical seriesdepicting aspects of change in employee com-pensation. The presentation facilitates analysisof the interrelations between the series and thedivergences in their movements.


Package cost data are used extensively byunion and management officials, for whom dataon developments in other firms and industriesoften provide an important criterion for theirown deliberations. In a different vein, the dataare examined by government officials andprivate analysts, concerned with the economicrepercussions of collective bargaining on thecosts of individual employers and on wage-price-employment relations within the eco-nomic system as a whole.

The user of the data should remember thatthe series does not purport to measure allchanges in average hourly expenditures foremployee compensation. Estimates are derivedunder the assumption that all factors affectingemployer outlays other than contract modifica-tions are constant.

Nevertheless, changes in the volume of over-time and shift work, in the composition of thework force, in the level and stability of employ-ment, in factors affecting incentive earnings,etc., are not unusual, and will influence outlaysfor employee compensation. In some instances,these changes are introduced by managementspecifically to offset costs of new labor agree-

"Thus, the series on effective package c.ianges essentially ments. In other cases, changes are the result ofis based on the pricing of items at the time settlements are modified production schedules or of tech-reached. The only exception is the subsequent addition ofcost-of-living escalator wage changes. nological developments independent of collec-

1139

PACKAGE COST ANALYSIS WORKSHEET

Name -

Location - Union

X

I

SCHED,

A.H.E. SIC

EMPLY.

Wages

Wages

DUR

WAB

OTL*

F.Y

F.Y.

Total

WAB

12131

MIEMMMEMEMMEMEEENMEMEMEEMEMMEMEMMMTEEMMEMMIMMMMMMIIMMMMTMMOTITIMMTMM:

2 3 4 5 6 7

sion

II

1.1112110111 C11111 MN 9036001 IMMO

Y0 M P A

yA

MIRE

00 0

Date Priced Out

3/15/71 Escal

Clause Infor. (if applicable)

Previous Expiration Date

12/31/70

Date Negotiated:

1/15/71

Effective Date of Agreement

1/i/7 1

New Expiration Date

-71'2/31/73

N/A

Increases

in

cents

STATISTICALINFORMATION

MEASURE:

Total

Percent

Increase

First-Year Wages & Benefits

Wages and Benefits

(over life of contract)

First-Year Wages Alone

Wages Over Life of Contract

Wages and Benefits Effective in





Wage's Alone Effective in

Wages Alone Effective in



Wages Alone Effectivein\

Other:

1971

23.54

$ 3.00

56.91

28.25

Annual

Rate of

Increase

7.1%

6.57.

$ 3.00

18.97

$ 4,00

19 72

22.69

$ 4.2825

1973

32.67

$ 4.509

N/A

N/A

N/A

N/A

N/A

N/A

1971

23.54

$ 3.00

1972

16.18

$ 3.235

1973

17.19

$ 3.397

N/A

N/A

N/A

N/A

N/A

N/A

N/A

N/A

N/A

7.8%

6.07.

7.17.

Provision

Exclud:With

creep

creep

as

Skill adjust

5C for 207

Impr. shift diff.

add. 2C hr. on 2nd shift

1/1/71

_1/1/71

for

6.11

1.00

1.20

307.

.6C

add. 3C hr. on 3rd shift for

207.

.6C

2% @ 3.172

5% @ 3.235

3% @ 3.397

27. @ 3.499

10/1/71

7/1/72

1/1/73

7/1/73

1.20

1.44

6.34

7.61

16.18

19.42

10.19

12.23

7.00

8.40

1/1/73

$3.499 @ 8 hrs. = 27.99

1972 = hrs.

1.42

Vacations:

Impr. averaging 20 add.

hrs. of vac./ vr. for all empls

$3.235 @ 20 hrs. = 64.70

1/1 72

1980 hrs

added impact in 1973

3.27

$3.499 @ 20 hrs. = 69.98

1972 hrc

5.3%

3.55C

(3.55C - 3.276 =

2R

7.2

Pensions add. 5C / hr. 4,fund 1/1/71

5C @ 2080 hrc

= q10A Afl ! 1079 =

N/A

N/A

7.8%

5.0%

5.1%

N/A

N/A

N/A

Remarks:

** For wages and benefits, divide by the base; forwages sleet.,

divide by the A.H.E.

For the effective measures, divide by the adjusted base or A.H.E.

as of the start of the appropriate year.

Priced Out By'

John Doe

Note:

7'7

1/1/73

Estimate avg. cost of

100/empl

$100 = 1972 =

5.07

TOTALS:

Wages and fringes --- hob, 83.61

c;64:31n;27111

Wages

* No data should be supplied in these columns

When verification is

made, the data will be supplied.

MEASURING COLLECTIVE BARGAINING SETTLEMENTS

tive bargaining, and may either add to orsubtract from the cost of the union-manage-ment settlement. In any event, an importantinfluence 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 suturechanges in employer outlays. As already em-phasized, completely accurate estimates shouldnot be expected. Secondly, the data apply pri-

207

marily to settlements for 5,000 workers ormore.

Although package cost estimates are ex-tremely valuable as comprehensive measures ofchange resulting from union-management ne-gotiations, to use the estimates as precise,unambiguous, and unfailing measures of theeconomic effects of collective bargaining isadding an assignment which the data are in-capable of fulfilling.


Number1. 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. "The 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 ofthe relation between package cost estimates and actual changes in hourly compensationexpenditures.

VICTOR J. SHEIFER

Chapter 24. Wage Chronologies and Salary Trend Reports

Background

Most Bureau of Labor Statistics series show-ing the movement of money wages-such asdata on average hourly earnings-apply tolarge aggregates of workers, e.g., all produc-tion workers in a given manufacturing indus-try either nationwide or in a particular Stateor local area. Wage chronologies and salarytrend reports, on the other hand, apply tomore narrowly defined employee groups.

Chronologies report on wage-rate changesmade by specific employers. Except for thestudy on Federal Classification Act employees,they deal with developments under collectivebargaining agreements. In all cases they reporton supplementary benefit as well as wagechanges.

Salary trend reports present and analyzechanges in salariel of selected categories ofgovernment employees.' In most instances,however, they do not contain separate data forindividual employers.

Both programs date back to the early post-World War II period. The wage chronologyprogram was instituted in 1948 and the firstsalary trend report was issued in 1950. To theextent possible, material for earlier years wasincluded in the initial reports.

Description of Program

Chronologies. The following 32 chronologiescurrently are being maintained:2

1. Aluminum Company of America2. American Viscose Division of FMC Corp.3. The Anaconda Company4. Armour and Company5. A.T. & T.-Long Lines Department6. Berkshire Hathaway Inc.7. Bethlehem Atlantic Shipyards8. Bituminous Coal Mines

I Although other BLS studiec repott on salary trends, theydo not have this specific title. The time series in these studiesare often by-products of repetitive Bureau survey activity.

2 Several chronologies have been discontinued, generallyeither because of fragmentation of bargaining units, decliningimportance, or because standardization of collective bargain-ing eliminated the value of more than one chronology in agiven industry.

-e2

9. The Boeing Company10. Commonwealth Edison Company11. Dan River Mills12. Federal Classification Act Employees13. Firestone Tire and Rubber Company and B.F.

Goodrich Company14. Ford Motor Company16. International Harvester Company16. International Paper Company (Southern Kraft

Division)17. International Shoe Company18. Lockheed-California Company (A division of

Lockheed Aircraft Corp.)19. Martin-Marietta Corp.20. Massachusetts Shoe Manufacturing21. New York City Laundries22. North America Rockwell Corp.23. North Atlantic Longshoring24. Pacific Coast Shipbuilding25. Pacific Gas and Electric Company26. Pacific Longshore Industry27. Railroads-Nonoperating Employees28. Sinclair 0i1 Cernpanies29. Swift & Company30. United States Steel Corp.31. Western Greyhound Lines32. Western Union Telegraph Company

Each chronology covers either a single wage-determination unit or a group of closely relatedunits. It may cover an individual company andunion (e.g., Ford Motor Company and noUnited Automobile Workers), a single com-pany and two or more unions (e.g., the Alumi-num Company of America and the UnitedSteelworkers and the Aluminum Workers), agroup 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., PacificCoast shipbuilding companies and a number ofcraft unions), or a governmental body (e.g.,the chronology covering Federal ClassificationAct employees).

The program is designed to summarize long-term wage-benefit developments in a varietyof industrial environments. Accordingly, chro-nologies cover groups that (1) have existedover a period of years; (2) are important intheir own industry; (3) employ a significantnumber of workers; (4) are of general publicinterest; and (5) are willing to cooperate withthe Bureau and for which appropriate infor-mation is available.

202 c266 209

.4

210 BLS HANDBOOK

Each chronology is divided into a narrativesynopsis of the collective bargaining or legisla-tive developments resulting in wage and benefitchanges and a tabular summary of the changesthemselves. Each chronology contains separatetables showing general wage changes andchanges in supplementary benefits.

As used in the wage chronology program,general wage changes are defined as upwardor downward changes that affect an entire unitcy. 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 ratesor incentive rates) that do not have an im-mediate and noticeable effect on the generalwage level. Because of the omission of non-general wage changes, fluctuation in incentiveearnings, and other factors, the sum of thewage changes listed in each chronology willnot coincide necessarily with the movement ofaverage hourly earnings over the same period.

The tables generally include benefits such asguaranteed minimum earnings, shift premiums,daily and weekly overtime, weekend premiums,pay for holiday work, paid vacations and holi-days, other paid leave provisions, reportingtime, waiting time, paid lunch periods, pay fortravel time, and health, welfare, and pensionbenefits.

When minimum plant rates, common laborrates, occupational wage rates, or rates forlabor 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 currentlyare issued for Federal classified employees,firefighters and police patrolmen, and urbanpublic school teachers. They all contain indexesof long-term salary movements. Data on recentand current salary levels also are provided.

Reports for firefighters, police patrolmen,and teachers basically apply to cities of 100,000population or more, and provide separate fig-ures for regions and city-size groups, as wellas overall national data. Material on individualcities, however, is not shown. Because it dealswith a single employee group, the report on

OF METHODS

Federal classified employees contains consider-able detail on developments in wage structure.

Data Sources

Both wage chronologies and salary trendreports are developed primarily from publisheddata. Collective bargaining agreements, pen-sion and welfare documents, and newspaperand periodical articles provide the main sourcesof chronology information. These are supple-mented by direct requests to the parties forinformation when available written records areinadequate. Thus, most of the information usedto compile chronologies is, in one form oranother, already a matter of public knowledge.In all instances, to avoid dissemination oferroneous material, pre-publication drafts ofreports are submitted to the parties for theirreview and comments.

Salary trend reports are prepared largelyfrom salary data collected by other groups.That for Federal classified employees is basedon data published by the U.S. Civil ServiceCommission in its annual report on Pay Struc-ture of the Federal Civil Service.

Reports on teachers and firefighters andpolice patrolmen are based mainly upon com-pilations of data for individual cities made bythe National Education Association and theInternational City Management Association,respectively. These are supplemented by annualsurveys of salaries and working conditionsconducted by the International Association ofFire Fighters and the Fraternal Order ofPolice and, when necessary, by direct inquiriesby the Bureau of Labor Statistics.

Statistical Procedures

Statistical analysis in these programs isconfined largely to the preparation of indexesof salary movements of groups of governmentemployees. Sampling problems do not arisesince in each case an effort is made to examinethe total universe, i.e., all Federal classifiedemployees, and all teachers, firefighters, andpolice patrolmen in cities of. 100,000 populationor more.

2C3

WAGE CHRONOLOGIES AND SALARY TREND REPORTS 211

Indexes generally are computed by a methodthat minimizes the effect of year-to-yearchanges in relative employments in the citiesor occupational categories covered. As a rule,chain indexes are employed, i.e., the index forthe current year is obtained by adjusting theindex for the prior year by the percentagechange in average salaries over the interveningperiod. Normally, to preserve a pure measureof salary change, average salaries for each ofthe two years are computed using current-yearemployments as weights.

Presentation and Analysis

Wage chronologies. Wage chronologies arepublished individually as BLS Bulletins andrevised periodically to incorporate materialresulting from new collective bargaining settle-ments or legislative developments. Bulletins areupdated after every other contract settlementor legislative enactment. Intervening develop-ments are reported in supplements to existingbulletins. Thus, when 3-year collective bar-gaining agreements are negotiated, a revisedbulletin 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 annuallyin Current Wage Developments. Press releasescontaining sumMary data for firefighters andpolice patrolmen precede publication of thearticles. Salary changes >r teachers are re-ported on in biennial CWD articles, since thebasic data are issued at two-year intervals.Reprints are available of all CWD articles. Inaddition, all articles issued up to the mid-1960'shave been collected and reprinted in the BLSBulletins listed in the technical references atthe end of this chapter.

Chronologies are primarily listings of wageand benefit changes, with background materiallimited to descriptions of the collective bar-gaining or legislative processes leading up tothe changes. Greater effort is made in salary

trend reports tc analyze the thita. Backgroundfactors are presented and the wage movementsare compared with wage changes for otheremployee groups.


Both wage chronologies and salary trend re-ports serve two audiences; they are useful assources of comparative wage data for union,management, and government officials engagedin wage setting and as research tools for eco-nomic analysts. Chronologies are particularlyuseful I -)r negotiators because the studies pre-sent dei,...tiled information on developments inunits that not infrequently provide wage lead-ership for their industries. Moreover, com-parisons of wage and benefit changes in suchunits provide valuable insights into wage set,ting in the American economy. In addition, thedata help to explain the movements in aggrega-tive statistics such as the Bureau series onaverage hourly earnings.' Salary trend reportsare noteworthy as one of the relatively fewsources of data on wage movements and levelsin the government sector.

Although wage chronologies describe changesin supplementary benefits, they do not measurethe effect of these changes on employers' hourlylabor costs. In this connection, decisions onwhether or not to adopt a given benefit changemay hinge on its cost, which may vary amongbargaining units with work forces of differingcomposition.

The temptation may be to use salary trendreports as indicators of salary movements forgovernment employees in general. However,the particular groups covered by these reportsare by no means a representative sample forthis purpose.

'For examples of the use of chronologies by economists,see Harold M. Levinson, Postwar Movement of Prices andWages in Manufacturing Industries, Study Paper No. 21, Studyof Employment, Growth, and Price Levels. Joint EconomicCommittee, 86th Cong., 2d Sess. (Washington: 1960); andJohn E. Maher, "An Index of Wage Rates for Selected In-dustries, 1946-1957," Review of Economics and Statistics,August 1961, pp. 277-282.

212 /374 HANDBOOK OF METHODS


Number ti

1. Arnow, Philip, Bloch, Joseph W., and ()mint, Willis C. "The New Wage Chronology Series,"illonthly Labor Review, December 1948, pp. 581-583.

Describes the aims of the chronology series al the time it was introduced.2. U.S. Department of Labor, Bureau of Labor Statistics. A Directory of Wage Chronologies,

1948June 1989 (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, 1939-64 (Bulletin 1444, 1965).5. _ Salary Trends: Firemen and Policemen, 1924-64 (Bulletin 1445, 1965).

Compilations of previously published salary trend articles.

VICTOR J. SIMPER

2,05

F.

Productivity and Technology

Chapter 25. Output Per Man-Hour: Private Sector

Background and Description ofMeasures

To provide information about the relation-ship between productivity, prices, wages, em-ployment, and economic growth, the Bureauof Labor Statistics publishes indexes of outputper man-hour, compensation per man-hour,unit labor costs, and related costs for broadeconomic sectors. Measures of output per man-hour have been developed for the total privatesector 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 andrelated 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 andcosts is available for the nonmanufacturing sec-tor and its component major industrial sectors.Productivity measures, first published in 1959,'represent the culmination of a long series ofdevelopments in productivity measurement inthe Bureau.2

The Bureau of Labor Statistics' output perman-hour measures specifically refer to theratio between dollar gross product (GNP)originating 3 in the private or individual sec-tors and the corresponding hours of all personsemployed. Indexes of output per man-hour in-dicate the relationship between output and la-bor input (man-hours). Index changes throughtime show the effectiveness of man-hours inproducing 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 ofthe productivity measurement program of the BLS, seeChapter 26.

3 Gross Domestic Product equals Gross National product lessthe net return on foreign investments.

lishment payroll data . on employment andhours. These man-hours refer to hours of workand paid time for vacation holidays and sickleave.

The Bureau also develops an output perman-hour series based on labor force man-hourdata from the Current Population Survey(CPS). The concept underlying the labor forceman-hours is hours worked, rather than hourspaid. Theoretically, the difference between thelabor 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 statisticalvariation and differences in method.

Indexes of compensation per man-hour meas-ure the hourly costs of wages and salaries, inaddition to supplemental payments, such as thecontributions of employers to social security,unemployment insurance tax, and payment forprivate health and pension plans. Measures ofreal compensation per man-hour reflect theadjustment of hourly compensation for changesin the Consumer Price Index.

Unit labor costs measure the cost of laborinput required to produce one unit of outputand is derived by dividing compensation perman-hour by output per man-hour. Unit non-labor payments (costs) measure the cost ofnonlabor inputs such us depreciation, rents,interest, indirect taxes, in addition to profitincome such as corporate profits and incomeof proprietorships and partnerships.

Data Sources and Estimating Procedures

Output. The real gross national product orig-inating in the private or individual sectors isused to prepare output per man-hour estimates.Gross national product is the market value offinal goods and services produced within acertain time period. It includes purchases ofgoods and services by consumers gross privatedomestic investment, net foreign investment,and purchases by Government. GNP is equalto income received by labor and property for

2C6213


services rendered in the current production ofgoods and services, in addition to capital con-sumption allowances, indirect business taxes,and several other minor items.

Gross national product in current dollarscannot be used directly as the output measuresince it reflects price changes in addition tochanges in physical volume. The Office of Busi-ness Economics (OBE), U.S. Department ofCommerce, prepares estimates of constant-dollar GNP for the total private sector andmajor sectors. These estimates indicate onlychanges in the volume of production and areused to develop output per man-hour measures.'

Compensation. The OBE develops compensa-tion of employees data as part of the nationalincome accounts. These data include directpayments to labor, such as wages and salariesinclusive of executive compensation, commis-sions, tips, and bonuses and payments in kindwhich represent income to the recipients andsupplements to these direct payments. The lastitem consists of employer contributions forsocial insurance, private pension and healthand welfare plans, compensation for injuries,doctors' fees, pay for military leave, etc.

Since these compensation measures referonly to wage and salary workers, they do notreflect the cost of labor expended by proprietorsand unpaid family workers. Labor costs can beseriously underestimated in sectors like thefarm where proprietor man-hours represent asubstantial portion of the labor input. The Bu-reau, therefore, imputes a payment for laborservices of proprietors and family workers.Compensation per man-hour of proprietors isassumed to be the same as that of the averageemployee in that sector.

Unit labor and nonlabor costs. The Bureau alsopresents data on labor and nonlabor costs perunit of output for the private sector and its

A detailed description of the methods and procedures ofestimating GNP and GDP in current and constant dollars isgiven in the 1954 National Income Supplement to the Surveyto Current Business, U.S. Department of Commerce. Furtherinformation on Gross Product Originating estimates formajor industry sectors is presented in the October 1962 issueof Survey of Current Business.

major components. Unit labor costs relatehourly compensation of all 1:ersons to outputper man-hour. Unit nonlabor costs are com-puted by subtracting compensation of all per-sons from current dollar GNP and dividing byoutput.

Labor input (man-hours). The primary sourceof man-hours and employment data is the BLSCurrent Employment Statistics (CES) pro-gram, which provides data on employment (allemployees and production or nonsupervisoryworkers) and average weekly hours of produc-tion workers in nonagricultural establishments.Jobs rather than persons are counted, so thatmultiple jobholders are counted more thanonce. Weekly hours are hours paid rather thanplant hours. These statistics are based on pay-roll records frox. a sample of establishmentswhere probability of sample selection is pro-portionate to the establishment size ; largeestablishments (relative to the sector) fallinto the sample with certainty. Data on em-ployment, hours, and earnings are collectedmon thly; however, the reference period forthese data is the payroll period including the12th of the month. Methods are described inChapter 2. Establishment data are publishedmonthly in Employment and Earnings and inan annual summary.

Since the CES includes only nonfarm wageand salary workers, data from other sources(National income accounts or the CPS) areused for the uncovered sectors (farm, proprie-tors, unpaid family workers, and privatehousehold workers) to develop employment andman-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 totalprivate and private nonfarm levels. Each man-hour is treated as a homogeneous unit ; no dis-tinction is made between workers who havedifferent skill levels or rates of pay.

In the manufacturing sector, separate esti-mates for production and nonproductionworker man-hours are derived and then aggre-gated to the manufacturing total. Productionworker and nonproduction worker employmentand production worker average weekly hours

OUTPUT PER MAN-HOUR: PRIVATE SECTOR 215

are taken directly from CES data. Averageweekly hours for nonproduction workers aredeveloped from BLS studies of wages. and sup-plements in the manufacturing sector whichprovide data on the regularly scheduled work-week of white-collar employees.5

For nonmanufacturing sectors, employmentand weekly hours paid are taken from the pay-roll series. Although payroll average weeklyhours data refer to nons.ipervisory workers forman-hours computation, the assumption is thatthe length of the workweek in each nonmanu-facturing industry is the same for all wageand salary workers.

When establishment employment data arenot available (proprietors, unpaid familyworkers, and private household workers),either labor force data (CPS) or national in-come employment data are used. On the otherhand, average weekly hours are based on laborforce data for hours worked rather than hourspaid. However, perscais who have a job but arenot at work are assumed to have been paid foran average workweek in the sector where thejob was located.


Ylndexes of output per man-hour showchanges in the ratio of output to labor input(man-hours). These indexes relate output toman-hours but should not be interpreted asrepresenting labor's sole contribution to pro-duction. Rather, they reflect the interaction ofmany forces in addition to labor and skill, suchas changes in technology and increased capitalinvestment per worker.

For manufacturing nonproduction workers, average hoursare estimated as follows: Estimates of vacation time, holidays,paid sick leave, and personal time off are subtracted froman estimate of scheduled annual hours paid nonproductionworkers; scheduled annual hours are derived by extrapolat-ing the 1959 level of scheduled weekly hours with data fromBLS Area Wage Surveys, and then multiplying by the num-ber of workweeks in each year; the level of scheduledweekly hours for nonproduction workers was calculated fromdata collected by BLS for the study, Employer Expendituresfor Selected Supplementary Remuneration Practices forProduction Workers in Manufacturing Industries, 1962, BLSBulletin 1428 (1965); estimates of vacation time, holidays,sick leave, and personal time off are derived primarily fromdata from Area Wage Surveys and Social Security Ad-ministration studies.

For economic aggregates like productivityin the private sector, changes over time reflectmovements 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 influencednot only by the increments or decrements inthe component sectors but also by the shiftfrom low to high productivity industries,Within industries, other forms of shifts alsotake place and are not accounted for ade-quately. In output measures, for example,changes in income and taste may be reflectedin shifting consumption patterns to higherquality goods or to services rather than goods.

Short-term movements in productivity andunit labor costs often result from cyclical varia-tion which tends to distort the underlyingrelationship between the numerator and de-nominator of the ratio. For example, because ofmarket imperfections, an employer often hasdifficulty in adjusting his staff immediately tochanges in production. Sharp increases or de-creases in de and usually are followed byerratic move ents in productivity and unitlabor costsrygn longer periods are analyzed,however, these fluctuations tend to iron rutand a secular trend is more apparent.

Long-term productivity trends tend to movemore smoothly over time and present a morestable picture of historical patterns than short-term movements. The latter tend to reflecttemporary changes in level of demand, utiliza-tion of productive resources, and other short-term phenomena, such as strikes, which affectcurrent conditions but should not be expectedto continue into the future. The period selectedfor analysis may affect these long-term trends,so that some element of judgment is present inany analysis of trend.

Indexes of output per man-hour, compensa-tion per man-hour, and related cost data arepublished quarterly in the BLS press release,Productivity and Costs. Historical indexes ofthese data are available on request. Detailedprocedures for developing output per man-hourindexes appear in Trends in Output per Man-Hour in the Private Economy, 1909-58 (BLS


Bulletin 1249, 1959). Indexes of output perman-hour and related cost data appear in Em-ployment and Earnings, the Monthly LaborReview, the Handbook of Labor Statistics, TheEconomic Report of the President, the Man-power Report of the President, and the Sta-tistical Abstract of the United States.


Data on output per man-hour, compensationper man-hour, and related costs are designedfor use in economic analysis and public andprivate policy planning. These data apply inareas such as wage determination and analysisof prices and living conditions.

An especially relevant use of output perman-hour and related costs is the relationshipof productivity, wages, prices, profits, and costsof production. Within the framework of na-tional income and product accounting, gross

g national product for a specific time periodrepresents the market value of all final goodsand services produced or the sum of all costsof productioncompensation, profits, deprecia-tion, interest, etc. Unit labor costs, or compen-sation per unit of output, represent a majorportion of total unit costs and reflect the com-bined effect of changes in output per man-hourand compensation per man-hour. An increase

in hourly compensation tends to increase unitlabor 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 thewage-price relationship. It indicates the extentto which compensation gains can occur withoutputting pressure on prices or reducing profits.

Certain characateristics of the output perman-hour and related cost data should be,-ecognized when applying them to specificsiivations. First, the data reflect not onlychanges in various component industries butalso changes in the relative importance of theseindustries. Second, these measures representthe culmination of a chain of economic events.Data for a single time period often do notshow direct causal relations and should beinterpreted in light of previous economic eventsas well as current conditions. Third, underlyingconcepts and data available for estimation limitto some extent measures of productivity, out-put, compensatiopi,!and employment. In fact, inseveral sectors,ffltre output data are difficultto obtain, output changes are equal to employ-ment changes. This definition understates pro-ductivity growth. Consequently, the meaningof these measures should be interpreted withcaution because of conceptual and practicallimitations and statistical errors which canarise in any numerical series.


Number1. Edward Denison, Why 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 factorscan 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., Princet3n University Press (1961).

A collection of papers and comments devoted to an appraisal of the measurement ofoutput, input, and productivity.

OUTPUT PER MAN-HOUR: PRIVATE SECTOR


4. 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 andcost measures for individual industrial sectors.

5. National Bureau of Economic Research, Conference on Income and Wealth, "Production andProductivity 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.

6. U.S. Department of Labor, Bureau of Labor Statistics, Treads in Output per Man-Hour inthe Private Economy, 1909-1958, (Bulletin 1249, 1959).

A presentation and analysis of output per man-hour indexes and trends for the totalprivate U.S. economy and major sectors. Includes a description of methods and sources fordeveloping output per man-hour measures.

7. U.S. Department of Labor, Bureau of Labor Statistics, Productivity: A Bibliography (Bul-letin 1514, 1966).

A collection of nearly 600 references concerning productivity cad productivity measure-ment. Each reference includes a brief annotation giving the gist of the subjects covered.

8. Mark, Jerome A., Wage-Price Guidepost Statistics: Problems of Measuremenl, 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 usedin the specification of the guideposts.

9. Ziegler, Martin, "Productivity in Manufacturing," Monthly Labor Review, October 1967.An analysis of the effect of changes in capacity utilization on productivity.

SHELBY A. HERMAN

t

I

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 BLSprogram. The first Commissioner of Labor,Carroll D. Wright, conducted a study of 60manufacturing industries in 1898. The findingsin his report on "Hand and Machine Labor,"provided striking. evidence of the savings inlabor resulting from mechanization in the lasthalf of the 19th century. CommissionerWright's study was prompted by concern onthe part of Congress that human labor wasbeing displaced by machinery. The impact ofproductivity advance upon employment re-mained an important focus of the BtS programthroughout the 1920's and 1930's. It was alsoduring this period after World War I that theBureau began the preparation and publicationof industry indexes of output per man-hour,based on available production data from theperiodic Census of Manufactures and employ-ment statistics collected by the BLS.

In 1940, Congress authorized the Bureau ofLabor Statistics to undertake continuing stud-ies of productivity and technological changes.The Bureau extended earlier indexes of outputper man-hour developed by the National Re-search Project of the Works Projects Admin-istration, and published measures for selectedindustries. This work was reduced in volumeduring World War II, owing to the lack ofmeaningful production and man-hour data formany manufacturing industries.

The advent of World War II also caused achange in program emphasis, from problems ofunemployment to concern with the most effi-cient utilization of scarce manpower. The BLSundertook a number of studies of labor require-ments for defense industries, such as syntheticrubber and shipbuilding. After the war, theindustry studies program resumed on a regularbasis, and was supplemented by a number ofindustry studies based on the direct collectionof data from employers. Budget restrictionsafter 1952 have prevented the continuation ofdirect collection of data. Consequently, thepreparation of industry measures is limited to

those industries where readily available datacan be utilized to construct measures.

In recent years, public interest in produc-tivity has grown, and there has been greaterycognition that increases in output per man-

hour are important indicators of economicprogress and the means to higher levels of in-come, rather than merely a threat to job oppor-tunities.

The Industry Studies program covers avariety of manufacturing and nonmanufactur-ing industries. For these industries, indexes ofoutput per man-hour, output per employee andthe related data on output, employment, andman-hours are prepared and published on anannual basis. The indexes are generally avail-able for most years from 1947 to the mostrecent year for which data are available, andfor many industries also for 1939.

Concepts

Industry indexes of output per man-hourmeasure changes in the relationship betweenthe Physical volume of output of an industryand the man-hours expended in that output.Although, traditionally, output per man-hourhas been the measure most frequently used,discussion of physical output per man-houroften is simplified if conducted in terms of itsreciprocal: man-hour requirements per unit ofoutput (unit man-hours). Therefore, this formof index is used in the following description.

For an industry producing a single uniformoutput, the unit man-hours index is simply theratio of the man-hours expended to produce aunit of output over two periods of time. Thisratio may be expressed as follows:

1/ = ,

" 4,

Where I represents the unit man-hour index,I represents the output per man-hour index,and 1, and denote unit man-hours expendedin the current and base periods, respectively.

For an industry producing a number of prod-uctsthe more typical casethe unit man-hours index is the ratio for two periods of the

2'1 onf/ 219

220 BLS HANDBOOK

total hours required for the output of a givencomposite of products. Indexes of such indus-tries vary with the composite of products spe-cified and can take many forms. Letting q, andq, represent basePeriod -and current periodquantities of a given product, respectively, twoof these forms are :

a. Using a current period compositen1,

1=_-7,

b. Using a base period composite

I.=An index constructed according to (a) corn-

pares the man-hours expended in the produc-tion of the current composite with man-hourswhich would have been required to producethe current composite in the base period. Anindex constructed according to (b) comparesthe man-hours required in both periods to pro.duce the base period composite. These indexesthus eliminate the effects of variations overtime in the relative importance of products onunit man-hours.

In either form, an index of unit man-hoursalso can be viewed as the quotient of an indexof man-hours and an index of output, i.e.,Man-hours Output index Unit man-hours

index (Laspeyres) = index (Paasche), 2144iii=

21.q. El.q,

Man-hours Output index Unit man-hours indexindex÷ (Paasche) (Laspeyres)

, 214q4 El4q4J.= :21.q.

The man-hours index measures the change inaggregate man-hours between the base andcurrent periods. The man-hours data are thetotal hours expended by employees in establish-ments classified in the industry, in producingthe base period and current period composites.

As can be seen in the formulas, the appro-priate output index is one which compares thequantities of the various products in the cur-rent and the base periods, each weighted bythe man-hours expended per unit produced ina given period. A current period weighted

OF METHODS

unit man-hours index employs a base periodweighted output index divided into the man-hours index. Conversely, a base periodweighted unit man-hours index is consistentwith an output index which utilizes currentperiod weights.

Methods and Sources

Output Per Marliour

The Bureau of Labor Statistics computes anindex of output per man-hour by dividing anoutput 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 andnonproduction workers are used.) 'Three cor-responding measures also are computed relat-ing output to the number of employees.

Output

BLS industry output indexes are basedprimarily on the physical output of the prod-ucts of the industry combined with fixed periodweights. However, the availability of quantitydata on physical output varies among indus-tries, and, for manufacturing and mining in-dustries, may vary depending on whether thedata are for a year when a Census was con-ducted or for a noncensus year. For manu-facturing and mining industries, quantity dataon physical output are usually most compre-hensive for years covered by a Census. Tomake maximum use of the comprehensivecensus data, output indexes are derived fromdata for two consecutive censuses; these in-dexes are referred to as benchmark indexes.For intercensal years, annual indexes are basedon either physical output data (generally, inless detail than for Census years) or if suchdata are not available, value of output adjustedfor price change (i.e., the value of output inconstant dollars). The annual series subse-quently are adjusted to the benchmark levelsfor the census years.

3

OUTPUT PER MAN-HOUR MEASURES: INDUSTRIES 221

Weights. The mathematical form of the outputindex implies use of unit man-hour weights,and such weights are used whenever possible.These weights are derived from special surveysor from data published for specialized estab-lishments in the Census of Manufactures, Insome industries, however, unit man-hour infor-mation is not available for individual products.Consequently, the BLS uses substitute weightswhich are assumed to be proportional to unitman-hour weights. Usually these are unit valueweights. Unit value weights are computed fromCensus or survey- data on the quantity andvalue of shipments of the primary products ofthe industry. The introduction of these substi-tute weights results in an industry output perman-hour index which reflects shifts in valueper man-hour of the various products in theindustry. Thus, a change can occur in the indexwithout any change in the output per man-hourfor any product of the industry.

The extent to which error or bias may beintroduced by the use of unit value weightsis not known. The index is equivalent to oneweighted with unit man-hours if the unit man-hours and unit values among the products areproportional or if there is no correlation be-tween the relative change in quantity and valueper man -hour' There is evidence that unitvalues are fairly reliable approximations forindividual products where wages constitute alarge proportion of total value of output. Theerror generated in the output index by an errorin the weights is generally considerably smallerthan the error in the weights themselves.

In some industries, unit value weights forspecific products and unit man-hour weightsfor product groups are used at different stagesin constructing the industry output indexes.When this procedure is used, the individualproducts are first aggregated into primaryproduct group indexes with unit value weights.

1 See Irving II. Siegel, "Further Notes on the DifferenceBetween Index Number Formulas," Journal of the AmericanStatistical Association, December 1941, pp. 519-524.

2 The "specialization ratio" is the value of shipments ofpriMary products of plants in the industry as a percent oftotal shipments of all products (primary plus secondary)made by these same establishments, The "coverage ratio" isthe value of shipments of the primary products made byplants classified in the industry as a percent of the total ship-ments of the industry's psmat:y products made by all pro-ducers, both in and out of the specdied industry.

These indexes in turn are combined into anindustry output index with primary productgroup man-hours. The primary product groupman-hours relate to a base period, as do valueweights.

To obtain primary product group man-hourweights, total man-hours for plants specializingin each primary product class are derived frompublished census data on production workerman-hours and nonproduction worker employ-ment. These are supplemented by unpublishedBLS estimates of nonproduction worker man-hours. (See page 223 for the procedures usedto estimate nonproduction worker man-hours.)Ratios of man-hours to value of shipments aremultiplied by the corresponding value of pri-mary products shipped by the entire industryto yield the estimated primary product groupman-hour weights. This procedure assumesthat the man-hours per dollar for each productclass shipped by the whole industry are thesame as those for plants specializing in theproduct group. This procedure is used onlywhen the "specialization" and "coverage"ratios of the industry are high and specializa-tion data for all or most of the product groupsare available.2

Most published industry indexes have used1947 weights for 1947-58, 1958 weights for1959-63, 1963 weights for 1964-67, and 1967weights for years after 1967. The Bureaupolicy is to revise the weights as more currentdata become available from the periodiccensuses.

Benchmark indexes. For most manufacturingand all mining industries, indexes reflectingchanges in output between census years areconstructed. These are called benchmark in-dexes.

For manufacturing industries, benchmarkindexes are developed using the following pro-cedure: Price indexes for each primary productclass are developed from data on the value ofeach individual product within the classwhether made in the industry or elsewhefe.Wholesale price indexes are used whereverpossible to convert the product values to con-stant dollar estimates. If a wholesale priceindex is not available, a price index is de-213


veloped using both the quantity- and value thitareported for the product in the Census ofManufactures. The primary product classprice indexes are derived from the sum of thecurrent dollar values and the sum of the con-stant dollar values.

These "wherever made" primary productclass price indexes are used to deflate the valueof primary products produced only by the in-dustry. This procedure assumes that the pricemovements of the primary products within theindustry are the same as the price movementsfor all primary products wherever made. Theseconstant dollar values are related to cor-responding base year values in order to deriveseparate primary product indexes within theindustry.

These separate primary product indexes inturn are combined with man-hour weights toderive the total industry primary product out-put index. The index of primary products ofthe industry is multiplied by a "coverage" ad-justment to represent the total output of theindustry. This "coverage" adjustment is theratio of the index of value of industry ship-ments (after inclusion of net additions to in-ventories) to the index of value of shipments ofprimary products. The final industry outputindex thus reflects inventory buildups andchanging proportions of secondary products.

For the mining industries, benchmark in-dexes are computed from unweighted tonnagedata as reported in the Census of MineralIndustries.

Annual Indexes: Annual output indexes areconstructed by the following described pro-cedures. For manufacturing and mining in-dustries, the annual indexes are adjusted, ifnecessary, to the levels of the benchmark in-dexes previously described. The adjustmentfactors 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 outputindexes are based on physical quantities ofproducts combined with fixed-period unit man-hour or unit value weights. The basic quantitydata are generally primary products of an in-dustry classified into product groups; the

greatest amount of detail available is used. Thequantity data relate to primary products"wherever made" and in some cases to ship-ments of the products.

The Bureau's annual measures of productionare constructed from data on physical quanti-ties of products comprising a high percentageof the total value of an industry's output.Coverage varies between 60 and 100 percent.Complete coverage generally is obtained inmining and other well-defined industries witha relatively homogeneous output.

2. Deflated value. When adequate annualphysical quantity data are not available, in-dexes are derived from data on the value ofindustry output, adjusted for price change.Since the adjustment for price change is mostoften downward, the indexes usually are called"deflated value" indexes. Such indexes areconceptually equivalent to indexes which usedata on physical quantities of products com-bined with unit value weights. To derive thisindex, data on the industry's value of outputare divided by an industry price index. Anindex of these deflated values shows the changein the real value of output between the pastand current periods.8

Often data on value of production are notavailable and data on value of shipments mustbe used. In this case, data on value of ship-ments for each year are divided by an industryprice index representing the average annualprice for the year. Beginning- and end-of-yearfinished goods and work in process inventoriesare also deflated. The estimated value of ship-ments in constant dollars is then adjusted bythe net change in inventories, also in constant

nFor example:Value Index Price Index (Paasche) = Output index

(Laspeyres)

ZP91 ZP.91M7,9. 21391 Min%

where p, and p. represent prices of products in the industryin the current and base periods, respectively. This indexrequires quantities of all items produced in each year. Thesedata are not available for the particular industries wherethis measure is used, and quantity data are usually availablefor the base year only, so that the deflated value indexes em-ployed usually take the following form:Value Index ± Price index (Laspeyres) = Output index

(Paasche)

ZI/d41 EP,9. IPAEPA, MM. Z13i9.

Z134%

OUTPUT PER MAN-HOUR MEASURES: INDUSTRIES

dollars, to yield an estimate of the constantdollar value of production.

Sources. industry output indexes are preparedfrom basic data published by various publicand private agencies, using the greatest amountof detail available.

The Bureau of the Census, U.S. Departmentof Commerce, is the major source of outputstatistics for, the manufacturing industries.The Bureau of Mines, U.S. Department of theInterior, compiles most of the information forthe mining, cement, coke, and metals indus-tries. Other important Government sourcesinclude the U.S. Department of Agriculture,the Fish and Wildlife Service, U.S. Departmentof the Interior, the Inters. ate Commerce Com-mission, the Internal Revenue Service, and theCivil Aeronautics Board. Important sources oftrade association data include the Textile Eco-nomics Bureau, Inc., National Association ofHosiery Manufacturers, Inc., National CannersAssociation, Rubber Manufacturers Associa-tion, and the American Iron and Steel Institute.

For deflated value series, industry price in-dexes are derived from wholesale price indexespublished by the Bureau of Labor Statistics.

Man-Hours

An index of man-hours is computed bV divid-ing the aggregate man-hours for each year bythe base-period aggregate. Man-hours aretreated as homogeneous and additive with nodistinction made between hours of differentgroups of employees. Data on changes in qual-itative aspects of man-hours, such as skill,efficiency, health, experience, age, and sex ofpersons comprising the aggregate, are not usedand generally not available. Man-hours indexesare developed for all employees, productionworkers, and nonproduction workers.

Sources. Industry employment and man-hoursindexes are developed from basic data compiledby the Bureau of Labor Statistics or the Bu-reau of the Census. For most private nonagri-cultural industries (including 'manufacturing),the Bureau of Labor Statistics publishes em-ployment and average weekly hours data for

223

production or nonsupervisory workers and em-ployment data for all employees. For manu-facturing industries, the Bureau of the Censuspublishes employment and aggregate man-hours data for production workers and em-ployment data for all employees (includingnonproduction workers). The Bureau of theCensus provides data in greater industry detailwithin manufacturing than BLS.

The two sources differ in their definition ofman-hours. The Census data include all hoursat the plant, worked or paid for, and excludepaid time for vacations, holidays, or sick leave,when the employee is not at the plant. Over-time and other premium pay hours are includedon the basis of actual time at the plant. Incontrast, the Bureau of Labor Statistics datainclude time for paid vacations, holidays, andsick leave, as well as plant man-hours. Differ-ences in the data from the two sources for thesame industry, however, also stem fromsampling and reporting differences.

Whenever employment and hours data areavailable from both"- the-Bureau of the Censusand the Bureau of Labor Statistics, the laborinput data which are used are those consistentwith the data on output. Thus, when outputdata from the Bureau of the Census are used,employment and hours data from the samesource usually are preferred.

Nonproduction Worker Hours. While both theBureau of the Census and the BLS providedata on production worker man-hours, neithersource provides annual data by industry onnonproduction worker nor all-employee man-hours. Therefore, these measures are esti-mated.

The estimates of aggregate nonproductionworker man-hours for the manufacturing in-dustries are derived from published employ-ment data, and estimates of average annualhours worked or paid per nonproductionworker.

The estimates of average annual hoursworked are calculated by multiplying the num-ber of work-weeks in the year times the sched-uled weekly hours. This produces an estimateof average annual hours paid. Estimated hoursfor vacations, holidays, disability, and personal


time off are subtracted from average annualhours paid, to obtain an estimate for averageannual hours worked.

Vacation and holiday trends are based ondata from various BLS surveys. Estimates ofdisability time are based on studies of the De-partment of Health, Education, and Welfare,and data from BLS surveys. Personal time offhas been estimated as a constant from refer-ences in relevant publications.

All employee man-hours estimates for manu-facturing industries are derived by summingthe aggregate man-hours for production work-ers, and the estimated aggregate man-hoursfor nonproduction workers.

Comparability of Output and Man-Hours Data

Man-hours data are based on total man-hoursof establishments classified in an industry,whether the man-hours are applied to produc-tion of primary or secondary products. Annualphysical output data, on the other hand, usuallyinclude only primary products of an industry.In addition, they are usually reported on a"wherever made" basis. Thus, there can besome discrepancy in the coverage of output andman-hours measures. This is not a seriousproblem unless there is considerable variationfrom year to year in the proportion of primaryproducts to total products of an industry, or ifthere is change in the proportion of primaryproducts which are made in other industries.The cGaiparability of the man-hours and out-put data is indicated by the specialization andcoverage ratios which the Bureau of the Censuspublishes. All industries in the BLS industrymeasurement program have high specializationand coverage ratios.

In selecting industries for the measurementprogram, attention is also given to changes inthe degree of vertical integration. Man-hoursrelate to all operations performed by establish-ments of an industry, while output usually ismeasured in terms of the final product. If es-tablishments undertake additional operations,such as the manufacture of components whichhad previously been purchased from suppliers,man-hours will increase but there will be nocorresponding 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 ofa change in the degree of vertical integration.

Presentation

BLS indexes are published annually in theform of a bulletin, Indexes of Output, Per Man-Hour, Selected Industries. As new industryindexes are developed, they are presented asarticles in the Monthly Labor Review. Thearticles contain an analysis of productivity,output and employment trends in the industry.Technical notes describing the methodologyused to develop the indexes are available onrequest.

Indexes of output per man-hour also arepublished in the- Statistical Abstract of theUnited States and in the Handbook of LaborStatistics. Some indexes for earlier years arepublished in Historical Statistics of the UnitedStates.


Industry measures of output per man-hourare particularly useful for studying changes inmanpower utilization, projecting future man-power requirements, analyzing trends in laborcosts, comparing productivity progress amongcountries, examining the effects of technolog-ical improvements on employment and unem-ployment, and analyzing related economic andindustrial activities. Such analysis usually re-quires that indexes of output per man-hour beused in conjunction with other industry data.For example, to study technological effects,related data on production and employment areuseful ; to study trends hi labor cos:,s, data onearnings and other labor expenditures arenecessary.

Although the measures relate output to oneinputlabor timethey do not measure thespecific contribution of labor, capital, or anyother factor of production. Rather, they reflect

OUTPUT PER MAN-HOUR MEASURES: INDUSTRIES 225

the joint effect of a number of interrelated in-fluences such as changes in technology, capitalinvestment per worker, utilization of capacity,layout and flow of material, skill 'and effort ofthe work force, managerial skill, and labor-management relations. Also, indexes which re-late output to one group of employees representthe total output of the industry resulting fromall employees and do not represent the specificcontribution 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 cf goods andservices produced. Second, although effortshave been made to maintain consistency ofcoverage between the output and labor inputestimates, some statistical differences may re-main. Third, changes in the degree of plantintegration and specialization often are notreflected adequately in the production statis-

tics. This may result in overstatement of pro-ductivity gains in some years, understatementin others. Fourth, indexes involving nonpro-duction worker man-hours are subject to awider margin of error than are the indexesusing production worker man-hours because ofthe technique for estimating average man-hours of nonproduction workers. Errors inestimating man-hours for nonproductibn work-ers, however, have a relatively insignificanteffect on the estimates of man-hours for allemployees. Fifth, year-to-year changes in out-put per man-hour are irregular, and, therefore,not necessarily indicative of basic changes inlong-term trends. Conversely, long-term trendsare not necessarily applicable to any one yearor to any period in the future. Because of theseand other statistical limitations, these indexescannot be considered precise measures; insteadthey should be interpreted as general indicatorsof movements of output per man-hour.


Number1. Dunlop, John T. and Diatchenko, V. Labor Productivity (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 conceptsand measurements of productivity; international comparisons of productivity; wages andproductivity; and technical, managerial, and organizational factors affecting productivity.

2. European Productivity Agency, Organization for European Economic Co-operation, Produc-tivity Measurement, 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 andWages; and Indices of Industrial Efficiency.

3. Evans, Duane and Siegel, Irving H. "The Meaning of Productivity Indexes," Journal of theAmerican Statistical Association, March 1942, pp. 103-111.

The nature, use, and limitations of productivity indexes.4. Fabricant, Solomon. "Of Productivity Statistics: An Admonition," Review of Economics and

Statistics, November 1949, pp. 309-311.The deficiencies and limitations in productivity measurement.

5. Greenberg, Leon. Data for Measurement of Industrial Productivity in the United States, U.S.Department of Labor, Bureau of Labor Statistics, 1965.

A review of the status of available statistical data and determination of what gaps existin order that improvements can be made.

6. International Labour Office. Measuring Labor Productivity, Geneva, 1969. Discusses meth-odology and problems in the measurement of productivity, analysis of national series andinternational comparisons of productivity measures.

7. Kendrick, John W., Postwar Productivity Trends in the United States, 1948-88, NationalBureau of Economic Research, forthcoming.

r.

t.



Presents trends in productivity by industry groupings from 1948-66 with preliminaryestimates 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.

8. Klotz, Benjamin P., Productivity Analysis in Manufacturing Plants, U.S. Department ofLabor, Bureau of Labor Statistics, 1970.

An econometric analysis of industry production functions derived from data for over1700 plants.

9. Klotz, Benjamin P. and Herman, Shelby A., Productivity in the Railroad Industry, U.S. De-partment of Labor, Bureau of Labor Statistics, March 1970.

An analysis of productivity in the railroad industry using econometric techniques.10. Mark, Jerome A. "Industry Indexes of Output Per Man-Hour," Monthly Labor Review, 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.

11. National Bureau of Economic Research, Conference on Research in Income and Wealth. Out-put, Input, and Productivity Measurement, Volume 25 of Studies in Income and Wealth,Princeton, N.J., 1961.

A collection of papers and comments devoted to an appraisal of the measurement ofoutput, input, and productivity.

12. Siegel, Irving H. Concepts and Measurement of Production and Productivity, Working Paperof 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 productionand productivity measurement.

13. U.S. Department of Labor, Bureau of Labor Statistics. Indexes of Output per Man-Hour,Motor Vehicles and Equipment industry, 1957-66, December 1968.

A study of productivity trends in the motor vehicles and equipment industry. Contains adetailed technical note describing the methodology used to derive the measures.

14. United Nations. Index Numbers of Industrial Production, Series F, No. 1 of Studies inMethods, New York: 1961.

Deals with the more technical aspects of compiling production indexes.15. Productivity: A Bibliography (BLS Bulletin 1226, 1958) and (BLS Bulletin 1514, 1966).

Two collections of annotated references concerning productivity and productivity mea-surement.

CHARLES W. ARDOLINI, ARTHUR S. BERMAN, AND JOHN L. CAREY

Chapter 27. Technological Change

Background

Studies of technological changes and theirlabor implications have been uncle :taken by theBLS over the years for a variety of purposes.During the 1930's, public interest focused onthe unemployed and reports were preparedon technological changes and displacement ofworkers in various industries. During WorldWar II, emerging technologies wo, studied ierpurposes of improving msrvi.ioai4i. utilization.

Beginning in the mid-19-6'0.s, nationwide at-tention was focused on the implications of newdevelopments classified under the general term"automation." The BLS made a series of stud-ies on a plant basis, in the insurance, petroleumrefining, bakery, air lines, and electronics in-dustries, to explore the manpower implicationsof various changes. Later, broader studies wereundertaken, including a survey of manpowerimpacts of changeover to electronic computersin 20 large companies and intensive studies oftechnological change in the coal and paper in-dustries.

These studies formed the basis, beginning inthe early 1960's, for a more systematic investi-gation of future changes. Research now under-way pinpoints technologies which will becomeincreasingly important over the next decadein key industries and attempts to provideadvance information about their manpowerimplications.

Description of Studies

The Bureau's research program on tech-nological change involves a variety of reportsand studies of different degrees of detail andapproach. The current program thus providesdetailed case studies of changes within a singleplant or office; summary reports surveyingtrends in major industries; detailed industrystudies; and studies of major technologicalinnovations, such as computers, that affectworkers in different industries.'

I 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 totechnological change. The framework of suchstudies is a single plant or office. A typical casestudy covers such topics as: management'sobjectives and results of introducing technolog-ical change at the plant or office; exent ofdisplacement and reassignment of employees;practices regarding transferring, retraining,and selecting employees for new occupations ;characteristics of employees whose jobs wereeliminated and who were assigned to new posi-tions; and implications of automation for olderemployees. Both qualitative and quantitativedata are presented.

The case study approach has also been usedto investigate in detail special aspects of ad-justments to technological change. Among thetopics covered by case studies are: the per-formance of older workers in industry retrain-ing programs, as shown in four plants indifferent industries; manpower planning pro-cedures followed in connection with a series oftechnological changes at a public utility andin telephone offices; the adjustments to elec-tronic data processing in a government agency;experience under an adjustment program inthe railroad industry; post-layoff work ex-perience of displaced workers in plants in dif-ferent industries and regions; and job redesignfor older workers at different plants. In thesestudies, information about single plants ispresented, as illustrative, rather than as rep-resentative cases.

Reviews of Technological Trends in MajorIndustries

To provide a broad overview of significanttrends in the economy, the Bureau prepares asummary report, applying to key industries onnew types of machinery, processes and prod-ucts which are believed likely to have an im-portant effect over the next 5 to 10 years. Theindustries covered comprise a cross section ofthe economy and include those where the pacemay be slow as well as those where change is

219 227

228 BLS HANDBOOK

rapid. The first version of this report, entitledTechnological Trends in 36 Major AmericanIndustries, was issued by the President'sAdvisory Committee on Labor-ManagementPolicy in 1964. A revised edition covering 40industries was published in 1966.2

The emphasis of the report is on technolog-ical developments within each industry in anearly stage of the innovation's commercialusei.e., the period after introduction on themarket but before widespread adoption. In-ventions and discoveries still in the "drawingboard" stage are considered unlikely to haveas much impact over the next decade as thosealready tested and are generally not discussed.

The report briefly describes recent technolog-ical developments, indicating insofar as prac-ticable some economic advantages of varioustypes of new equipment, processes or products;their importance in terms of the man-hoursengaged in the operations affected; estimatedextent of use currently and in 5 to 10 years,and some factors affecting adoption such asthe volume of investment and expenditures forresearch and development. The advantages de-scribed include not only labor savings perunit, but also quality improvements, fuel andmaterial economies, greater accuracy, newmarkets, etc.

In assessing the employment implications oftechnological changes, account is taken of thepossible rate Of growth in output per man-hourand in the industry's total output. Appraisalalso is made of the changes in occupationalstructure and of some issues and examples ofadjustment that are taking place.

Outlook Reports on Industry TechnologicalTrends

Intensive studies are made of selected majorindustries where far reaching changes, on alarge scale, are taking place, such as coal, rail-roads, and textiles. These studies involve de-tailed analysis of the economic implications ofmajor technological developments within in-dividual industries. Factors analyzed includeinvestment trends and factors affecting the

See technical reference 12.3 See technical references 9, 10. and 12.

OF METHODS

prospects for the diffusion of recent technolog-ical advances, such as the structure of theindustry. Estimates are developed of the dis-placement of present by new methods over thenext 10 years. Unit labor requirements undernew and old technologies are compared,wherever possible. Since the focus of the studyis on the industry as a whole, data on recentindustry trends in output per man-hour, pro-duction, and employment are examined in re-lation to long-term trends and projections offuture trends are developed.

Technological Innovation Studies

Some technical innovations have applicabilityin many industries. Among these are such de-velopments as computers, numerical control ofmachine tools, materials handling equipment,and control instruments .2 Because of their farreaching impact, special studies have beenmade of the nature, status, prospects for adop-tion, and implications for unit labor require.ments, occupational change training needs, andproblems of industrial relations. In analyzingtheir impact in different industries, differencesas well as similarities are revealed.


A variety of data sources and collectionmethods are utilized in making studies of tech-nological change and its impact.

Personal Interviews

In making case studies, analysts personallyconduct intensive interviews with plant man-agers, personnel directors, and other officialswho have direct knowledge of changes at theirplant. Union officials at the plant, and in somecases, individual workers are interviewed. Theanalyst uses a checklist of questions in con-ducting informal interviews in order to elicitthe maximum amount of data. Plants andoffices included in these studies are selected onthe basis of having recenily made a majorchange in their equipment, products, or meth-ods of production.

ei) ti1J=

TECHNOLOGICAL CHANGE

Personal interviews also are utilized to helpdetermine industry trends. Informal inter-views are conducted with engineers, scientists,economists, and other experts in companieswhich produce and use new technology, unions,trade associations, government agencies, uni-versities, etc., who have specialized knowledgeof particular technological development of in-dustry trends. One objective in these cases isto obtain their expert judgment about thenature, pace of introduction, and possible im-pact of developments with which few plantshave had any experience. The emphasis in theseinterviews is on the technological changerather than on experiences in adjusting.

Trade and Technical Publications

Important sources of information concerningtechnical trends are trade journals, technicalmagazines and books, conference proceedings,government hearings, and company reports.Annual reports of leading corporations andcompany house organs often contain usefulinformation on current technical developmentin some industries. In making studies of indus-tries, these publications are reviewed to obtaininformation about the status and prospects ofimportant developments and to ascertain whichcompanies and plants merit more intensive fieldvisiting. Reports and publications of firms thatproduce particular types of equipment oftenare found useful in studies of industries thatuse such equipment.-

Statistical Data Sources

Quantitative information about the status ofspecific technological developments is fragmen-tary and scarce. The Bureau makes use ofavailable data from many public and privatesources. These sources include, for example:Office of Management and Budget; annualinventory of computers in the Federal Gov-ernment; Business and Defense Services Ad-ministration: annual survey of numericallycontrolled machine tools; McGraw-Hill Com-pany: survey of computers in industry; Ameri-can Bankers' Association : survey of bankingautomation; American Machinist Magazine:

. 229

inventory of metalworking machinery; andBureau of Mines : annual reports on miningequipment.

Statistical information on industrywidetrends are useful in analyzing the economicimplications of technological change. Amongthe important sources used in preparing stud-ies are the Bureau's indexes of output perman-hour and related series on production,employment and hours; the Bureau of Censusdata on expenditures on plant and equipment ;and the National Science Foundation's esti-mates of research and development.

Mail Surveys

Because of the complexity of the subject,relatively little use has been made of mailsurveys in studying the impact of technologicalchange. This technique, however, is useful inobtaining information on a broad scale nototherwise available, to supplement detailed in-formation collected through case studies. Thus,to determine the extent and manpower impactof computer use in the insurance industry, theBureau conducted a two-part survey of 400companies employing most of the employeesin 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 agroup consisting of all large offices, and a sam-ple of small companies. From information re-ceived from this screening questionnaire anumber of companies were selected to whom amore detailed questionnaire was sent, askingfor information about computer uses, numberof employees engaged in electronic data proc-essing, planned applications, etc.

The mail survey technique also was used infollowup surveys of workers who have beenlaid off as a result of technological and eco-nomic change.5 The names and addresses ofsuch workers were obtained from plant person-nel records, selecting only a sample in caseswhere a large number of workers had beendisplaced. The questionnaire was mailed atleast 6 months after the layoff occurred, inorder to allow a period of time for some adjust-

"See technical reference 5n See technical reference 9.

221

230 BLS HANDBOOK

ment to take place. Information collected fromthe 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 forcestatus at time of survey, type of job held afterlayoff, source of jobs, etc.

Plant Records

In making detailed studies of the impact oftechnological change on individual workerswithin a plant, analysts sometimes can obtainfrom employers' files, data on such aspects asthe age, sex, and related personal charac-teristics of employees whose jobs are elimi-nated and the jobs in the plant held by eachindividual affected before and after the change;similar data are collected on individuals whoare selected for the positions created in con-nection with automated equipment.°

,-Expert Review

In. preparing forecasts of future technolog-ical trends, a critical step is the review ofpreliminary reports with outstanding expertsin each. industry? Drafts of industry reportsare mailed to company executives, union re-search directors, trade association officials,technical journal editors, and university andgovernment specialists for their assessment ofthe validity and adequacy of projected trends.Over 450 persons were contacted in thi g wayin the preparation of a report on technologicaltrends im major industries. Some experts arevisited personally to review draft statements indetail. Through this means, reports on tech-nological prospects are designed to reflect, asmuch as possible, the authoritative view;; of anumber of persons who have expert, first handknowledge of each industry.

Analysis and Interpretation

For a better understanding of research re-sults in this field, it is important to keep inmind the meaning of certain key ideas and

OF METHODS

concepts. Some of the key problems of interpre-tation and analysis in this type of research aretherefore set forth, briefly.

Definition of Technological Change

Technological change is defined broadly inthe BLS studies, as encompassing significantchanges in processes and equipment, productsand services produced, and materials, fuels, andenergy used. The term "automation, which issometimes popularly used as a synonym for"technological change," designates, strictlyspeaking, a particular type of current de-velopment. It has been variously defined, forexample, as "automatic operation," "the mech-anization of sensory, control and thoughtprocesses," and "a concern with productionprocesses as a system."

While BLS studies have been concerned withdevelopments in "automation," particularly inanticipating long-term trends in the future,they are not the only technological changestaking place that affect labor requirements andindustrial relations. For example, new ways ofgenerating 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 usuallycovered by technical definitions of "automa-tion," but having significant manpower im-plications.

Impact on Productivity

Since one of the principal consequences oftechnological change, so far as manpowerutilization is concerned, is an increase in pro-ductivityi.e., output per man -hour, specialattention is given in BLS studies to analyzingchanges in industrial productivity. Such trendanalysis is a useful method of measuring thepace of technological change. Changes in pro-ductivity, however, also reflect changes incapacity utilization and many other nontech-nical factors; it is important to recognize that

See technical references 1 and 6 for use of plant records.7 See technical reference 12 for further detail.


the productivity trend is only a partial meas-ure of the rate of technological change.

In determining the impact of a specifictechnology, BLS studies try to indicate thereduction in unit labor requirements that thenew processes are designed to achieve. In somecases, estimates of labor savings are derivedon the basis of comparisons with the estimatedaverage technology of the industry understudy; in others, with the best equipment thatis available; or in actual plant studies, withthe technology that is actually displaced.

It is also important to distinguish betweenthe impact on productivity of the operationdirectly affected and on productivity of theplant as a whole. An advanced machine tool,for example, may result in a relatively largereduction in unit labor requirements in themachining operation, but would have little im-pact on finishing, and assembling, and mayeven require additional labor in engineeringand maintenance work.8 The impact on plantproductivity, therefore, would be considerablyless than the effect on productivity of any de-partment or operation directly affected.

Impact on Employment

In assessing the impact of technologicalchange on employment, it is necessary to con-sider the implications of plant manpowerpolicies and the effects of economic changes,with which technical changes interact. Analysisof the impact of technological change purelyin terms of machinery is incomplete.

At the plant level, for example, the substitu-tion of machinery for labor may substantiallyreduce job opportunities in operations directlyaffected. If efforts are made, however, to elimi-nate these jobs by not filling vacancies createdby quits, deaths, and retirement of employees,or by transfer of affected workers to otherpositions in the plant or office, labor savingscould be achieved without displacing the work-ers affected from the plant.

Moreover, the employment impact of tech-nical change is also interrelated with the effectsof the business cycle. Thus, workers whose jobsare eliminated by technical changes may notbe displaced from a plant until a decline in

231

demand result! in layoffsa long time afterthe change has been made in some cases. Inthe subsequent recovery, however, they maynot be hired back because their jobs no longerexist.

Since many changes exert their effects onemployment through the competitive market,the employment trend for the industry as awhole must also be examined. The plant whichreduces its unit costs through technologicalimprovement may be able to gain a larger shareof the market and increase its employment, butat the expense of the less technically advancedcompeting plants, which may be forced to shutdown, displacing workers far from the locationof the change.

Because of the whole complex of economicfactors that operate through the market, in-cluding changes in demand, location, foreigncompetition, merger, and consumer taste, it isvery difficult*to isolate the expanding and dis-placing effects of technological change.

Impact on Occupations

Two aspects of occupaLional change result-ing from technological changes are examined.Changes in job structurethe distribution ofthe plant or office work force by function orbroad skill groupingare studied to determinethe extent of upgrading or downgrading. Sincethe content of jobs may be altered as a resultof changes in equipment or processes, attentionalso is directed to intensive before-and-afteranalysis of job duties and the knowledge andabilities required to perform these duties asindicated by job descriptions and observation.The content of newly created jobs, such asprogrammer, also is studied and the .qualifica-tions required and personal characteristics ofindividuals selected for these new positions aredescribed, so far as possible.°

Adjustments to Technological Change

Technological change has important implica-tions for personnel management and collectivebargaining within plants. The introduction of

See technical reference 10 for further discussion.See technical references 1, 6. and 10.


new machinery, products, or processes oftenrequires movement of workers among jobswithin the plant or office by transfer or promo-tion, the setting of wage rates, and selection ofpersons for new jobs. Often the adjustmentproceeds according to rules established in ad-vance through collective bargaining. Provisionsto assist workers whose j'obs are eliminatedinclude severance pay, retraining and earlyretirement. Besides analyzing the optraticin offormal provisions under collective bargaining,Bureau studies describe informal efforts to pro-vide training, to utilize attrition, and to obtainjobs for displaced workers elsewhere.1° Thelimitations of these measures as well as theiradvantages 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 objectivesand responsibilities of the Manpower Develop-ment and Training Act. Under this act, theSecretary of Labor is required "to establishtechniques and methods for detecting in ad-vance the potential manpower impact of auto-mation, technological progress, and otherchanges in the structure of production." Aspart of such an early warning system, BLSstudies and reports of technological change areuseful to managers, union leaders, educators,economists, government officials, and others inplanning policies to cushion the impact of

10 See technical references 1 and 8 for further discussion.

change. The study of emerging technologicaltrends and possible implications, moreover,provides a basis for more valid projections ofproductivity and economic growth. They alsoare useful in pinpointing manpower problemsand determining the most productive directionof future research to obtain possible solutions.

Some limitations of the Bureau's studies oftechnological change must be kept in mind inassessing their appropriateness for particularuses. In general, it is important to recognizethat judgments about the future direction andpace of technological change and its implica-tions are necessarily complex and difficult. Therate of introduction of new technology dependsnot only on technical advantages- but also onmany economic factors, such as the volume ofinvestment, market prospects, and the avail-ability of trained workers, all of which are sub-ject to significant variations. Moreover, sincethe period of introduction generally spans anumber of years, new developments are con-stantly appearing so that assessments of theoutlook must be reappraised from time to timein the light of new information.

Finally, studies of the impact of technolog-ical change deal primarily with changes withinindividual industries. But these changes ofteninvolve changes in the type and amount ofgoods and services purchased from other in-dustries and could therefore have importantimplications for production and employment inindustries supplying inputs. The accumulationof information on interindustry relationships,through the Bureau's economic growth studies,will provide a quantitative basis for analyzingthis aspect of technological change.

224



Number1. Adjustment to the Introduction of Office Automation, (BLS Bulletin 1276, 1960).

A study of manpower implications of the installation of electronic data processing in20 offices in private industry, with special reference to older workers. Discusses problemsof measuring impact of electronic data processing on employment and occupational re-quirements.

2. Automation: A Discussion of Research Methods, Labour and Automation Bulletin No. 1,Geneva, International Labour Office, 1964.

Fourteen papers on problems of research methods in studying manpower impact ofautomation at the plant and industry level.

3. Automation and Technological Change (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 ofmeasuring employment impact.

4. Case Studies of Displaced Workers, BLS Bulletin 1408 (1964).Studies of post-layoff experiences of nearly 3,000 workers formerly employed at five

different plants. Discusses use of mail quastionnaire in followup of displaced workers.6. Impact of Office Automation in the Insurance Industry, (BLS Bulletin 1468, 1966).

Mail survey of over 400 insurance companies. Covers extent, pace, and employment im-plications of electronic data processing.

6. Impact of Office Automation in the Internal Revenue Service, (BLS Bulletin 1364, 1968).Detailed case study of large-scale changeover to electronic data processing. Migrates

use of internal personnel records in analyzing effects of office automation.7. Industrial Retraining Programs for Technological Change, BLS Bulletin 1368 (1963).

Four case studies of experience of older and younger workers in industrial retrainingprograms, based on plant records. Discusses some problems of measuring comparative per-formance.

8. Manpower Planning for Technological ChangeCase Studies of Telephone Operators, (BLSBulletin 1574, 1968).

Case studies of the manpower policies and experiences of several telephone companies incushioning the impact of technological change on their employees.

9. Outlook for Computer Process Control, (BLS Bulletin 16E8, 1970).A study of manpower implications of the introduction of computers to control produc-

tion processes in six process industries.10. Outlook for Numerical Control of Machine Tools, (BLS Bulletin 1437, 1965).

A study of manpower implications of a key technical innovation affecting metalworkingindustries. Discusses problem of generalizing about change in productivity as result ofspecific change.

11. Technology and Manpower in the Textile Industry of the 1970'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.12. Technological Trends in Major American Industries, (BLS Bulletin 1474, 1966).

Description of outlook for major technological developments based on a variety of datasources and expert review.

EDGAR WEINBERG

225

233

'4.

Chapter 28. Construction Labor Requirements

Background

New construction is a major component ofthe Nation's output of goods and services anda major source of employment. The jobs itcreates occur not only at the site of employ-ment, but also in the many manufacturing,trade, transportation, and service industrieswhich furnish the materials and services re-quired in the construction process. Because ofthis far-reaching employment impact, the crea-tion of new construction projects often isregarded as a means of counteracting cyclicalunemployment.

To assist in assessing the extent of the impactof construction expenditures on employment, aseries of labor and material requirements stud-ies for different types of construction wasstarted in 1959. The program was establishedas a result of Congressional legislation, requir-ing the Bureau of Labor Statistics to provideestimates of the amount of total employmentgenerated directly and indirectly by variouskinds of construction per dollar of expenditure.

Earlier Bureau efforts to relate employmentand volume of construction included the "LaborRequired for New Construction" series, withreference mainly to on-site man-hours. Thisseries, started in the early thirties, appearedintermittently through the years, but was notbased on actual up-to-date surveys and wasfinally halted in the mid-fifties because thefactors employed were found to be obsolete.There was also a series of Put; lie Works Ad-ministration studies published in 1940, cover-ing Federal public works projects constructedin the mid-thirties; and a few individual stud-ies of specific types of construction.

The present studies include the major typesof building construction (schools, hospitals,public and private housing, etc.) and also heavyconstruction (highways, dams, etc.). However,only one or two selected types of constructionare surveyed in any given year. Selected typesof construction are resurveyed periodically.These resurveys may, in addition to providingcurrent information on labor requirements,contribute information useful in preparing con-

struction cost indexes and estimates of changesin productivity of on-site construction labor.


The surveys are designed primarily to deter-mine the number of man-hours represented bya fixed dollar volume ($1,000) of contract con-struction. Man-hours, as defined by the surveys,include both on-site construction employmentand the off-site employment required to pro-duce and deliver materials used in the con-struction. Data for on-site labor include totalman-hours for the supervisory, engineering,clerical, and custodial employees, as well asthose for workers in each construction tradeat the site of construction. Data for off-sitelabor include employment in the off-site (e.g.,office and warehousing) activities of construc-tion contractors ; in building materials andequipment manufacture and distribution; andfinally, employment in all the other industrieswhich are affected directly or indirectly by theproduction and distribution of building ma-terials from raw materials to the final manu-factured product. Man-hours are also estimatedfor the employment created by overhead ex-penditures of contractors.

Certain types of employment are not coveredby the survey. For example, no estimate ismade of the employment used in the planningand designing of the projects studied. Alsoexcluded are estimates of employment requiredin government and public utility agencieswhich might be affected by the constructionbeing studied. Employment created by the re-spending of wages and profits of the workersand their employersthe multiplier effectare not considered within the scope of thestudies.


The surveys cover construction which isnonfederally assisted and/or that which maybe totally or partially financed through Federal

0123V/236


funds. Although the type of construction laborand material requirements data sought aresimilar for both federally and nonfederallyaided projects, the sources for the data aredifferent, particularly for the on-site man-hourinformation.

For the construction of nonfederally aidedprojectsthose financed entirely by variouslevels of local government, private individualsor voluntary groupson-site man-hour data,by occupation, are obtained by field represents -,tives from local authorities, architects, con-tractors, and other direct participants in theprojects. These individuals provide the desiredman-hour information from payroll records,daily work force reports, or summary timereports.

For those projects financed wholly or in partby Federal funds, on-site employment informa-tion generally is obtained from contractors'payrolls submitted to the government underregulations of the Davis-Bacon Act or otherFederal legislation covering federally aidedconstruction.

These payrolls furnish the data for estimat-ing on-site man-hour requirements, as well asdata on wages for all hourly rated workers onthe 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 possiblethe collection and presentation of additionaldetail for the projects studied. This includesinformation on wage relationships, timing ofconstruction operations, and requirements bytype of contractor.

Information on material and equipment costsfor the projects studiedi.e., the items whichare used for estimating 8ff-site employmentis collected by the field representatives fromthe individual contractors and subcontractorsengaged in the construction of the sampleprojects.


Sampling procedures vary with the type ofconstruction being studied. The "universe" ofprojects for a specified study generally repre-

sents all of the projects known to have beencompleted during a selected 1-year period.

The projects are then stratified into cellshaving similar characateristics which mayaffect 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)' and (e) amount of total constructioncontract (i.e., size of project).

One or more projects are selected at randomfrom each cell and assigned weights to givetotal representation of the cell. Substitutions,when necessary, are made from the same cells.

Data from the sample projects are combinedin aggregate form to represent the total dollarvolume of each cell.

As indicated previously, the off-site employ-ment estimates are derived from the materialsand equipment cost information obtained fromthe contractors and subcontractors cooperatingin the studies. The contractors provide a list ofthe value of each type of material used in theconstruction of sample projects. These materiallistings are classified into categories consistentwith a 4-digit Standard Industrial Classifica-tion code as used in Census of Manufacturesproduct groups. For each of these productgroups, average amounts of material (in dol-lars) required for each $1,000 of contractconstruction cost are determined. The value ofmaterials is reduced by a ratio representingthe difference between valuation by the pur-chaser and valuation by the producer. (Thisratio is based on valuation data provided bythe U.S. Department of Commerce, Office ofBusiness Economics.) This latter step is re-quired because all data reported by contractorsare in purchasers' value, and reduction to pro-ducers' value is necessary to obtain figuresconsistent with Census data which are usedto calculate manufacturing employment.

Primary employment in manufacturing isconsidered to be that required to produce theconstruction bill of materials in their finalstage of fabrication. In this stage, man-hoursare developed by multiplying average pro-

' For examPle, elementary or secondary schools.

Irt.(7'7

rF

4

CONSTRUCTION LABOR REQUIREMENTS

ducers' value of each construction material bythe ratio, of manufacturing man-hours per$1,000 of production. (This ratio is establishedby using the Census of Manufactures.)

Primary man-hours in the trade, transporta-tion, and service industries are estimated fromthe difference: between producers' and pur-chasers' value for each construction material.The value differences are alloted to trade andtransportation, and primary man-hours foreach component of transportation and tradeare then estimated from labor factors providedby BLS.

Secondary employment is defined as the em-ployment in all industries involved in produc-tion and transportation of building materialsand equipment from basic extraction to, butnot including, the final manufacturing stage.The Interagency Input-output Study is used toobtain these estimates.2 For each group of ma-terials, the interindustry study provides infor-mation on the amount of secondary productsrequired from each of its 78 industry sectors.The product data are converted to man-hoursby use of output per man-hour ratios for eachof the sectors. Adjustments for price and pro-ductivity are made to provide estimates con-sistent with the year of construction and bill ofmaterials.

For each off-site stage (primary and secon-dary), a man-hour figure per $1,000 for theconstruction being studied is obtained. Whenthese 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, isdetermined.

An exact study of the off-site employment ofeach construction contractor is not attempted,since it is almost impossible to relate accuratelysuch employment to the projects studied. Build-ers' off-site employment is occupied not onlywith the sample projects studied, but also withother current or future projects. The estimateof contractors' off-site man-hours for each$1,000 of construction contract is based on thedifference between construction worker em-ployment and total employment in the con-

See "Input-output Structure of the U.S. Economy: 1963."Survey of Current Business. November 1969. pp. 16-47.

237

struction industry, with adjustments foron-site supervisory and administrative employ-ment.


The construction labor and materials re-quirements studies are published in Bureaubulletins. Summary articles, based on the sur-vey findings, appear in the Monthly LaborReview. The bulletins and articles highlight thetotal man-hours generated per dollar volume ofconstruction expenditures, with subtotals foron-site and off-site man-hours. In presentingthe labor and materials data, the statisticaltabulations are supplemented with an analysisof the various factors which apparentlyaffected the man-hour requirements for thespecific types of construction studied. The bul-letins contain information on the characteris-tics of the sample projects and the man -hoursper square foot and per $1,000 of contractcost, with an analysis of the variations in re-quirements and costs arising from differencesin design, type of structure, and regionalor metropolitan versus nonmetropolitan in-fluences.

The reports include information and analysisof : on-site man-hour requirements by occupa-tion; the employment share of the general andspecial trades contractors engaged in the con-struction work ; direct on-site wage cost ; thedistribution of employment by periods of con-struction time; and the cost of major materials.

When feasible, comparisons of unit man-hourand material requirements for earlier periodsare made and analyzed.


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, congressionalcommittees, and industry research and tradeorganizations, to assist in gauging the impactof planned expenditures for construction onemployment, and the economy, generally. Ofspecial interest to market research analysts

22811-0


and companies manufacturing materials foruse in construction, are the materials listingsper $1,000 of construction contract.

While the overall estimates of employmentare believed to be reasonably accurate, the de-tailed data would have a wider margin ofsampling error and may be subject to other

limitations. Man-hour and material require-ments are affected by a number of factors suchas location, size of project, type of structure,architectural design, availability of certain ma-terials or equipment, labor skills, and localbuilding codes and customs. The effects of theseseparate factors cannot be isolated.

Technical ReferencesLabor and Material Requirements Studies

Number1. Civil Works Construction by the Corps of Engineers (BLS Bulletin 1390, 1964).

A study of on-site and off-site man-hour and wage requirements for dredging and land-type projects in the U.S. Corps of Engineer& civil works program from 1959 to 1960.

2. College Housing Construction (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 primarilyto determine the man-hours required for $1,000 of college housing construction.

3. Construction of Federally Aided Highways, 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 censtruction. For on-site construction, the study also provides a comparisonof annual man-hour requirements for 1947-64.

4. Federal Office Building Conatruction (BLS Bulletin 1331, 1962).A statistical study of on-Bite and off-site labor requirements for constructing 22 Fed-

eral office building projects in various localities of the United States over a 3-year periodfrom the fall of 1957 to 1960.

5. Hospital Conatruction (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 theUnited States between mid-1958 and mid-1959.

6. -Hospital and Nursing Home Construction (BLS Bulletin 1691, 1971).A study similar to the 1962 study but with data shown per square foot as well eh per

$1,000 of construction contract.7. Private One-Family Housing Conatruction (BLS Bulletin 1404, 1964).

j A statistical study of on-site and off-site labor requirements for constructing a sampleof one-family houses built in 1962 in various localities of the United States.

8. Public Housing Construction (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 variousStates to be representative of four broad geographical regions of the conterminous United

tates.9. Sc ool Construction (BLS Bulletin 1299, 1961).

A study of primary and secondary man-hours required per $1,000 of new school con-st ction based on contracts awarded throughout the United States for 85 elementary and43 junior and senior high schools.

10. School Construction (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 studyalso includes data on the types and value of materials used, wages paid, occupations dis-tributed and use of apprentices.

11. Sewer Works Construction (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.

MARTIN ZIEGLER

e-V74911

Occupational Safety and Health

Chapter 29. Occupational Safety and Health Statistics

Background

Passage of the Occupational Safety andHealth Act of 1970 (Public Law 91-596)marked a major departure in the collection ofwork-injury statistics. The act is comprehen-sive; it provides that every place of employ-ment subject to the act shall be free fromrecognized hazards which are likely to causedeath or serious physical harm. To assist inattaining this goal, the act provides for record-keeping and reporting procedures which willidentify the seriousness of on-the-job accidentand job related illnesses.

The Bureau of Labor Statistics has beenassigned responsibility for developing and im-plementing the statistical program which isrequired to achieve the objectives of the act.

The Bureau has been concerned for manyyears with standardizing the methods for com-piling work-injury statistics. As early as 1911the Bureau called a formal conference to dis-cuss the matter. The work of the conferencewas continued by the International Associationof Industrial Accident Boards and Commissions,resulting in publication of the first standardizedprocedures in .1920.1 In 1937 the first work-injury standard was published by the AmericanStandards Association, now the American Na-tional Standards Institute. The most recentrevision is the Standard Method of Recordingand 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 aletter to the American National Standards In-stitute that the proposed legislation includeda national system for the collection of safety

Standardization of Industrial Accident Statistics (BLSBulletin 276, 1920).

2 Proposed National System for Uniform Recording andReporting of Occupational Injuries and Illnesses, AmericanNational Standards Institute, New York, New York, 1970.

and health statistics. He requested that theInstitute evaluate whether the standard methodcontained in Z16.1-1967 was appropriate forthe broad universe of employers who would besubject to the proposed legislation. An informalconference called by the Institute concludedthat there was cause to believe that the Z16.1standard was inappropriate for the proposeduse and recommended that a study group beformed to consider the matter further and, ifnecessary, develop a simple method of report-ing injuries. This recommendation was basedupon a belief that the Z16.1 standard : (1) hadgrown too complex, through efforts to makeit equitable, to form a basis for a mandatorynational reporting system, (2) did not ade-quately reflect trends in injury experience foremployers with good safety records, (3) wasnot 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 andillnesses was needed. The study group's pro-posal was published by the Institute in Decem-ber 1970.2 The recordkeeping regulations whichhave been issued by the Secretary draw heavilyon the proposals of the study group.

Records Required to be Kept

The Occupational Safety and Health Act of1970 directs the Secretary of Labor to issueregulations which require all employers sub-ject to the act to maintain accurate recordsof work related deaths, injuries, and illnesses.However, records do not have to be maintainedfor minor injuries requiring only first aidtreatment and which do not involve medicaltreatment, loss of consciousness, restriction ofwork or motion, or transfer to another job.

The regulations concerned with recording ofoccupational injuries and illnesses (29 CFRPart 1904) became effective on July 1, 1971.They require employers to maintain in eachestablishment a log of occupational injuries

230289

240 BLS HANDBOOK

and illness. Within 6 working days of receiv-ing information that a recordable case has oc-curred, the employer must enter the factscalled for in the log. At the end of each yearflu. employer is required to complete an an-nual summary of occupational injuries and ill-nesses. Information for the summary is drawnfrom the log. The summary is to be preparedno later than one month after the close of theyear and is to be posted prominently in eachestablishment in a place accessible to the em-ployees.

The log and summary must be kept availablein the establishment for a period of 5 yearsfollowing the end of the year to which theyrefer. In addition, a supplementary record mustbe maintained for recordable injury or illness.A form is provided for that purpose (OSHAForm 101) but in most cases a workmen's com-pensation, insurance, or internal form whichmust be completed for other purposes will bean acceptable record in place of form 101.

All logs, summaries, and supplementary rec-ords must be available in the establishment forinspection and copying by compliance officersor statistical agents of the Federal or StateGovernments.

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 ona calendar year basis. Collection will be pri-marily by mail questionnaire. The respondentwill merely transcribe the injury and illnessdata from the summary form to the question-naire. The questionnaire will also seek informa-tion about the numbers of employee-hoursworked and a statement of the product of theestablishment.

Coverage

Data will be collected from a representativesample of establishments. The sample for thefirst report period (July through December1971) is designed to produce injury data atmost of the 2-digit SIC industry levels. The

OF METHODS

sample for the first full calendar year (1972)of operation under the act will be approxi-mately 200,000 establishments, providing dataat the 4-digit SIC industry level in manu-facturing and at the 3-digit level in nonmanu-facturing.

State Participation

The act encourages the States to assumeresponsibility for the administration and en-forcement of occupational safety and healthlaws, and for carrying out an occupationalsafety and health educational program.

The Secretary of Labor is authorized tomake grants to the States to assist them indeveloping and maintaining an effective pro-gram of collection, compilation, and analysisof work-injury and illness statistics. A con-siderable number of States are expected toparticipate in these grant programs.

Existing Work-Injury Data

Work-injury data have been collected by theBureau for many years. These data, which havebeen collected through 1970, will not be com-parable to data compiled under the Occupa-tional Safety and Health Act. Methods used inthe collection of these data are described in theprevious edition of this Handbook. (See Chap-ter 26 of the Handbook of Methods for Surveysand Studies, U.S. Department of Labor, Bureauof Labor Statistics, Bulletin 1458, 1966).

Description of Surveys and Methods

Programs for the collection, tabulation,analysis, and publication of occupational safetyand health statistics are being formulated. De-tails of these programs will be published assupplements to the BLS Handbook of Methodsprobably about mid-year 1973. These supple-ments can be obtained after mid-1973 fromthe Bureau of Labor Statistics, U.S. Depart-ment of Labor, Washington, D.C., 20212, orfrom any of its regional offices listed on theinside back cover of the Handbook.

231LYLE R. SCHAUER

ii

Due to budget cuts and personnel ceilings, Labor Developments Abroad, basic countrystudies, labor digests, and country or area bibliographies will be discontinued by June30, 1972.

Economic Trends and Labor Conditions

Chapter 30. Foreign Labor Conditions, InternationalComparisons, and Trade Research

Background

Almost from its inception, the Bureau ofLabor Statistics (BLS) conducted research onlabor conditions and developments abroad. TheBureau carries on two research activities ofparticular interest to users of research reports :analysis of research data on labor in foreigncountries and the preparation of international

'comparisons.' The Bureau undertook these re-search programs because (1) summary anddetailed information on labor conditions pub-lished by a majority of foreign countries is notavailable in English (or in any form readilyusable by U.S. employers, labor unions, Gov-ernment officials, and others) ; (2) users cannotkeep pace with the variety of source materialsfrom nearly 200 independent countries and de-pendencies; and (3) often, none but an expertcan judge the quality of source materials.

Description of Reports

The foreign labor research reports publishedby the Bureau are in part general and in partstatistical. The principal types may be listedas follows:

1. Labor Developments Abroad, a monthlypublication which covers important develop-ments in labor in foreign countries and pro-vides data on living costs abroad, furnished bythe 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. policyand program officers on labor in foreign countries. and pro-viding training or orientation to foreign statisticians andother foreign visitors. especially participants in exchangeprograms and technical cooperation program, on U.S. laborstatistics methods and on the economic conditions of U.S.workers.

2. Basic country studies, primarily mono-graphs on labor law and practice in selectedcountries. These reports are factual, but notprimarily statistical.

3. Brief labor digests on single countries,published either separately or with directoriesof labor organizations in major geographicareas.

4. Bibliographies, chiefly for a country oran area, often with annotations or briefsummaries.

5. International comparisons presented infeature articles and bulletins. These are mainlystatistical in nature and contain explanationsof the source data and, where possible, adjust-ments for differences in definitions and meth-ods. So far, principal comparisons publishedhave concerned hourly compensation, outputper man-hour, unit labor costs, unemployment,and price trends.

6. Reports on trade and labor developmentsin selected industries, including the relation-ship of imports to domestic employment.

The Bureau also provides data in responseto individual inquiries from Government of-ficials, U.S. businessmen, labor unions, andstudents.

Data Sources and Collection MethodsThe Bureaux receives much materialde-

scriptive, statistical, and analyticalon labordevelopments abroad, which it maintains infiles classified by country and by subject. Thematerial includes (1) current reports fromlabor attaches and other officers of the U.S.Foreign Service throughout the world ; and(2) foreign data from various sources. Thelatter include periodicals and other publica-tions issued by agencies of foreign countries(labor ministries, bureaus of statistics, and

241


others) ; international organizationsgovern-mental, business, and laboras well as the In-ternational Labour Organization ; and privateagencies like banks and economic researchorganizations in the United States and abroad.When specific supplementary information isneeded, the Bureau often requests it throughForeign Service reporting channels.

Staff members keep informed by contact withresearch institutions, with individual scholarsspecializing in international labor, and withU.S. firms employing labor abroad. In addi-tion, they are briefed by labor attaches andother Foreign Service officers when theyvisit Washington. Also, from time to time,members of the research staff study foreignlabor conditions at first hand by travel abroad.

Analysis and Uses

With the exception of international com-parisons, the major focus in the presentationof reports is geographic, although an occasionalsummary by topic is used. Within an individualcountry study, the organization is by subject.For the labor law and practice series, subjectsinclude : (1) social, economic, and politicalbackground ; (2) manpower (including laborforce, employment, unemployment, manpowerplanning and utilization); (3) labor law andadministration; (4) wages, supplementarybenefits, 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 thebasic situation at the time the report waswritten or with a recent development or trend.With a few exceptions, of which real wages isone, less attention is given to hbitorical de-velopment of long-term trends. The extent ofmaterial covered is so great that often only themost summary information for a given countrycan be written more often than every 10 years;many countries have never been covered inindividual monographs. Reports often containan annotated bibliography for readers in-terested in more detail.

International comparison studies were begunin 1960 on a regular basis. They deal with in-dividual subjects and are primarily statisticalrequiring specific information about details ofdefinition, and information is seldom coveredfully by regularly published sources. Therefore,great reliance is placed by the Bureau on re-quests to the Foreign Service and statisticalagencies abroad for information specificallyrequired.

Among subjects covered are internationalcomparisons of labor costs and productivity.For instance, articles containing basic infor-mation on "The Role of Labor Cost in ForeignTrade" and on "International Comparisons ofUnit Labor Cost : Concepts and Methods" havebeen published. Absolute comparisons havebeen issued also for the iron and steel industryin four countries, as well as trend data in indexform for all manufacturing in ten principaltrading natictns and in the United States. Rela-tive labor cost trends are useful for analysisof trade competitiveness, the balance of pay-ments, and inflationary developments. Othertopics studied have included unemployment,work stoppages, job vacancies, and pricetrends.

The greatest problem involved in comparingunit labor costs among countries arises fromthe fact that labor cost data are available byindustry, but output is reported by product.Since nearly all industries produce a numberof 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 comparablefrom country to country, and because morethan one set of data exists for most free indus-trial countries, considerable effort is spent inanalyzing the individual series and selectingthose most consistent and valid. In most coun-tries where it was available, the Bureau de-cided (1) to use as the measure of labor costthe compensation of employees in manufactur-ing, from the national accounts, and (2) to useas the measure of total output the real gross

g33,

FOREIGN LABOR CONDITIONS, INTERNATIONAL COMPARISONS, AND TRADE RESEARCH 243

in approach. The methods are often intricate,national product originating in manufacturing.The chief reason for this decision has been thegreater comparability and comprehensivenessof these measures.

As regards unemployment, the interest ininternational comparison centers on the per-centage of the labor force unemployed. Datapublished by the individual countries on a cur-rent basis are of different types. However,most industrial countries of the free worldhave made one or more sample surveys of theirlabor forces, using definitions and methodsfairly comparable tc those in the United States.The Bureau uses these latter data (adjustedas carefully as feasible to uniform definitionsand brought up to date by the best data cur-

rently available) as the basis for internationalcomparisons of unemployment rates.

A program is under development for con-ducting studies of the relationship of foreigntrade to U.S. labor. Analyses will be preparedfor selected industries in order to understandthe factors contributing to employment changesand the effects of rising imports on employ-ment. Since the relationship of trade to em-ployment is usually indirect and difficult tomeasure, a multiple approach will be taken,including review of plant closings, masa layoffsof employees, and the effects of changes indemand, imports, and productivity on employ-ment.

-JULIET F. KIDNEY AND JOHN CHANDLER

Chapter 31. Economic Growth Studies

Background

The Bureau of Labor Statistics has de-veloped a program of economic growth studiesaimed at providing a more comprehensive andintegrated framework for analyzing the prob-lems of long-run economic growth in relationto employment opportunities. Because thisprogram has important implications for manyagencies of the Government, an interagencyplanning and coordinating committee providesthe guidelines for the program. The chairmanof the committee is from the Council of Eco-nomic Advisers; other members come from theBureau of Labor Statistics, U.S. Departmentof Labor ; Office of Business Economics, U.S.Department of Commerce; and the Office ofManagement and Budget.

A primary objective of the program is todevelop projections, under alternative assump-tions, of the rate and pattern of economicgrowth in the United States. These provide aframework for assessing a number of impor-tant economic problems, including problems ofmanpower utilization over the next decade.

Methods

An economic growth model has been de-veloped to serve as a tool in making economicprojections. This model begins by developinga potential economic growth rate for the UnitedStates. To do this, it must project the laborforce to the target year, assuming a specifiedunemployment rate, and project the rates ofchange in productivity and average hours paid.This potential growth in gross national prod-uct (GNP) is distributed among the four majorcomponents (or demand categories) of GNP:consumer expenditures; domestic investment;government expendituresFederal, State, andlocal; and net foreign demand.

The next stage is to develop a projection ofthe industry composition of demand for eachof the four major demand categories. For ex-ample, the consumer expenditures categoryincludes the amounts spent on rents, automo-

biles, medical expenses, and other goods andservices purchased by consumers. For each ofthe four major demand categories, a differentprocedure is followed in allocating demand tothe producing industry. The industry detail towhich the categories of demand are allocatedmatches the input-output classification.

Demand as used in the national income ac-counts refers only to final demand, i.e., that ofthe ultimate consumer. To place a value on theoutput of an industry whose products are notsold to ultimate consumers, but are used in-stead by other industries in the course of theirown production, an addition set of calculationsis necessary using input-output. The input-output system translates final demand for agiven product into the output that is requiredfrom all other industries to produce the ma-terials needed to manufacture that product orservice.

The third stage is to project the input-outputcoefficients. The input-output tables used as abase in the economic growth model are pub-lished by the Office of Business Economics, U.S.Department of Commerce. However, theseinput-output tables incorporate the technologyand product mix for a base year, and may notreflect adequately the technology and productmix which may prevail during the period forwhich the projection is being made. To accountfor this difference, two methods are used: De-tailed analyses are made of the changes takingplace in the technology of various industriesas well as the changes expected in product mixdue to differing growth rates of product groupswithin industries. For industries for whichdetailed stu.lies are not made, analyses aredeveloped to determine the direction and mag-nitude of change in the use of its products byother industries.

Next, employment estimates by industry aredeveloped. This is accomplished by use of a setof industry productivity projections. The finalstage is to balance the model. Projections con-tain many complex relationships among eco-nomic variables that were developed througha lengthy sequence of operations. It is neces-sary to have a set of checks and balances toinsure that the various states of the projections

-,-s11045drip orlt,


make up an internally consistent model. Theeconomic growth model is designed to providea feed-back and balancing procedure with re-spect to three of its elements: imports, invest-ment, and employment. In practice, all three ofthese elements must be brought into balancesimultaneously.

Uses

The projections developed in the economicgrowth program serve a number of uses. Theemployment projections by industry are usedin developing occupational outlook projections.The projections developed by the Bureau forman important part of the U.S. Government'sreport to OECD on long-term economic outlookfor the United States. In addition, other Gov-

Technical

ernment agencies use various facets of the eco-nomic growth projection to develop projectionsin their specific areas of responsibility. Outsideof government, the projections of GNP and in-dustry growth patterns are important sourcesof information for industry analysts. Their useof these projections are primarily in diversifi-cation studies, market analysis, and long-termcapital planning.

The structure of the economic growth modelis developed so that other analytical use is madeof the program. Specifically, the program isused to analyze the industry by industry effectof changing levels and patterns of various typesof demand such as defense expenditUres. Simi-lar types of analysis also is performed for ex-ports and imports to provide information on theemployment requirements for foreign trade.

References

Number1. U.S. Department of Labor, Bureau of Labor Statistics, Patterns of U.S. Economic Growth:

1980 Projections of Final Demand, Inter-industry Relationships, Output, Productivity, andEmployment (Bulletin 1672, 1970).

2. Projections 1970: Intei-industry Relationships, Potential Demand, Employment (Bul-letin 1536, 1966) .

3. Oliver, Richard P., "Increase in Defense-Related Employment During Vietnam Buildup,"Monthly Labor Review, February 1970, pp. 3-10.

4. Bowman, Charles T., "Report on Employment Related to Exports," Monthly Labor Review,June 1969, pp. 16-20.

RONALD E. KUTSCHER

r.

Appendix A. The BLS Seasonal Factor Method

Background

An economic time series may be affected byregular intra-yearly ("seasonal") movementswhich result from climatic conditions, modelchangeovers, vacation practices, holidays, andsimilar factors. Often such effects are largeenough to mask the short-term underlyingmovements of the series. By isolating and re-moving the effect of such intra-yearly repeti-tive movements, the current evaluation of aseries may be made more perceptive.

Seasonal movements are found in almost alleconomic time series. They may be regular, yetthey do show variation from year to year andare also subject to changes in pattern overtime. Because these intra-yearly patterns arecombined 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 withexactness.

More than a half-century ago, attempts weremade to isolate seasonal factors from timeseries. Some early methods depended uponsmoothing curves by using personal judgment.Other formal approaches were periodogramanalysis, regression analysis, and correlationanalysis. Because these methods involved alarge amount of work, relatively little applica-tion of seasonal factor adjustment procedureswas carried out.

In the mid-1950's, new electronic equipmentmade more elaborate approaches feasible inseasonal factor methods as well as in otherareas. Using a computer, the Bureau of theCensus developed seasonal factors based on aratio-to-moving-average approach. This was amajor forward step, as it made possible theuniform application of a method to a largenumber of series at a relatively low cost.' Sub-sequent improvements in methods and in com-puter technology have led to more refinedprocedures which are both faster and cheaperthan the original technique.

The Bureau of Labor Statistics began itswork in seasonal factor methods in 1959, pri-marily to correct a deficiency in the method

then used. Prior to this time, when additionaldata became available and seasonal factorswere generated from the lengthened series, thenew factors sometimes differed markedly fromthe corresponding factors based on the shorterseries. This difference could affect any portionof the series. It was difficult to accept a processby which the addition of recent informat:oncould affect significantly the seasonal factorsfor periods as much as 15 years earlier, es-pecially since this meant that factors couldnever become final. The first method de-veloped by BLS and introduced in 1960 hadtwo goals : first, to stabilize the seasonal fac-tors for the earlier part of the series; second,to minimize the revisions in the factors for therecent period.

Since 1960, the Bureau, through continuedresearch, has made numerous changes and im-provements in its techniques and in methods ofapplying them. These changes were'described 2as they were introduced. The method intro-duced in May 1966 is described in the sectionswhich follow.

Characteristics of BLS SeasonalFactor Method

The BLS method is a ratio-to-moving-average method. It assumes that the threecomponent partstrend-cycle, seasonal, andirregularare multiplied together (multiplica-tive assumption) to give the original observa-tions.3 (See illustration in charts 1 and 2.) TheBLS method differs from other similar methodsin the following respects:

1. The initial trend cycle is improved byrestoring to it any residual trend-cycle whichmay have found its way into the irregular com-ponent. This adjustment for the deficiency intrend-cycle is developed explicitly and is avail-

Julius Shiskin, Electronic Computers and Business Indica-tors. New York, National Bureau of Economic Research,(1957). Occasional Paper No. 57.

'See U.S. Department of Labor items in the list of Tech-nical References at the end of this appendix.

*A parallel method assuming an additive relationshipamong the components also has been prepared for experi-mental use.

237247

248 BLS HANDBOOK

able for review in the form of a table of trend-cycle correction values.

2. The BLS method provides for changesover time in the producing mechanism (changesin samples, method of collection, method ofestimation) or in economic factors (introduc-tion of guaranteed annual wage plans, etc.)'

3. The BLS method calibrates each observa-tion in order to assign a supplementary weightwhich is used in the various averaging proc-esses. These "credence factors" reduce theeffect of observations having large irregulari-ties. They increase the smoothness with whichthe seasonal factors change over time, and theyalso keep large irregularities out of the finaltrend-cycle.

4. A second way in which the BLS methodattempts to protect the final trend-cycle fromlarge irregular fluctuations is by using modifiedoriginal values for computing the centered12-month moving average in the later stages ofthe procedure. The credence factors are used inobtaining the modified original values.

The BLS method is very complex and re-quires an enormous amount of arithmeticalcomputations. Therefore, the method has beenadopted for use on electronic computers. Atpresent, two program decks are available ; onewritten in FORTRAN II and another inFORTRAN IV. Special emphasis has beenplaced on keeping the handling and clericalrequirements to an absolute minimum, on pro-viding as many aids as possible for analysts orothers using the results, and on making theapplication of the computer program as simple

4A 5-year moving period is used to reflect such changes.The use of the 5-year moving period allows the full impactof a change to be reflected in a relatively short period oftime. 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 wasin the early 1950's. This reflects the improvements made inthe survey, such as the expansion in the number of sampleareas, the increase in the number of households covered, thechanges in the estimating' methods, and the improvements inthe design of selecting h ,useholds so that there would be athree-quarters overlap in adjacent months and a one-halfoverlap over the year.

Upon request, the Bureau of Labor Statistics will make itsseasonal factor computer program available in the followingform: (1) A source deck: of punched cards containing theprogram instructions; (2) A deck of punched cards contain-ing a text problem which will enable the user to verify theoutput: (3) Two copies of a document containing operatingand reference material.

OF METHODS

and efficient as possible. Only the data and onespecification card are needed to produce a setof completely labelled tables for each series.Any number of series with varying character-istics (length of series, starting and endingdates, magnitude of original observations, finaltable patterns desired) can be processed in asingle running without manual intervention.The computer program for applying the BLSprocedure is available on request.5

Basic Approach

The BLS method attempts to separate aneconomic time series into three constituentparts : the underlying movement or trend-cyclewhich is a combination of the long-term,cyclical, and subcyclical movements (T) ; theannual repetitive movement within the year,or "seasonal" (S) ; the irregular or residualunexplained movement (I). These three com-ponents, when multiplied together, completelyand exactly exhaust the original observations(0). (See charts 1 and 2.) The exact allocationamong 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 goodbalance between the conflicting objectives ofsmoothness in the trend-cycle, stability in theseasonal, and randomness in the irregular. Theprocess is an iterative one; each successiveiteration provides an improved estimate foreach of the components of the original series.

Three iterations are used, each having twophases. The first phase of each iteration startswith a centered 12 -month moving average asthe estimate of trend-cycle. Seasonal-irregularsare then secured and partitioned into seasonaland irregular components. The second phase ineach iteration starts with an improved trend-cycle in which the centered 12-month movingaverage has been improved by recovering theresidual trend-cycle from the irregulars of thefirst 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 phaseof each iteration to reduce the effect of large

38

0(0

ON.

Co

nla A

X t'q4

p4s

0z

wtn cl)

la4

E-4

'74

Tt;

KC

a.0a)

UI

rn

THE BLS SEASONAL FACTOR METHOD

ul 0CN1

to 0 to0

0N

;

0)fA CD

a)co

CD

coco

toCD

CD

07CO

-4,04,1;.

Y*.1

0CO

4tirc!;..41161,.?

zoi

V9=11,,i1W,W.T-,40

751dee9;1

o .4eet<.::v ki,,P1

... to p toco

...447,07=CYCZI

t<VTA.VIL.--h

0

44,4;:ttVt.

?: 6 tonron-..r...wuot vow» WVcfla

2,39

0

a)a)

co

en

Co

CO

0)

249

250

(.2

ON


0ODD NO

1/

0 0 0 0. Ni M 0

//N

C 0

0))-

0

01

0)co

i "...7.7.)

PO

nj it)

0 -5(A'

0

yr N.wb A

4' 1-4

7.0 ore U3

P 4 14

U2

to0

44ro

eia)

5

0tD

rn

tr.

qttat.f;Y-1.4:e37: k3ivw

eZ1 ;14

-41:411$::S.:;(3

2

0 0(3) 038 0 0 0 0

N 52,

-1-1;°1.;

0 0 8 8°

0)a,a)nr0)

5

THE BLS SEASONAL

irregularities in the original observations. Theseare supplementary weights in which observa-tions with small0 irregulars are given moreweight than observations with large irregularsin calculating the trend-cycle and the seasonalfactors.

Detailed Procedure of BLS SeasonalFactor Method

First iteration

First Phase. The initial estimate of trend-cycle(T) is the centered 12-month moving averageof the original observations. The trend-cyclevaluta for each month is divided into its cor-responding original value (0) to produce aseries of seasonal-irregular (SI) ratios(0/T=S1). Treating each month separately(i.e., all January's, all February's), the SIratios are arrayed by years and moving 7-termweighted averages 7 (S') are secured (in per-centage form) as estimates of the unforcedseasonal factors. The unforced seasonals (S')for each year are then adjusted to yield amonthly average of 100.0 for the calendaryear.R These forced seasonal factors (S) arethe seasonal factors for the first phase of thefirst iteration. Each seasonal-irregular ratio(SI) is then divided by its forced seasonalfactor (S) to yield the random or irregularratio (S//S=/). The irregular ratios at thisstage may contain some residual trend-cycle.°To separate the residual trend-cycle from thetruly random variation contained in the ir-regulars, a 9-term weighted moving average 10(trend-cycle correction) of the first phase ir-regulars is secured. This completes the firstphase of the first iteration.Second Phase. The second phase of the firstiteration starts with an improved estimate ofthe trend-cycle. This is secured by multiplyingthe trend-cycle used in the first phase (thecentered 12-month moving average of originalobservations) by the trend-cycle correction(weighted moving average of first phase ir-regulars). Using this improved trend-cycle, thesecond phase repeats the computational stepsof the first phase to develop new SI ratios, new

FACTOR METHOD 251

unforced seasonals (S'), new forced seasonals(S), and new irregulars (I). At this point, thegenerated trend-cycle and seasonal componentsrepresent the components of the original seriesfairly well except for the effect of highly

"References to "large" or "small" irregulars have to dowith the deviation of the irregular component from 1.000 inrelation to the standard deviation of such irregulars. A valueof 1.423 or .577 would usually be considered large, while .997or 1.003 would be small.

T The weight patterns used are:

Weight pattern assigned toseven consecutive

Seasonal seasonal-irregular ratios1st year _ .281 .270 .242 .207 ___ _2nd year .213 .221 .213 .191 .162 ___3rd year _ _ .160 .179 .10.5 .179 .160 .137Middle years .120 .141 157 .164 .157 .141 .1203rd from end .137 .160 .179 .185 .179 .1602nd from end .162 .191 .213 .221 .213End year .207 .242 .270 .281

The underlined value indicates the year to which theweighted average applies.

*The forcing is performed in two siages: If the unforcedseasonals do not start in January, the first 12 unforced sea-sonals are summed and the total divided into 1200 to providea forcing factor. This factor is then multiplied by the un-forced seasonals for the partial year only (through the firstDecember value) to provide the forced seasonals for theincomplete at the beginning of the series. A similar pro-cedure is followed at the end of the series if the unforcedseasonals do not end in December. For each of the fullcalendar years between, a forcing factor is computed bydividing the sum of the unforced seasonals for the year into1200. This factor is then multiplied by the unforced seasonalsin that year to produce the forced seasonals.

'This is because the 12-month moving average is not verygood at following abrupt or curvilinear changes in the levelof the original series, particularly in the vicinity of peaksand troughs. Since the trend-cycle, seasonal, and Irregularcomponents completely and exactly exhaust the originalseries, any deficiency of the first estimate of trend-cycle istransferred 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 sevendifferent years, with the result that the deficiency of the12-month moving average, as an estimator of trend-cycle,appears mainly in the irregular component. The periodswhere the 12-month moving average is a poor estimator oftrend-cycle, usually have a run of consecutive irregularratio all on the same side of the base reference ratio of 1.000instead of being scattered above and below this base.

1"The weight patterns used are:

Month Weight pattern assigned to nine consecutive irregulars1st month _ .33Q .341 .256 .115-.082 _.2nd month .254 .276 .254 .191 .086-.081 ___3rd month . .160 .214 .231 .214 .160 .072-.051 ___

4th month _ .067 .150 .199 .216 .199 .150 .067-.048Middle

months ____-.050 .071 .157 .209 .226 .209 .157 .071-.0504th month

from end . -.048 .067 .150 .199 .216 .199 .150 .067

3rd monthfrom end ___-.051 .072 .160 .214 .23/ .214 .160

2nd monthfrom end .086 .191 .254 lig .254

End month __ _ _ _-_-.082 .115 .258 .341 .171The underlined weight indicates the month to which the

weighted average applies.

241


deviant original values. In the BLS method,the influence of such values is diminished byusing supplementary weights in addition to theregular weights given in footnotes 7 and 10.The use of these supplementary weights (called"credence factors") gives less impact to theextreme observations (those with large irregu-lar components) and relatively more to theneighboring values with smaller irregular com-ponents."

With the preliminary credence factors, theSI ratios of the second phase are repartitionedinto seasonals and irregulars. The adjusted un-forced seasonal is a 7-term moving average ofthe SI ratios, using as relative weights theproduct of the standard 7-term weights (givenin footnote 7) and the preliminary credencefactor associated with each value. These un-forced seasonals (S') are forced in the usualmanner to total 1200 for the calendar year,after which adjusted irregulars are calculatedby dividing the adjusted forced seasonals (S)into the SI values.

This repartitioning of the SI ratios removesthe large irregular variation from the seasonalcomponent and puts it in the irregular com-ponent instead. However, the trend-cycle is stillcontaminated by deviant original values, be-cause it was calculated before the credencefactors were developed.

The second iteration is designed to providean improved 12-month average for use in de-veloping revised components. In preparationfor this, the adjusted irregulars of the firstiteration are used to develop revised (inter-mediate) credence factors. These intermediatecredence factors are computed from the ad-justed irregulars in exactly the same way (de-scribed in footnote 11) as the preliminarycredence factors are obtained from the earliersecond phase irregulars.

The first use of the intermediate credencefactors is to create a modified original serieshaving no large irregularities. Each originalvalue having a credence factor less than 1.000is replaced by a modified value in which theirregularity has been reduced." The creationof the modified original series marks the endof the first iteration.

Second iteration

First Phase. The second iteration is similar tothe first iteration, except that the intermediatecredence factors are used along with the weightpatterns of footnotes 7 and 10. The seconditeration, like the first, has two phases. Thefirst phase begins with the centered 12-monthmoving average of the modified original seriespreviously described. SI ratios are secured bydividing the actual original series (not modi-

* fled) by this 12-month average. The SI ratiosare arrayed by month and moving 7-term aver-ages, using the weights of footnote 7 and theintermediate credence factors, are taken toyield unforced seasonals (S'). (The credencefactors prevent extreme observations fromaffecting the seasonals.) The forcing process isthen applied to yield seasons (S) which aver-age 100.0 for the calendar year. Irregulars (I)are secured by dividing the latest SI ratios bytheir corresponding forced seasonals (S).These irregulars may include some residualtrend-cycle because of the failure of the 12-month moving average to fully penetrate the

It The credence factors are computed as follows. Find. atest is made to locate extreme irregulars falling outside ofthe 3 sigma limit. These irregulars are replaced by a valueof 1.000 for a new sigma calculation. A moving 61-termstandard deviation of the irregulars is computed for calibrat-ing the irregular associated with the middle (31st) term. Forthe 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 differencefrom the mean of the 61 terms used to secure its standarddeviation, 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 is1.000. For a standardized irregullar of 2.800 or more, thecredence factor is 0.000. For a standardized irregular between1.000. For a standardized irregular of 2.800 or more thestandardized irregular. The 2.8 sigma limit makes it ex-tremely unlikely for a "good" value, not affected by a strikeor other such aberration, to be assigned zero credence. Onlyone-half of one percent of the values in the normal distribu-tion lie beyond this limit. On the other hand, "bad" valueswhich deserve to be disregarded have a much higher prob-ability of falling outside the limit.

" The amount of reduction for each observation is suchthat the deviation of its new irregular from 1.000 is theproduct of the credence factor and the deviation of its oldirregular component. Thus, each modified original value isthe product of the trend-cycle and seasonal components de-veloped in the first iteration, times an irregular which hasbeen scaled down if it exceeded one standard deviation. Tnemodified original values are used only to secure an improved12-month moving average with which to start the seconditeration; seasonal-irregulars are always calculated from theactual original values.

Iv jgV

THE BLS SEASONAL FACTOR METHOD 253

peaks and troughs of the modified originalseries. A trend-cycle correction is computed byarraying 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 thefirst phase of the second iteration.Second phase. The second phase of the seconditeration starts with an improved trend-cycle,which is the product of the centered 12-monthmoving average of the modified originals andthe trend-cycle correction previously described.The second phase repeats the steps and pro-cedures of the first phase to develop new SIratios, new unforced seasonals (S') makinguse of the intermediate credence factors, newforced seasonals (S), and new irregulars (I).

These second phase irregulars are used tocalculate final credence factors in the same wayas 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 finalcredence factors. This completes the seconditeration.

Third iteration

The third iteration carries the refinementprocess still further. It follows the same stepsas in the second iteration, from the centered12-month moving average of the newly modi-fied original series up to the development of theirregular component near the end of the secondphase. This completes the partitioning of theseries into the final trend-cycle, seasonal, and

1 Additional iterations yield little further modification. Thedecision to stop with 3 iterations was based on the verysmall changes occurring after the third iteration, the rea-sonable fit of the trend-cycle to the original data, and thecost of additional processing.

irregular components." As supplementary in-formation to aid in analysis and evaluation, thefinal irregular is standardized using 61-termmoving sigmas, which are printed also. Theseasonally 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 beaware that the result of combining series whichhave been adjusted separately will usually be alittle different from the direct adjustment ofthe combined series. For example, the quotientof seasonally adjusted unemployment dividedby seasonally adjusted labor force will not bequite the same as when the unemployment rateis adjusted directly. Similarly, the sum of sea-sonally adjusted unemployment and seasonallyadjusted employment will not quite match thedirectly adjusted labor force. Separate adjust-ment of components is usually preferable iftheir seasonal patterns- are different, providedthe increased measurement errors in the com-ponents are not excessive and that the amountof work does not proliferate unduly.

Finally, it is worth noting that the avail-ability of a fast, efficient procedure for makingseasonal computations can easily lead to theprocessing of large numbers of series withoutallotting enough time to review the results. Nostandard procedure can take the place of care-ful review and evaluation by a skilled analyst.A subjective review of all results is stronglyrecommended. The computer program for ap-plying the BLS method facilitates such reviewby providing the needed materials in a logicaland easily used format.



Number1. Barton, H. C., Jr., "Adjustment for Seasonal Variation," Federal Reserve Bulletin, 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., The Smoothing of Time Series. 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, Seasonal Adjustment on Elec-

tronic Computers, 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. Includestheoretical sections relating to calendar (trading day) variation and general properites ofmoving averages.

4. Shiskin, Julius, Electronic Computers and Business Indicators, New York, National Bureauof Economic Research (1957). Occasional Paper No. 57. Also published in Journal of Busi-ness, Vol 30, October 1957, pp. 219-267.

Describes applications of the first widely used computer program for making seasonaladjustments.

5. U.S. Department of Labor, Bureau of Labor Statistics, The BLS Method of Deriving Sea-sonal Factors. Paper presented at the Interstate Conference on Labor Statistics, Newport,R.I., June 16, 1960.

6. U.S. Department of Labor, The BLS Seasonal Factor Method; paper presented at the 1960Annual Meeting of the American Statistical Association, Palo Alto, Calif., Aug. 23, 1960.

7. U.S. Department of Labor, New Seasonal Adjustment Factors for Labor Force Components;Special Labor Force Report No. 8 (1960). Includes article by Morton S. Raff and Robert L.Stein published in the Monthly Labor Review, August 1960, pp. 822-827.

8. U.S. Department of Labor, The Bureau of Labor Statistics Seasonal Factor Method; Paperpresented 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 Factor Method (1963Revision); paper presented at the Interstate Conference on Labor Statistics, San Fran-cisco, Calif., June 27, 1963.

10. U.S. Department of Labor, The BLS Seasonal Factor Method, Its Application by an Elec-tronic Computer (1963).

11. U.S. Department of.Labor, The BLS Seasonal Factor Method (1964). (1964).12. U.S. Department of Labor, The BLS Seasonal Factor Method (1966). (1966).

2 4

Appendix B. Industrial Classification

Much of the usefulness of BLS statistics pre-sented by industries is due to the fact that theycan be compared with other types of data forthe same industries. This is possible becauseBLS and other Federal and State agencies fol-low as closely as possible a single system todefine and classify industries in the U.S. eco-nomy. The Government publishes a StandardIndustrial Classification Manual (SIC) of in-dustries based on principles set forth by atechnical group made up of government andindustry experts) The Bureau of Labor Sta-tistics took part in the development of the SICover a long period of years and continues towork actively with the Office of Managementand Budget and other agencies in seeking toimprove the system.

Four basic principles were followed in de-veloping the classification :2

(1) The Classification should conform to theexisting structure of American industry. (2)The reporting units to be classified are es-tablishments, rather than legal entities orcompanies. (3) Each establishment is to beclassified according to its major activity. (4)To be recognized as an industry, each groupof establishments must have significance fromthe standpoint of the number of persons em-ployed, volume of business, and other impor-tant economic features, such as the numberof establishments.

As there are thousands of products and ac-tivities, the system provides for grouping theseinto categories, both narrow and broad, to en-hance the value of industrial statistics forusers interested in different levels of detail.8

Using the SIC as a guide, the Bureau classi-fies the reports received from each factory,shop, or store according to major product oractivity. The SIC is used in the same way bythe agencies supplying the Bureau with itsuniverse lists and benchmark data. Hence, ahigh degree of orderliness and consistency isattained, which benefits not only the users ofall BLS establishment statistics, but also theusers 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 statisticalprogram to another. For example, there maybe differences in the way in which "major ac-tivity" is determined; or changes in the majoractivity of individual establishments whichoccur over time may be handled in statisticaltime series in different ways. Consequently, theuse of the same manual and following a com-mon set of principles of application do notalways 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 normalmethod of handling industrial classificationwill 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 isstated as follows:

An 'establishment' is an economic unit which pro-duces goods or servicesfor example, a farm, a mine,a factory, a store. In most instances, the establishmentis at a single physical location; and it is engaged inonly one, or predominantly one, type of economic ac-tivity for which an industry code is applicable.

Where a single physical location encompasses two ormore distinct and separate economic activities for which

Office of Management and Budget, Standard industrialClassification Manual, 1987.

'Ibid. pp. IX. X.'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. Contractconstruction; D. Manufacturing: E. Transportation, com-munication, electric, gas, and sanitary services; F. Wholesaleand retail trade; G. Finance, insurance, and real estate; H.Services; L Government; and .1. Nonclassifiable establish-ments. At the 2-digit level all products and services arecombined into 99 "major groups." Thus, in the ManufacturingDivision, establishments engaged in manufacturing machinery.apparatus, and supplies for the generation, storage, trans-mission, transformation, and utilization of electrical energyare combined into Major Group 38, "Electrical machinery.equipment, and supplies."

The 3-digit level provides several hundred categories. Inthe electrical machinery major group the SIC provides 8groups of industries; 361. Electric transmission and distribu-tion equipment; 362. Electrical industrial apparatus; 303.Household appliances; 304. Electric lighting and wiring equip-ment: 365. Radio and television receiving sets, exceptcommunication types: 368. Communication equipment; 307.Electronic components and accessories; 369. Miscellaneouselectrical machinery, equipment, and supplies.

Thousands of products and activities are distinguished atthe 4-digit level. For example, in Group 367, five industriesare defined: 3671. Radio and television receiving type electrontubes. except cathode ray; 3872. Cathode ray picture tubes;3673. Transmitting, industrial, and special purpose electrontubes; 3674. Semiconductors and related devices: 3879. Elec-tronic components and accessories, not elsewhere classified.

245255


different industrial classification codes seem applicable,such activities should be treated as separate establish-ments and classified in separate industries, provided itis determined that: (1) such activities are not ordi-narily associated with one another at common physicallocations; (2) no one industry description in the Stan-dard Industrial Classification includes such combinedactivities; (3) the employment in each such economicactivity is significant; and (4) reports can be preparedon the number of employees, their wages and salaries,

and other establishment type data. An establishment isnot necessarily identical with the business concern orfirm, 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 ofan establishment are included in the industry descrip-tions of the Standard Industrial Classification.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 thesenew States, the effect on national time serieswas small. Where statistics are published byStates, data for these States and other smallStates are shown where reliable data areavailable.

Generally speaking, BLS assigns an estab-lishment to the State in which its employeesare reported for payroll tax purposes, sincethese 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 notonly on a national and State basis, but also forselected metropolitan areas. Comparability ofarea statistics among Federal agencies is main-tained by means of a set of published standarddefinitions.' Defining metropolitan areas forstatistical purposes is done on the basis of in-formation about population and about economicand social ties among cities. To qualify as aStandard Metropolitan Statistical Area undercurrent rules, an area must have at least onecity with 50,000 or more inhabitants, or twocontiguous cities meeting certain other cri-teria.' Which counties are to be included in thestandard area is determined on the basis ofcriteria relating to metropolitan character andextent of economic and social communicationamong counties.'

The Bureau of Labor Statistics is repre-sented on the Federal Committee on Standard

Metropolitan Statistical Areas, and plays anactive role in the Technical Committee on AreaDefinitions. Both interagency groups are underthe chairmanship of the Office of Manaiementand Budget.

State Economic Areas

State economic areas are relatively homo-geneous subdivisions of States developed by theU.S. Department of Commerce's Bureau of theCensus, in cooperation with the U. Depart-ment of Agriculture's Bureau of AgriculturalEconomics, and several State and private agen-cies. They consist of single counties or groupsof counties which have similar economic andsocial characteristics. The boundaries of theseareas have been drawn in such a way that eachState is subdivided into relatively few parts,with each part having certain significant char-acteristics which distinguish it from adjoiningareas.4

-ROBERT B. STEFFES

I U.S. Office of Management and Budget, Standard Metro-politan Statistical Areas, 1967.

I 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 singlecommunity with a combined population of at least 50,000, thesmaller of which must have a population of at least 15,000.

2. If two or more adjacent counties each have a city of50,000 inhabitants or more (or twin cities under 1 (b)) andthe cities are within 20 miles of each other (city limits tocity limits), they will be included in the same area unlessthere 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.XXVII.

* U.S. GOVERNMENT PRINTING OFFICE, 11173 0-430-7110

247257

BUREAU OF LABOR STATISTICS

REGIONAL OFFICES

ostoo

*

San Francisco

etolsci°9"REGION II

U.S.

VIRGIN ISLANDS

Qo

HAWAII

DallasREGION VI

Region I

1603-JFK Federal BuildingGovernment CenterBoston, Mass. 02203Phone: 223-6762 (Area Code 617

Region II341 Ninth Ave., Rm. 1025New York, N.Y. 10001Phone: 971-5405 (Area Code 212)

Region III406 Penn Square Building1317 Filbert St.Philadelphia, Pa. 19107Phone: 597-7796 (Area Code 215)

Region IVSuite 5401371 Peachtree St. NE.Atlanta, Ga. 30309Phone: 526-5418 (Area Code 404)

Region V8th Floor, 300 South Wacker DriveChicago, Ill, 60606Phone: 353-1880 (Area Code 312)

Region VI1100 Commerce St., Rm. 6B7Dallas, Tex. 75202Phone: 7493516 (Area Code 214)

Regions VII and VIIIFederal Office Building911 Walnut St., 10th FloorKansas City, Mo. 64106Phone: 374-2481 (Area Code 816)

Regions IX and X450 Golden Gate Ave.Box 36017San Francisco, Calif. 94102Phone: 556-4678 (Area Code 415)

Regions VII and VIII will be serviced by Kansas City.Regions IX and X will be serviced by San' Francisco.

ell

PUERTO RICO

document resume ed 067 505 bls handbook of methods for … · *methodology; *national surveys;...

Documents

document resume ed 067 505 bls handbook of methods for … · methodology; national surveys;...