Automation and the Workplace: SelectedLabor, Education, and Training Issues

March 1983

NTIS order #PB83-191320

Library of Congress Catalog Card Number 83-600508

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

Preface

This Technical Memorandum is an interim product of OTA’s assessment, “Com-puterized Factory Automation: Employment, Education, and the Workplace.” The assess-ment is examining the nature and development of automation technologies (such asrobotics, computer-aided design and manufacturing, and automated materials handling,storage, and retrieval). The assessment is also evaluating the structure and competitiveconduct of industries producing and using programmable automation technologies. Final-ly, the implications of the production and use of programmable automation for laborand for education and training activities are being examined. The Joint Economic Com-mittee, the Senate Committee on Labor and Human Resources, the Senate Committeeon Commerce, Science, and Transportation, and the Subcommittee on Labor Standardsof the House Committee on Education and Labor requested the assessment.

This Technical Memorandum discusses procedures for evaluating potential employ-ment change associated with automation, and outlines associated problems. It also pro-vides descriptions of the nature and modes of delivery of education, training, and retrain-ing for persons holding or seeking employment in manufacturing industries. The materialdraws in part on the products of a July 1982 OTA workshop on labor markets andindustrial relations and an August 1982 survey commissioned by OTA of education,training, and retraining activities and trends.

OTA is grateful for the assistance of the assessment advisory panel, the Labor Mar-kets and Industrial Relations Workshop participants, the contractors, and the manyothers who provided advice and information. However, OTA assumes full responsibilityfor this technical memorandum, which does not necessarily represent the views of indi-vidual members of the advisory panel.

JOHN H. GIBBONSDirector

. . .Ill

OTA Automation and the Workplace Advisory Panel

Roy Amara, Panel ChairmanPresident, Institute for the Future

William D. BeebyPresidentWilliam Beeby & Associates

Erich BlochVice President, Technical Personnel DevelopmentIBM Corp.

Barbara A. BurnsManufacturing Technology Group EngineerLockheed Georgia

Jack CahallManager, Training and DevelopmentCincinnati Milacron, Inc.

Dennis ChamotAssistant Director, Department for

Professional EmployeesAFL-CIO

Robert ColeProject Director, Joint U.S.-Japan Automotive StudyUniversity of Michigan

Alan E. DraneManager of Flexible Automated SystemsEmhart Corp.

Audrey FreedmanChief Labor EconomistThe Conference Board, Inc.

Sheldon FriedmanDirector, Research DepartmentUnited Automobile, Aerospace and Agricultural

Implement Workers of America

Theodore W. KheelOf CounselBattle, Fowler, Jaffin & Kheel

James F. LardnerVice President, Governmental Products and

Component SalesDeere & Co.

Eli LustgartenVice PresidentPaine Webber Mitchell Hutchins, Inc.

M. Granger MorganProfessor, Engineering and Public PolicyCarnegie Mellon University

George J. PoulinGeneral Vice PresidentInternational Association of Machinists and

Aerospace Workers

Bernard M. SallotExecutive Director, Robot Institute of AmericaSociety of Manufacturing Engineers

Harley ShaikenResearch Fellow, Science, Technology,

and Society ProgramMassachusetts Institute of Technology

Kevin G. SnellVice President, Forward Planning and DevelopmentCareer Works, Inc.

Alfred P. TaylorManager, Factory Automation Planning Services

OperationGeneral Electric Co.

Philippe VillersPresidentAutomatix, Inc.

Victor C. Walling, Jr.Coordinator, Business Futures ProgramSRI International

Dennis WisnoskyVice President, Industrial Systems GroupGCA Corp.

Michael J. WoznyProfessor and Director, Interactive Computer

Graphics CenterRensselaer Polytechnic Institute

Robert ZagerVice President for Policy Studies and Technical

AssistanceWork in America Institute

iv

OTA Automation and the Workplace Project Staff

John Andelin, Assistant Director, OTAScience, Information, and Natural Resources Division

Fred W. Weingarten, Program ManagerCommunication and Information Technologies

Marjory S. Blumenthal, Project Director*

Beth A. Brown, Senior Analyst

Jean E. Smith, Analyst

Margaretta McFarland, Research Assistant

Elizabeth A. Emanuel, Administrative Assistant

Shirley A. Gayheart, Secretary

Theresa Richroath, Secretary

Patricia Gleason, Temporary Secretary

Jeanette Contee, Wordprocessor**

Contractors

Veronica F. Nieva and Robert SchrankAnn Majchrzak, WESTAT

Publishing Staff

John C. Holmes, Publishing Officer

John Bergling Kathie S. Boss Debra M. Datcher Joe Henson

Doreen Cullen Donna Young

*Assistant Project Director until September 1982. Zalman A. Shave]] served as Project Director from April to August 1982.**Until October 1982.

Labor Markets and Industrial Relations Workshop Participants

Eileen Appelbaum*Department of EconomicsTemple University

William Cooke*Krannert School of ManagementPurdue University

Fay Duchin*Institute for Economic AnalysisNew York University

Alan FechterNational Science Foundation

Sheldon FriedmanUnited Automobile Workers of America

Louis Jacobson*Center for Naval Analyses

Robert Levy*Center for Naval Analyses

Michael PilotBureau of Labor Statistics

Markley Roberts*AFL-CIO

Myron RoomkinKellogg School of ManagementNorthwestern University

Jim SmithThe Rand Corp.

Robert WestbrookCommunications Workers of America

● Contributed papers appear in app. C.

vi

Contents

Chapter Pagel. Introduction and Concepts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3

Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3Programmable Automation Technologies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4Dimensions of Technological Change . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5

2. Labor Markets and Working Environment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11Potential for Occupational Change: An Overview.. . . . . . . . . . . . . . . . . . . . . . . . . . . . 11Occupational Change Forecasting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12Beyond Hardware and Software: Other Factors To Reconsidered . . . . . . . . . . . . . . 19Labor Supply . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21Working Environment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25

3. Education, Training, and Retraining . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31Technology and Manufacturing: Changes in Instructional Requirements . . . . . . . . . . 35New Types of Instructional Programs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37Survey of Current Views of Education, Training, and Retraining Requirements . . . 41Selected Critical Issues for Instructional Programing . . . . . . . . . . . . . . . . . . . . . . . . . . . 43

AppendixesA. Survey Methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47B. Industrial Relations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53C. Papers Prepared for Workshop . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60

List of Tables

Table No. Pagel. A Comparison of 1980 and 1970 Decennial Census Occupational Categories . . . . . . . 132. Comparison of Robot v. Human Skills and Characteristics . . . . . . . . . . . . . . . . . . . . . . . 163. Noninstitutional Population and the Labor Force, 1929-82. . . . . . . . . . . . . . . . . . . . . . . . 234. Wage and Salary Workers in Nonagricultural Establishments, 1929-82 . . . . . . . . . . . . . 24S. Labor Management Committees on Industrial Relations Issues, Safety,

and Productivity by Industry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26

Figure

Figure No, Pagel. Bureau of Labor Statistics Employment Projections System . . . . . . . . . . . . . . . . . . . . . . . 18

vi i

Chapter 1

Introduction and Concepts

Chapter 1

Introduction and Concepts

OVERVIEWProgrammable automation technologies are at-

tracting attention as outgrowths of the evolutionof computer and communications technologiesand as instruments of potentially far-reachingchange in the operations, structure, competitive-ness, and hiring patterns of many industries, par-ticularly in manufacturing. While popular recog-nition of programmable automation seems to beconfined mostly to one of its forms (robotics),programmable automation comprises other typesof hardware, software, and systems. * The fami-ly of programmable automation technologies, asapplied in manufacturing, is the subject of anOffice of Technology Assessment (OTA) study,“Computerized Factory Automation: Employ-ment, Education, and the Workplace, ” which isscheduled to be completed in late 1983. Thistechnical memorandum, which is an interim prod-uct of that assessment, presents a set of conceptsand background materials that are fundamentalto the analysis of the labor and education andtraining implications of programmable automa-tion technology.

The OTA assessment is examining the develop-ment and production of programmable automa-tion technologies and their use in discrete productfabrication and assembly. It is examining the ap-plication of programmable automation to the en-tire manufacturing process, from design throughproduction. The assessment is concerned in partwith the economic and social aspects of the pro-duction and use of programmable automation,including:

• impacts on the types and mix of productsmanufactured,

● the structure and competitive behavior ofmanufacturing industries,

*Besides robotics, programmable automation includes computer-aided design (CAD) and manufacturing (CAM), computer-aidedprocess planning (CAPP), automated materials handling (AMH),and automated storage and retrieval systems (AS/RS).

●

●

the numbers and skill mix of peopleemployed in manufacturing, working condi-tions in manufacturing jobs, andthe education and training requirements im-plied by growth in the production and useof programmable automation.

Although it addresses the potential of program-mable automation for the manufacturing sectoras a whole, the assessment highlights implicationsfor the transportation equipment, industrialmachinery, and electronics industries, where thegreatest impacts may occur in the next 10 to 15years.

Early work on this assessment revealed thatanalysis of employment change depends critical-ly on methodology, while analysis of instructionalrequirements demands appreciation of the existingnature of, and delivery system for, education andtraining. These fundamental issues are the sub-jects of this technical memorandum. The remain-der of this introduction provides a brief reviewof the evolution of programmable automation andsets out several factors that influence the socialand economic consequences of new technologies.The labor chapter (ch. 2) discusses methodologyand provides background material useful for eval-uating employment and working environmentchanges. It draws on the products of a workshopheld by OTA in July 1982, where questions con-cerning the analysis of labor markets and in-dustrial relations were debated. The education andtraining chapter (ch. 3) examines the current statusof education and training provided by schools,labor, industry, and others to persons holding oraiming to hold jobs in manufacturing industries.It draws on the results of an August 1982 surveyconducted by an OTA contractor.

‘Exploratory Wcwkshop on the Social Lmpacts of Robotics—ABackground Paper (Washington, D. C.: U.S. Congress, Office ofTechnology Assessment, February 1982), OTA-BP-CIT-11.

3

4

PROGRAMMABLE AUTOMATION TECHNOLOGIES

Programmable automation may be viewed asthe latest development in a long process of en-hancing and augmenting human labor with vari-ous devices. Throughout history, people havecombined human effort and skill directly with thecutting and shaping abilities of tools. With thedevelopment of machines, people drew on me-chanical and other external sources of power,reducing the amount of human effort, and to someextent skill, needed for production. Automation,in turn, represents an advance over simple ma-chines consisting of a transfer of skills and effortsfor operating and controlling equipment and sys-tems from people to machines. Conventionalautomation has improved production efficiencywhere automated machinery has been tailored tospecific applications and devoted to the produc-tion of single products produced in large quan-tities. Programmable automation, which wedscomputer and data-communications capabilitiesto conventional machine abilities, increases theamount of process control possible by machinesand makes possible the use of single pieces ofequipment and systems for multiple applications.This flexibility may make programmable automa-tion more economical than conventional automa-tion across a range of applications from large vol-ume production to production of small batchesof products. Consequently, differences betweenlarge-scale, batch, and even custom productiontechniques may diminish and traditional ways ofthinking about manufacturing may ultimatelychange.

Programmable automation technologies are notnew, at least in concept; they have been intro-duced and refined over the past two to three dec-ades. Many date the launch of programmableautomation to the mid-1950's, when numericalcontrol (NC) for machine tools (currently consid-ered as part of computer-aided manufacturing)was developed and commercialized. The interven-ing years have seen growth in the capabilities anduse of NC, the introduction of industrial roboticsin the 1960’s, and the initial applications of com-puter technology to manufacturing design, pro-duction, planning, and analysis in the 1960’s and1970’s. During this period, capabilities and appli-

cations for programmable automation havegrown, while associated unit costs—at least forthe computer aspects— have declined.2 The tech-nologies and their potential markets appear tohave developed sufficiently to lead many manu-facturing industry analysts to anticipate sub-stantial growth in the production and use of pro-grammable automation in the 1980’s and 1990’s.However, current use of programmable automa-tion is limited. For example, while the Robot Insti-tute of America reports that less than 5,000 robotswere believed to be in use in the United States in1981, the National Machine Tool Builders’ Associ-ation reports that over 2.6 million machine toolswere in use in U.S. metalworking industries aloneby the late 1970’s.3 4

At this time it is possible to identify four attri-butes of programmable automation, as comparedwith conventional automation, that may havemajor ramifications for labor and for educationand

•

●

●

●

training:

capacity for information processing as wellas physical work, in connection with suchprocesses as planning, routing, design, fabri-cation, assembly, monitoring, and diagnos-ing process problems;capacity for quality enhancement, throughreliability, precision, and adaptive control ofthe production process;capacity for application to the production ofa diverse mix of products, through repro-grammability; andcapacity for integrating production equip-ment and systems with each other and withdesign, analysis, inventory, and other aspectsof the manufacturing process.

These attributes will influence: 1) the types andthe range of human skills and other abilities thatcan be replaced by machines, 2) the types of new

2“SME Golden Anniversary Issue, 1932-1982: A Review of Manu-facturing and the Society Which Guides Its Progress,” ManufacturingEngineering, January 1982.

3 WorMvi& Robotics Survey and Directory (Dearborn, Mich.:Robot Institute of America, 1982).

4Ekonomic Hmdbook of the Machine Tool Industry (McLean,Va.: National Machine Tool Builders Association, 1982-83).

5

applications within which both men and machinescan be combined, 3) the types of skills requiredto produce programmable automation, 4) thetypes of products (existing and new) for whichprogrammable automation may be used, 5) thecosts of producing given quantities of differentproducts, and 6) the organization and manage-ment of the manufacturing process. Consequent-ly, they may give rise to changes in the numbersand types of people employed, and thereforechanges in requirements for education, training,and retraining. Distinctive attributes of program-mable automation will also influence the work-ing environment of people employed in manufac-turing. How much,

DIMENSIONS

ment patterns and working environments changewill depend on how automated equipment andsystems are designed and implemented.

Conventional automation and other types ofmanufacturing technologies have traditionally af-fected—both positively and negatively—the em-ployment and working environment of manualworkers. Because of its capacities for performinginformation processing work and for integratingthe manufacturing process, programmable auto-mation may also have significant impact on othertypes of workers, the so-called white- and gray-collar workers, including managers.

an-d in what ways employ-

OF TECHNOLOGICAL CHANGEIt is possible to relate the emerging capabilities

of programmable automation technologies tochanges in employment, and therefore educationand training requirements, in the abstract. How-ever, the effects of programmable automation onlabor overall, as well as the experiences of specificgroups of people, depend on how programmableautomation technologies are implemented. Thedevelopment and implementation of programma-ble automation, or other new technologies, canbe appraised according to three factors: 1) the rateof technological change, 2) the nature of thechange, and 3) the pattern of technological diffu-sion associated with programmable automation.These factors, which reflect a combination of tech-nological and industrial/economic factors, arecentral to assessments of the social and economicimpacts of new technology. They are reviewedbriefly below.

Rate of Technological Change

There are two components to the rate of tech-nological change, the rate at which new technol-ogies are created and the rate at which they areadopted by users. For appraising the impacts oftechnology on employment and related educationand training needs, the rate of adoption is key;it determines whether changes in requirements for

different types of labor affect primarily existingor future/prospective employees. Although newtechnologies may be created at varying rates, theconventional view among economists is that theuse of new technologies spreads relatively slow-ly. It is commonly assumed that firms adopt newtechnology in a rational fashion, meaning thatthey strive to use the most affordable processesto avoid the costs of prematurely scrapping facil-ities and to adapt technologies to their individualneeds. This view implies that, since firms typicallydo not adopt each technological advance as soonas it is developed and since firms experience somenormal level of employee turnover, employees arenot (repeatedly) subject to catastrophic displace-ment. A more elaborate presentation of this viewand a discussion of supporting research can befound in a paper by L. Jacobson and R. Levy, ap-pendix C.

Although the notion that there is a lag betweenthe introduction of a new technology and its wide-spread use is commonly recognized, there is dis-agreement as to whether recent innovations basedon microelectronics technology have or will con-tinue to spread as slowly as previous ones. Forexample, innovation and associated employmentchange in the printing industry have proceededquicker than the conventional view might lead oneto expect.

6

In West Germany, for example, employmentamong printers dropped by 21.3 percent between1970 and 1977, while productivity per hour roseby 43.5 percent . . . .5

There is also evidence that the use of computershas spread relatively rapidly, a phenomenon thathas prompted many scientists and engineers totake steps to refine their skills. A discussion oftechnological diffusion and its impacts on scien-tists and engineers is found in a paper by W.Cooke, appendix C. Additional material on engi-neering education is presented in chapter 3 of thistechnical memorandum.

A central question for an analysis of the socialand economic impacts of programmable automa-tion is whether programmable automation is like-ly to spread especially rapidly among firms andindustries, and why. Answering that question re-quires appraising the influence of the internationalnature of markets producing and using program-mable automation and the influence of cyclicaland structural change in the U.S. economy on therates of adoption and production of program-mable automation in the United States.

Nature of Technological Change

The way in which technological change affectsemployment and instructional requirements de-pends on the nature of the technology. The aspectsof the technology that are relevant to an examina-tion of labor impacts fall into three categories:

1. process v. product technology,2. embodied v. disembodied technology, and3. capital intensity of technology.

Process technologies are technologies of pro-duction, while product technologies pertain to theattributes of a finished product. Programmableautomation, which comprises a set of goods andservices used by businesses to make other prod-ucts, has elements of both, but is primarily re-garded as process technology.

The product-process distinction is importantbecause, historically, process changes have beenmore likely to affect employment adversely than

‘Cohn Norman, Microekwtronics at Work: I+oductivity and Jobsin the World Economy, Worldwatch Paper 39, October 1980.

have product changes. New products (such as pro-grammable automation equipment and systems)create new markets and new sources of employ-ment (although net employment growth dependson whether—and when—new products replaceolder ones). New processes, however, are oftenadopted because they are considered efficient,using l%wer resources than older processes to yielda product of given quality. * If the conserved re-source is labor, a company adopting a more effi-cient process will need fewer employees for a fixedoutput level. If the company faces a mature mar-ket for its end product (i.e., sales volume is notlikely to grow significantly), overall employmentwill fall, but, if the company faces a growing mar-ket, it might experience stable or growing employ-ment. Also, some new processes may be adoptedto improve product quality without necessarilydiminishing company employment. Discussionsof programmable automation in the trade andbusiness press typically note its potential for bothefficiency and quality enhancement. These discus-sions, which separate quality gains from costreductions, recognize that process improvementsmay facilitate output growth but do not assurethat companies can sell larger volumes of output.

Embodied technologies are associated withphysical entities such as pieces of equipment. Forexample, mechanical adding machines and elec-tronic calculators embody different technologiesto perform the same functions. Disembodied(sometimes called soft) technologies constituteways of organizing and managing production thatare not locked into tangible items. An exampleis the just-in-time system of inventory manage-ment, wherein suppliers deliver materials for vir-tually immediate use (rather than interim storage).The contrast between embodied and disembodiedtechnologies is important for appraisal of pro-grammable automation because the spread andthe ultimate utility of programmable automationis linked to associated changes in the organiza-tion of production and the structure of companiesand industries.

In evaluating embodied technologies used inmanufacturing and elsewhere, it is important to

● Sometimes this characteristic is referred to as productivity im-provement, since productivity refers to the amount of output derivedper unit of input to production.

7

recognize that disembodied technologies can oftencomplement or even substitute for them. For thisreason, comparing counts of different types ofequipment (e.g., robots) used by different coun-tries may be misleading. As comparisons of auto-mobile production in the United States and abroadreveal, it is possible to produce the same productusing equipment and systems that differ in sophis-tication under different principles of organizationand management. Also, because embodied anddisembodied technologies are combined in pro-duction, simple attributions of employment orworking environment variations to changes inequipment and systems are hazardous; they ig-nore the role of management, organization of pro-duction, and other “soft” factors.

Capital intensity refers to the amount of invest-ment in plant and equipment needed to producea given level of capacity, relative to the amountof other inputs, such as labor. A capital-usingchange in technology is defined to be one that re-quires more investment to produce a unit of prod-uct than the original technology; a capital-savingchange, one that requires less investment; and acapital-neutral change, one that requires the sameinvestment per unit of product.

Generalizations about how programmable au-tomation may affect capital intensity in differentmanufacturing applications are difficult to makeat this time because of limited experience with thetechnologies and uncertainty about the evolutionof the technologies and their markets. However,an understanding of the capital intensity aspectsof programmable automation is important forunderstanding the long-term employment andwage impacts of programmable automation. Inbrief, capital intensity affects the flexibility em-ployers have for accommodating different em-ployment and wage levels, given company levelsof sales volume and of output per worker. * Theramifications of varying levels of capital intensityare examined in a paper by E. Appelbaum, appen-dix C.

*Capital intensity may also affect the distribution of wealth gener-ated through production—a shift to capital-using technologies may,for example, be associated with growth in profits (return on capital)relative to wages. Changes in the distribution of wealth in turn mayaffect employment and wage levels because those realizing incomeas profits may spend and invest in different markets than those realiz-ing income as wages.

Pattern of Technology Diffusion

The impacts of programmable automation tech-nologies will depend on where they are used aswell as when. New technologies may spread with-in industries among all firms, among firms in onlycertain industry segments, or among large or lead-ing firms only. They may be used in isolated in-dustries, interdependent industries, and/or in in-dustries in different sectors of the economy. Theimpact of programmable automation technologieson employment (and therefore on instructionalrequirements) in the United States will depend,in particular, on global trends in the productionand use of programmable automation, since themarkets for automation and for many productsmade with it are international.

Preliminary observations presented by industryand labor representatives and technology analystsat the 1981 OTA Exploratory Workshop on theSocial Impacts of Robotics and elsewhere indicatethat programmable automation may eventuallybe diffused more broadly than conventional auto-mation. b While conventional automation has beenapplied primarily in large-volume or mass-pro-duction manufacturing industries, programmableautomation offers potential value to use in small-er volume, batch manufacturing applications,which are the majority of manufacturing applica-tions. 7 Whereas conventional automation is de-voted to production of single products, program-mable automation equipment and systems can beadapted, through reprograming, for productionof multiple products, each of which may be de-sired in limited quantities. Since production equip-ment and systems themselves are often manufac-tured in small-quantity batches, their manufacturemay be automated.

Finally, equipment and systems similar (and insome cases identical) to those used for program-mable automation in manufacturing are beingadopted in nonmanufacturing settings, with multi-ple impacts on employment opportunities. A pos-sible consequence of the spread of office automa-tion, for example, is a decline in the growth rateof clerical employment. On the other hand, large

bExploratory Workshop on the Social Impacts of Robotics-ABackground Paper, op. cit,

71bid.

8

investments in office automation equipment and automation equipment is imported. The implica-systems imply potential employment gains in tions of the pattern of technology diffusion formanufacturing industries supplying office automa- employment in different sectors of the economytion, although who benefits from such employ- are discussed in a paper by E. Appelbaum, appen-ment gains depends on the extent to which office dix C.

Chapter 2

Labor Markets and

Chapter 2

Labor Markets and Working Environment

INTRODUCTIONThe use of computers in manufacturing has

aroused concern since the late 1950’s and early1960’s, when awareness of the potential of com-puter technology began to emerge and when ap-plications of more conventional automated manu-facturing were accelerating. During that periodpublic interest in the social ramifications ofautomation and computers was greater in Europethan in the United States. However, official U.S.concern led to the formation in 1965 of a specialFederal study commission, the National Commis-sion on Technology, Automation, and EconomicProgress, charged with the tasks of: 1) assessingthe effects, role, and pace of technological change;2) describing changes in employment demandsand working conditions associated with techno-logical change; 3) defining “unmet community andhuman needs” that technology can help to meet;and 4) identifying policy options for implement-ing new technologies. After meeting for a year,the Commission issued a report that foreshadowscontemporary discussions of job displacement,changing working conditions, and instructionalneeds.

From the 1950’s through today, labor-relatedconcerns associated with automation and com-puters have tended to fall into three not-wholly-distinct categories: 1) labor markets or employ-ment, 2) working environment (job content andoccupational safety and health), and 3) industrial

or labor-management relations. Of these threecategories, labor market issues have been mostsalient in popular (and political) discussions ofautomation, because employment is widely seenas reflecting the economic vitality of a countryor region. By contrast, working environment is-sues may be more subtle and more likely to beappreciated by those groups of people in directcontact with specific working environments. Fi-nally, industrial relations both influence and areinfluenced by changes in labor markets and work-ing environments that are associated with newtechnology and other factors.

In order to analyze the labor market implica-tions of programmable automation, it is necessaryto be able to measure and forecast the degree andtypes of changes in employment that may accom-pany the spread of this technology. The varietyof claims as to the eventual employment impactsof programmable automation that are being publi-cized by the media suggests that such evaluationsare straightforward. However, there appears tobe no accepted methodology for making such em-ployment forecasts reliably, a problem that wasemphasized in debates among participants of theOTA Labor Markets and Industrial RelationsWorkshop. This technical memorandum pointsout some of the shortcomings of many publicizedforecasts and some of the requirements for satis-factory forecasts.

POTENTIAL FOR OCCUPATIONAL CHANGE: AN OVERVIEWA first step in measuring or forecasting how

programmable automation or other new technol-ogies may affect employment-by occupation andindustry-is to assess: 1) how programmableautomation affects the activities performed bypeople working in user industries and occupa-tions, and 2) what types of activities maybe per-formed by people engaged in producing auto-mated equipment and systems. Unfortunately,

there are few empirical data describing relevantactivities. Moreover, what data may exist (e.g.,in case studies) may have little general valuebecause early programmable automation applica-tions have been limited in number compared toapplications of other types of equipment and sys-tems, and they have been tailored to individualcompany needs. Early applications also are like-ly to be different from later applications involv-

11

12

ing more sophisticated equipment and systems,especially since future applications are expectedto feature greater computer integration of produc-tion and other company activities.

At this time it appears that the range of activ-ities undertaken by manufacturing firms and vul-nerable to change in connection with programma-ble automation is not limited to the fabricationand assembly of products. Employment that maybe directly affected by the production and use ofprogrammable automation is associated with awide range of activities, including research anddevelopment; the design, fabrication, assembly,distribution, and servicing of products; and man-agement.

Production Activities. —The types of new ac-tivities associated with production of program-mable automation, as compared with productionof conventional factory equipment, are those thatpertain to its computerization aspects, namely thedevelopment, distribution, and/or adaptation ofcomputer hardware and software. Computeriza-tion, or more broadly a shift to microelectronicsfrom mechanical or electromechanical compo-nents, may also alter other activities associatedwith production of programmable automation.For example, the use of microelectronic com-ponents affects fabrication and assembly tech-niques, in part because individual microelectronic

components can often do the work of multiplemechanical ones.1 Finally, like the production ofconventional equipment, production of program-mable automation also entails applications engi-neering, technical support, installation, sales, andclerical activities.

Use Activities.—Activities associated with theuse of programmable automation are broadlysimilar to those associated with the productionof programmable automation, since both produc-tion and use of programmable automation aremanufacturing endeavors. Nevertheless, variationamong user industries (including users who alsoproduce programmable automation) by size andby nature of product will determine the specifictypes of tasks and occupations affected amongusers. The types of tasks that maybe created withthe use of programmable automation also pertainto computerization (e.g., programing, mainte-nance of electronic equipment, and data basemanagement). The types of tasks that may beeliminated are those tasks sensitive to the internal-ization of information flows (e.g., for certain cler-ical operating and supervisory tasks), or to thereplacement of physical labor (e.g., for welding,assembling, materials handling, and drafting).

IRoy Rothwell and Walter Zegveld, Technical Change and Em-ployment (New York: St. Martin’s Press, 1979).

OCCUPATIONAL CHANGE FORECASTINGHistorically, attempts to forecast detailed

changes in occupational employment have metwith limited success. As the Bureau of Labor Sta-tistics (BLS) has noted in evaluating its ownforecasts, it is easier to predict directions of changefor broad categories of employees than magni-tudes of change for relatively specific groups. Thissituation is unfortunate, since the more detailedthe occupational differentiation, the more precisemay be the evaluation of employment variationamong occupations and industries and thereforethe identification of people who may benefit orbe harmed by technological change.2

‘Max L. Carey and Kevin Kasunic, “Evaluating the 1980 Projec-tions of Occupational Employment, ” Monthly Labor Review, July1982.

Note that in practice, very detailed occupational analyses maybe less accurate than more aggregated analyses because of nonsam-

The occupations of people who maybe direct-ly affected by the spread of programmableautomation include professional specialty; exec-utive, administrative, and managerial; techniciansand related support; machine operators, assem-blers, and inspectors; precision production, craft,and repair; and handlers, equipment cleaners,helpers, and laborers. Table 1 contains the fullcurrent and prior lists of major census occupa-tional groups. While this set of categories can beused to describe the occupational mix of any in-dustry and the labor force as a whole, it is toobroad to describe more than gross shifts in oc-

pling errors in occupational title classification and analysis. SeeHarvey Goldstein, “Occupational Employment Projections for LaborMarket Areas: An Analysis of Alternative Approaches” (WashingtonD. C.: U.S. Department of Labor, 1981), R&D Monograph 80.

13

Table 1.—A Comparison of 1980 and 1970 Decennial Census Occupational Categories

1980 1970

Broadest groupingsManagerial and professional specialty White-collarTechnical, sales, and administrative support Blue-collarService ServicePrecision production, craft, and repair FarmOperators, fabricators, and laborersFarming, forestry, and fishing

Major occupational groupsExecutive, administrative, and managerial Professional and technicalProfessional specialty Managers and administrators, except farmTechnicians and related support SalesSales ClericalAdministrative support, including clerical Craft and kindredPrivate household Operatives, except transportProtective service Nonfarm laborersService, except private household and protective service Private householdPrecision production, craft, and repair Other service workersMachine operators, assemblers, and inspectors Farmers and farm managersTransportation and material moving Farm laborers and supervisorsHandlers, equipment cleaners, helpers, and laborersFarming, forestry, and fishing

SOURCE: John E. Bregger, “Labor Force Data From CPS to Undergo Revision in January 1983,” A40rrth/y Labor 17evlew, November 1982.

cupational proportions. Within each category,hundreds of occupations can be differentiated.Aggregating occupational categories may resultin uncertainty about future change in such detailedoccupations as “robot technician, ” where thespecific designation falls within a broader cate-gory, such as science and engineering technicians.Another cost of aggregation is generality—theaverage pattern of change within an industry maynot correspond to actual changes experienced byindividual companies or people, in part becauseindividual companies vary in their use of employ-ees with very specific skills, as well as in their useof production technologies. However, even a de-tailed occupational breakdown may mask changesin job content that may arise with new technol-ogy.

Most analyses of employment change use ag-gregated occupational descriptions because collec-tion and manipulation of more detailed occupa-tional data are costly, and because the most de-tailed descriptions fall easily out of date. Manyexperts believe that analysts have been handi-capped by the kinds of data available. For exam-ple, the most recent edition of the Dictionary ofOccupational Titles (DOT), which describes200,000 occupations, was published 6 years agoin 1977 (the previous edition was published in

1965). The DOT does not contain an entry for“robot technician, ” and the most similar entry,“automated equipment engineer technician,” refersto an individual who works with machinery pro-ducing items from paper or cardboard stock (asopposed to metal, plastic, or other materials withwhich robots or other forms of programmableautomation might be used).

How can the effects of programmable automa-tion on employment levels and distribution amongoccupations be gaged? Already, there are manyestimates of the overall and occupational employ-ment impacts of programmable automation ap-pearing in the trade, popular, and businesspresses. Examples include the following:

Automotive industry sources say the generalformula is that 1.7 jobs are lost for every robotintroduced.3

“Automation will cause a 20 to 25 percent de-cline in the factory work force over the next dec-ade, ” says Thomas G. Gunn, managing directorof Arthur D. Little’s computer integrated manu-facturing group. An internal study done by GEshows that it is now technologically possible forthe company to replace half of its 37,000 assem-

3Joyce Price, “With Robots On the Way, GM Workers Worry,”The (Baltimore) News American, Sept. 27, 1982.

14

bly workers with machines. Company officialsare quick to point out that they have no plansto do that and where GE is automating existingplants—at Erie, for instance—it is retraining thedisplaced workers. Sometimes extensive automa-tion also creates new jobs even as it destroysothers. The new automated parts factory inFlorence, Ky., for example, will allow Yamazakito expand production at its manned machine-toolassembly plant nearby; 100 workers will be hiredto fill the new jobs.4

Experts estimate that on the order of 45 millionexisting jobs—45 percent of all jobs, since thereare about 100 million people at work—could beaffected by factory and office automation. Muchof the impact will occur before the year 2000. . .The United Auto Workers, one of few unions thattries to anticipate automation expects its auto in-dustry membership to drop to 800,000 from 1million between 1978 and 1990, even assuminga 1.8 percent annual increase in domestic autosales . . . . Harvey L. Poppel, a senior vice-president with Booz, Allen& Hamilton, Inc., esti-mates that 38 million of more than 50 million ex-isting white-collar jobs eventually may be af-fected by automation. Paul A. Strassmann, vice-president of strategic planning for Xerox Corp.’sInformation Products Group, predicts that 20million to 30 million of these jobs will be affectedby 1990.5

Forecasting is, at its best, imprecise. However,the impact of robotics will definitely mean theelimination of some blue-collar jobs and the crea-tion of jobs that didn’t exist as recently as 10 yearsago. It’s estimated that there are currently 10,000workers involved in robotics in some form or an-other throughout the world. That includes every-one from assembly line workers to designers, en-gineers, company presidents, clerical help, main-tenance people and all of the support necessaryfor a young, developing industry.’

The above sources have derived their estimatesthrough various means. The estimation proce-dures used appear to fall into two categories:“engineering” and “economic.” Both categoriesderive labor requirements from other phenomena:

‘The Factory of the Future,” Business Week, Sept. 6, 1982.“’Changing 45 Million Jobs,” Business Week, Aug. 3, 1981.bJoel Weber, “Can Robots Do a Better Job?” D&B Reports, Janu-

ary/February 1980.

equipment in the former category, and demandfor finished products in the latter. These proce-dures are reviewed below to illustrate how limitedcurrent understanding and modeling of program-mable automation employment impacts really are.

Engineering Estimates

Engineering estimate is the term used in thisreport to refer to an estimate based more or lessexclusively on technical aspects of technologicalchange. Although engineering analyses may beused to support economic analyses of employmentchange, they are frequently used on their own.Most of the employment (or, in particular, unem-ployment) estimates cited in popular discussionsof programmable automation seem to be of thistype.

Engineering estimates are made by describingthe capabilities (for physical and mental work) ofnew automation technologies, projecting capabili-ty improvement over time, comparing the capabil-ities to tasks performed by humans, relating hu-man tasks to different occupations, and derivingthe number of jobs, by occupation, that could beassumed by new and future improved types ofequipment. This is done by comparing guesses asto the percentages of work that could be trans-ferred to programmable automation with countsof the numbers of people currently doing thatwork. For example, the employment impact of awelding robot might be estimated by identifyingthe types of welds the robot can perform, measur-ing the number of welds the robot can performin a given period of time, and calculating thenumber of “jobs” that might be displaced by com-paring the number of robots needed to achievea given volume of welds with the number of hu-man welders who could achieve the same volumeof welds, given contemporary hiring patterns.Projected improvements in robot welding, orother changes in the basic assumptions can be ac-commodated by modifying the calculations.

Similar calculations are used to derive the em-ployment requirements for producing the supplyof robots necessary to achieve a given level ofdisplacement —estimate the type of tasks requiredto produce robots, the number of tasks of eachtype required per robot, the allocation of robot-

15

production tasks between humans and equipment,and combine with the number of robots desiredin a given period to forecast producer employ-ment requirements.

Shortcomings of Engineering Estimates

The engineering approach is easily understood,adaptable to different assumptions, and useful asa first step in estimating the potential employmentimpacts of programmable automation. However,it has many limitations—in its application, if notits concept —which are largely functions of thenarrowness of the technological and/or economicassumptions chosen. Shortcomings of engineeringestimates may include some or all of the follow-ing:

●

●

●

These estimates are easily confounded by er-rors in projecting future technologicalcapabilities. Although providing a range ofassumptions may improve the usefulness ofthe estimates, there remains a problem of in-ability to foresee all possible developments,especially in new technologies.Both the development and the analysis ofautomation technologies (conventional andalso programmable) often rely heavily onpoint-by-point comparisons of electronic andmechanical capabilities with human capabil-ities, an orientation that lends itself tocalculations of how and where automationequipment and systems may replace or sub-stitute for human activities. See table 2.However, this orientation fails to capture thepotential for programmable automation ei-ther to perform jobs in ways other than sim-ulation of human behavior, or to performjobs that are poorly done or not done at allby humans because of human limitations.This failure may lead to overestimation orunderestimation of job displacement.Engineering estimates may be misleading be-cause they tend to yield a “technically” idealmix of humans and equipment, while the ac-tual mix may reflect complex managementand implementation considerations that areindependent of the capabilities of specificequipment or systems. For example, mana-gers mav be motivated out of risk aversion.

●

to provide redundant capabilities in the formof “extra” workers (or overskilled workers)to provide manual performance backup ormonitoring services, at least in the short termwhen programmable automation is relativelyunfamiliar. Varying assumptions about themix of humans and equipment would easethis problem.Engineering estimates are frequently based oncurrent or recent labor force characteristics.This practice assumes that users will buy anduse programmable automation to serve rela-tively constant production needs, and thatworkers will seek different jobs at constantrates. However, the job displacement andcreation consequences of programmable au-tomation will depend not only on how pro-grammable automation affects the numberand type of tasks per worker, but also onhow sales volume and the mix of products—which determine the total number of tasksdone at all—change. These quantities mayvary in response to factors other than tech-nological change, such as shifts in consumertastes. In addition, the employment conse-quences of programmable automation willdepend on the numbers and types of peoplewilling and able to work at different typesof jobs, which also may vary independentlyof technology.

Engineering analyses are useful for identifyingthe types of people (excluding, perhaps, managers)who may be affected by programmable automa-tion. As currently used, they are often too simplis-tic to provide realistic estimates of industry oreconomywide employment change. The chiefproblem with available engineering estimates ofnational employment impacts seems to be a lackof consideration for variations in economic con-ditions, trade patterns, and labor supply, althoughthese factors probably could be accommodatedby engineering analyses. Nevertheless, the engi-neering approach provides a framework that canbe used to evaluate the employment consequencesof alternative strategies for implementing pro-grammable automation, and a mechanism forevaluating specific variations in production proc-esses.

16

Table 2.—Comparison of Robot v. Human Skills and Characteristics

Robot a Human

A Act/on and manipulation1. Manipulation abilitiesa. One or more arms. Automatic hand change is

possible.b. Incremental usefulness per each additional arm can be

designed to be relatively higher than in humans.c. Requires the same amount of feedback throughout

operation.

d. Movement time related to distance moved by speed,acceleration and deceleration, and will increase with higheraccuracy requirements.

●

B. Brain and control1. Computational capabilitya.

b.c.

d.e.

f.g.

h.

i.Jk.1.

2.a.

b.c.d.

3.a.b.

E

3.a.

b.c.d.

Fast, e.g., up to 10 K bitrdsec for a small minicomputer control.

Not affected by meaning and connotation of signals.No valuation of quality of information unless provided byprogram.Error detection depends on program.Very good computational and algorithmic capability bycomputer.Negligible time lag.Ability to accept information is very high, limited only bythe channel rate.Good ability to select and execute responses.

No compatibility limitation.If programmable—not difficult to reprogram.Random program selection can be provided.Command repertoire limited by computer compileror control scheme.

MemoryMemory capacity from 20 commands to 2,000 commands,and can be extended by secondary memory such as cassettes.Memory partitioning can be used to improve efficiency.Can forget completely but only on command.“Skills” must be specified in programs.

IntelligenceNo judgment ability of unanticipated events.Decisionmaking limited by computer program.

Miscellaneous factors

TrainingRequires training by teaching and programing by anexperienced human.Training doesn’t have to be individualized.No need to retrain once the program taught is correct.Immediate transfer of skills (“zeroing”) can be provided.

4. Social and psychological needsNone

5. Individual differencesOnly if designed to be different.

a.

b.

c.

d.

● ●

a.

b.c.

d.e.

f.g.

h.

i.jk.1.

a.

b.c.d.e.f.

a.b.

● *

● *

a.

b.c.d.e.f.

a.

b.

Two arms—two legs—multipurpose hands.

Two hands cannot operate independently.

Feedback requirements (type and quantity) change withpractice—initially relatively higher than robot; visual feedbackdominates other sources of feedback.Movement time and accuracy governed by Fitts law. Highprecision movements may interfere with calculation processes.

Slow—5 bits/see.

Affected by meaning and connotation of signals.Evaluates reliability of information.

Good error detection/correction at cost of redundancy.Heuristic rather than algorithmic.

Time lags increased, 1 to 3 sec.Limited ability to accept information (10 to 20 bits/sec).

Very limited response selection/execution (l/see); responsesmay be “grouped” with practice.Subject to various compatibility effects (RR, SR, SS).Difficult to reprogram.Various sequence/transfer effects.Command repertoire limited by experience and training.

No indication of capacity imitation.

Not applicable.Directed forgetting very limited.Memory contains basic skills accumulated by experience.Slow storage access/retrieval.Very limited working register = 5 items.

Can use judgment to deal with unpredicted problems.Can anticipate problems.

Requires human teacher

Usually individualized is best.Retraining often needed due to forgetting.Zeroing usually not possible.Very costly.Not everyone can be taught.

Emotional sensitivity to task structure—simplified/enriched;whole/part.Social value effects.

100 to 150 percent variation may be expected.

a~obot ~armeter “alue~ are cited from currently available Industrial robot literature.

SOURCE: Nof, Knight, and Salvendy, “Effective Utilization of Industrial Robots—A Job and Skills Analysis Approach,” AHE TransactIons, vol. 12, No. 3, September 19S0.

77

Economic Estimates

Economic estimates is the term that will be usedin this technical memorandum to refer to projec-tions based on macroeconomic models. Economicestimates are better than engineering estimates forprojecting aggregate changes in employment pat-terns because they are inherently more com-prehensive. On the other hand, economic esti-mates may not be practical or useful for gagingpossible employment change at the company levelbecause they tend to be highly aggregated.

Economic estimates are made by explicitly eval-uating several factors, in addition to technology,that impinge on employment demands. For exam-ple, prices and production levels of goods andservices are typically considered, taking into ac-count, in turn, the forces that affect these factors,such as international trade and projected shifts inthe relative strengths of different sectors of theeconomy. Economic estimates place substantialemphasis on descriptions of employers in termsof different sectors of the economy and differentindustries within sectors. They rely on engineer-ing analyses for descriptions of alternative effectsof technologies on industry requirements for suchproduction inputs as labor (by occupation), equip-ment, and materials.

Economic estimates of employment change aremade using mathematical models of productionfunctions, which describe how different inputs toproduction are combined to yield a given levelof output. Some models pertain to single indus-tries, while other, more elaborate models also takeinto account the interactions among industries.The most detailed economic estimates come fromlarge-scale models, in particular those based onso-called input-output (I-O) models, which en-compass entire (regional, national, or global)economies. Technologies are defined in I-O mod-els as the structure—number, type, and propor-tions—of inputs associated with the productionof a unit of output of a given product.

The employment forecasts (total and by occu-pation) of the BLS draw on large-scale economicmodeling. They are generated with an I-O modelof the U.S. economy in combination with othermodels that forecast change in the labor force and

in the level and pattern of economic activity. Alsoincluded are descriptions of staffing patterns (themix by proportion of different types of workers)for each industry included, obtained from periodicsurveys. Since the BLS estimates are widely used,and since the procedures are substantially similarto procedures used by others who forecast withlarge-scale economic models (indeed, other modelsoften use the same data), a description of the out-lines of BLS forecasting procedures can serve asa description of economic employment forecastingprocedures in general (although individual modelsand procedures do differ in their details). *

Figure 1 shows the different computational ele-ments that contribute to BLS forecasts. The firstset of procedures is the projection of labor forcecharacteristics. The second set of procedures is theprojection of overall economic activity and result-ing gross national product. These projections re-quire estimation of the types and volumes ofgoods and services the economy can produce orsupply in both private and public sectors, andthose that will be demanded by the public andprivate sectors. The third set of procedures trans-lates overall economic projections into projectionsof industry activity, allocating estimated consum-er spending among product groups and allocatingproducts to producing industries. Estimated grossprivate domestic investment is in turn allocatedbetween changes in business inventories and in-vestments in construction (residential and nonresi-dential) and producer-durable goods (e.g., ma-chinery and tools). The fourth set of procedurestranslates projections of industry output into pro-jections of industry employment. This is done bya combination of procedures for estimating laborproductivity (defined as output per unit of laborinput) and weekly hours of work for each indus-try.

The final set of procedures yields projectionsof employment by occupation and by industry.It combines descriptions of staffing patterns ob-tained by periodic surveys with estimates of the

*Note that BLS has recently contracted with Chase EconometricsAssociates, Inc., to use the Chase macroeconomic model to developprojections of aggregate economic activity, using assumptions andvariables chosen by BLS. This arrangement will supplement in-houseBLS modeling and analysis.

98-951 0 - 83 - 4

18

Figure 1 .—Bureau of Labor Statistics EmploymentProjections System

“- - - : -

Census Age/sex/racep o p u l a t i o n “ f o r c e ‘

projections rates, .:

,I

SOURCE: U.S. Department of Labor, Bureau of Labor Statistics, “BLS Eco-nomic Growth Model System Used for Projections to 1990.”

number of jobs per industry. All of these proce-dures are described in detail in the BLS publica-tion, BLSEconomic Growth Model System Usedfor Projections to 1990, April 1982.

Shortcomings of Economic Estimates

As the description of the BLS procedures shows,large-scale economic models can take into accountthe growth and decline of different industries, thelikelihood that individual industries adopting newtechnologies may maintain or increase output lev-els, and the responsiveness of industry employ-ment levels to industry technology change. Thisframework prevents overattributing employmentchanges to single influences such as technologychange, as it shows the consequences of combina-tions of influences. In their detail, however, thevalidity of the projections generated depends onthe assumptions that underlie the formulation andoperation of each aspect of the model and the inte-gration of the different aspects. Moreover, the useof large-scale economic models carries the risk ofoversimplifying complex processes and conveyingan impression of greater analytical thoroughnessthan may actually exist.

Several questions have been raised about theassumptions used in large-scale economic forecast-ing models. The following list of some of theshortcomings of economic estimates reflects con-cerns raised by participants at the OTA LaborMarkets and Industrial Relations Workshop, whodebated whether economic models could ade-quately evaluate the impacts of programmableautomation on employment. It also reflects con-cerns raised by others regarding economic mod-eling in general and modeling of technologicalchange impacts in particular.

● Labor Supply. The growth of the labor forceand change in labor force participation ratesof specific groups depend in complex wayson demographic and economic factors. Theserelationships may not be captured in eco-nomic models which project labor supply andindustrial output profiles separately. * Also,variations in the quality, rather than thequantity, of available labor maybe beyond

*BLS is currently working to improve its treatment of demographicand economic influences on the labor force.

19

●

●

the scope of contemporary large-scale eco-nomic models. Consequently, the output oflarge-scale economic models may best beviewed as projected demands rather thanemployment levels, per se.

Technological Change. It is unclear how welllarge-scale models account for changes inequipment technologies. Although the com-mon practice of projecting future capitalstock by extrapolating from past use of plantand equipment and past descriptions of in-dustries and products suggests that economicmodels may be unable to capture the impactsof nontraditional equipment, experts disagreeas to whether measures of specific new tech-nology attributes are necessary for derivingeconomic estimates of employment change.See papers by L. Jacobson/R. Levy and F.Duchin, appendix C. In addition, economicmodels typically are constructed using theassumption that technological change isadopted to reduce unit costs, although it mayalso be adopted for other reasons (e.g., tomeet health or pollution standards) leadingto cost increases.

Staffing Patterns. Employment change dueto reorganization of production associatedwith programmable automation may not becaptured where occupational employment isprojected using staffing patterns derived fromprior practices. Similarly, changes in occupa-tional content may not be accounted for.BLS, for example, has found that many ofthe largest errors in its past estimates of occu-pational employment “resulted primarilyfrom misestimates of industry-occupational

staffing patterns.”7 The development of ade-quate staffing patterns would appear to re-quire engineering analyses that take into ac-count possible variations in the implementa-tion of programmable automation, altern-ative levels of integration of manufacturingactivities, and alternative approaches to ac-commodating existing company work forces.

Like engineering estimates, economic estimateshave several shortcomings. However, while engi-neering estimates tend to highlight job displace-ment impacts of new technology, economic esti-mates are better suited for evaluating whether per-sons displaced from particular industries may findjob opportunities in other industries requiringtheir skills, and therefore whether job displace-ment is likely to be associated with unemploy-ment. How well they do this depends on how wellthey capture the different components of the econ-omy and their interactions. Similarly, while engi-neering estimates may establish new needs for in-dividuals with certain skills, economic estimatesmay more readily provide perspective on econ-omywide demand for such individuals and there-fore whether demand for certain skills or occupa-tions is likely to exceed or fall short of supply.These differences arise because economic analysesas a rule model the interactions among segmentsof the economy, while engineering analyses donot, even though they may apply to the nation-wide use of a technology. However, valid infer-ences regarding future unemployment and laborshortages require that engineering analysis, eco-nomic and industry analysis, and labor supplyanalysis be considered together.

‘Carey and Kasunic, op. cit.

BEYOND HARDWARE AND SOFTWARE:OTHER FACTORS TO BE CONSIDERED

In general, satisfactory projections of the mag- among industries, and in the overall mix of em-nitude and distribution of employment shifts asso- ployment opportunities in the economy. Theseciated with programmable automation should changes will reflect the basic parameters describedtake into account a variety of factors that contrib- in the introduction (rate, nature, and diffusionute to the direct and indirect effects of the new pattern of technological change) and also the in-technology. Among these are changes in the orga- fluence of institutional factors such as labor-man-nization of production, in the level of output agement agreements and norms, which affect the

20

rate and manner of application of new technol-ogies. Labor-management relations are examinedin appendix B.

Organization of Production

Change in the mix and volume of activitiesamong users of programmable automation willdepend on alteration of the organization of pro-duction (and concomitant changes in productlines) that may occur as a result of its use. Asdiscussed in chapter 1, it is anticipated that thespread of programmable automation will involveboth technologies embodied in automated equip-ment and systems and disembodied technologiesin the form of organization and managementchanges. These changes may be most pronouncedin small-volume or batch production settings:

For a long time the functional layout in batchproduction, that is, all machines of the same kindare gathered in groups, has been as natural as thetransfer line in mass production. Through thefunctional layout, machine utilization can be kepthigh, but at the expense of complex routing ofparts through the shop and large buffers and in-ventories. . . . In the new manufacturing meth-ods the main principle is to organize the factoryaccording to product-oriented layouts. All ma-chines needed to produce one product or one setof products are grouped together in a “subfac-tory,” sometimes with its own administration.Each worker in product-oriented layouts attendsseveral machines. In the functional layout we canwith some simplification say that the materialswait for the machines while the machines in theproduct-oriented layout wait for the materials.The lead time (defined as the total time neededfor material to be processed into a finished prod-uct) can thereby be reduced dramatically.8

Production may also be reorganized betweenfacilities, as programmable automation facilitatesregional and even international reorganization ofproduction activities. For example, Ford MotorCo.’s Erika project (which resulted in the Escortand Lynx cars in the United States and similar carsin other markets) used “the largest collection ofcomputer design hardware under one roof” to

‘The Promotion of Robotics and CAD/CAM in Sweden,” reportfrom the Computers and Electronics Commission, Ministry of In-dustry (Sweden), 1981.

pool U.S. and European product design and anal-ysis efforts, eliminating separate parallel effortson different continents.9 Although there has beenmuch speculation among technology and industryanalysts about potential employment effects ofproduction reorganization, little reorganizationappears to have taken place, in part because busi-ness management has either failed to understandor resisted such change, and in part because theintegration aspects of programmable automationappear insufficiently developed.l”

Output Level

The employment consequences of programma-ble automation production and use depend notonly on the mix of manufacturing activities, butalso on production volume for both automationand end products made with it. Since program-mable automation will be sought by both newusers and customers previously using other typesof equipment, production volume should be eval-uated by taking into account possible reductionsin volume of other, older technology equipmentand systems. This offset problem is generallyrecognized in evaluating the impacts of microelec-tronics-based (and other) technologies found inboth new products and new production processes.

(I)t is clear that microelectronic technologieswill create jobs in those industries manufactur-ing novel electronic products. The $4 billion nowbeing lavished on electronic watches, calculators,games, and other microelectronic products hasspawned a whole industry that did not even ex-ist a decade ago. According to a projection by. . . Arthur D. Little, the manufacture of theseitems, together with computers and other elec-tronic equipment, could create about 1 millionnew jobs between 1977 and 1987 in the UnitedStates and Western Europe combined. About 1.5million people are now employed in the electron-ics industry in the United States. But these jobswill not represent net additions to the work force,for they will be offset to some extent by job lossesin the manufacture of goods with which the newmicroelectronics-based products are competing.11

‘Automotive News, Feb. 15, 1980.IO%e for example: Bela Gold, “CAM Sets New Rules for Produc-

tion,” Harvard Business Review, November-December 1982.llcolin Noman, ‘~icrWIWtronim at Work: Productivity and

Jobs in the World Economy,” Worldwatch Paper 39, October 1980.

21

The net effects of programmable automationon user employment will depend on the effect ithas on end-product prices and on foreign trade,product specialization, and other conditions inuser markets in the United States and abroad.These factors, together with technology and gen-eral economic conditions, determine growth indomestic company sales volume.

Employment Opportunity Mix

Overall, employment effects of programmableautomation will also depend on changes in em-ployment opportunities throughout the economy.EconomyWide changes in employment activitiesdepend in part on the pattern of diffusion of pro-grammable automation and in part on the pat-tern of change in the mix of products available.

LABOR SUPPLYWhile employment demands may change be-

cause of the characteristics of programmable auto-mation technologies and of industries producingand using them, change in employment (and un-employment) patterns also depends on the charac-teristics of the supply of labor: who is availableto do the work offered by employers, how ablepeople are to do different types of work, andwhether there are too many or too few peoplewith different abilities to do the work offered. Thefollowing is a brief overview of labor supply at-tributes and concerns.

Demography

The number of people willing and able to work,usually counted between the ages of 16 and 65,depends on several factors, including natural pop-ulation growth, * immigration and emigration pat-terns, public health conditions, the age structureof the population (the proportions by age), andthe willingness of people to work, given the levelsof available wages and salaries and alternativesources of income. The overall size, growth rate,

● Natural population growth reflects mortality and in particularfertility (childbearing) rates, both of which may vary geographically,and among subgroups.

Thus, although the apparent long-term decline inthe manufacturing share of total U.S. employment(which began prior to widespread use of program-mable automation) reflects the adoption of labor-saving technologies, the slow long-term growthin the absolute level of U.S. manufacturing em-ployment illustrates the importance of sales vol-ume and market growth (including the introduc-tion of new products). It can be misleading toevaluate the employment impacts of new proc-esses from the perspective of a constant mix offinished products because the number as well asthe mix of goods and services provided to bothproducers and consumers is dynamic. Such evalu-ations are common, however, because the existingmix of products is known, while future productarrays are not.

and age structure of the population are importantmeasures of the availability of people in grossnumbers to do work using particular technologiesto support a given level of economic activity. At-titudes toward work and other social factors,which vary among geographic areas and ethnicgroups, contribute to the actual numbers andtypes of people participating in the labor force.

Age structure and fertility patterns are particu-larly important influences on the makeup of thelabor force. Fertility patterns, in combination witheconomic conditions and social norms, influencethe labor force participation of women as well asthe age structure of the population. The earlierand more frequently women give birth, for ex-ample, the younger the population is likely to beand the greater the (eventual) influx of new en-trants to the labor force. Delays in and decreasesin the incidence of marriage and childbearing overthe past two decades have been causing the U.S.population to age by reducing the proportion ofchildren. The age structure, in turn, influences:1) the proportion of the population which is tooyoung and/or too old to work and therefore de-pendent on the economic activity of the working-age population, 2) the overall rate of populationgrowth, and 3) the numbers of new entrants to

22

the labor force. Consequently, differences in agestructure among countries influence national dif-ferences in employment patterns, preferences, andpolicies. The Japanese, for example, are reportedto have shown early interest in programmableautomation in part because “aging” of their popu-lation limited the supply of young workers.12

The composition of the American populationhas shifted toward older age groups more slowlythan that of the Japanese population, but the sup-ply of new entrants to the labor force is expectedto begin a long-term decline in the 1980’s. Federalprojections of the U.S. population through theyear 2050 show the number of teenagers to peakin 1980. The U.S. elderly population is expectedto grow from the 1980 level of 25.7 million to 67.1million by 2050, increasing from 11 to 21.7 per-cent of the population .13 Unless the propensity ofthe elderly to work increases dramatically, thispopulation shift will reduce the overall exposureof the U.S. labor force to job displacement, andit may eventually increase demand for labor-saving technologies.

Qualitative Attributes

Other characteristics of the labor force impor-tant to understanding employment trends arequalitative. They include level, type, and qualityof education or training; skills; and preferencesregarding different types of work. Education andtraining are important determinants of skills andtherefore of the types of work individuals can do.However, educational attainment is an imprecisemeasure of the qualities of workers, since skillscan be obtained through means other than for-mal instruction. A discussion of education, train-ing, and retraining can be found in chapter 3.

Occupational Structure

The characteristics of the labor force, togetherwith the array of jobs available, contribute to theoccupational structure of an economy-the distri-bution of workers among occupations. Labor

IZG. K. Hutchinson, “Flexible Manufacturing Systems in Japan”(Milwaukee, Wis.: University of Wisconsin Management ResearchCenter, November 1977).

‘3Robert Pear, “Population Drop Predicted in U.S.,” New YorkTimes, Nov. 9, 1982.

force attributes, and occupational structure in par-ticular, change over time with changes in demog-raphy and with changes in social norms, both ofwhich reflect economic conditions. For example,the absolute and relative growth in service sec-tor employment has been associated with thegrowth in female labor force participation.

Key attributes of the 1980’s labor force in theUnited States include growing proportions of old-er workers, relatively large proportions of womenand minorities, relatively large proportions ofcollege-educated workers, and declining numbersof people willing to work in low-level occupa-tions.14 Tables 3 and 4 display basic characteris-tics of the U.S. labor force.

It is important to note that, as long as differentgroups don’t radically change their propensitiesto seek employment, it is relatively easy to de-scribe the physical characteristics of the laborforce 10 to 12 years into the future since these peo-ple have already been born. However, describ-ing future occupational preferences and distribu-tion is less straightforward, since there are manypaths—not all measurable—for moving into dif-ferent jobs and occupations and many alternativepaths into, out of, and through the labor force.

Adaptability of Labor

A key issue in evaluating the adaptability ofthe labor force to changing labor demands—andtherefore the likelihood of unemployment as aconsequence of technology change-is the willing-ness and ability of people to perform differenttypes of jobs if the jobs they have held, or wouldprefer to hold, become unavailable. Because thisflexibility depends in part on “objective” workertraits such as specialized skills, and in part on“subjective” traits, such as personal preferencesfor certain kinds of jobs, it is difficult to evaluatethe true fit between labor supply and labor de-mand in the wake of circumstances such as tech-nology change that alter employment require-ments. A poor fit may be revealed in under-

“StW, for example, Howard N. Fullerton, “HOW Accurate Were

Projections of the 1980 Labor Force?” Monthly Labor Review, July1982.

23

Table 3.–Noninstitutional Population and the Labor Force, 1929-82(monthly data seasonally adjusted, except as noted)

I 1 Civilian labor force

Non ins t i -

Civilian Iabor forceUnemploy- pa r t i c i pa t i on r a te r

menl rate[pe rcen tof civilian.

l a b o r T o t a l M a l e s F e m a l e s

Year of montht u t i ona l A r m e dPopula- F o r c e s

Emp loymen t

t i o n I Total Unem-

Total Agri- : ’

Nonagr, ployment

Cultural culturalforcej I

Thousands of pesons 14 years 01 age and over Percent

321929

1933

1939

19401941194219431944

1945)9461947

194719481949

195019511952195331954

19551956195719581959

260

250

370

I M39709020

11410

1 I 44034501590

Thous

159114561616

16493098359335473350

30482856279926362551

25142572282721372138

27223122344635343 506

31882816244923262229

21802144213321172088

210221422179

208120862090209220882092

49180

51590

55230

556405591056410555405A 630

538605752060168

Is of person

47630

38760

45750

4752050350537505447053960

528205525057812

s 16 ye,

570385834357651

10450 37180

10090 28670

9610 36140

1 550

12830

9480

81205%026-501070

670

MO22702356

231122763637

328820551 ea318343532

2852275028594 SJ323740

38524714391140703786

33662875297528112832

40935016488243655156

79297406699162026137

76318273

10678

667466746717734879708063

835382878044811180297796

8 C4880327967786081338047

785480538271867390259389

100380101520102610103660104630

105530106520107608

103418104527105611

106645107721108823110601Ill 671

112732113811115065116363117881

119759121343122981125154127224

129236131180133319135562137841

140272143033146574149423152349

153333158294161166164027166951

169848172272I 144.1

168625168846169073169289169494169735

249

172

146 5579 9 56047 57219 5s712 586

I 5724; ‘ 55839 I 568

837843856864870

848826840

9540 379809100 412509250 445009080 453908950 45010

282287313360365

8580 442408320 469308256 49557

359312310

of age and Over

491485071449993

5175853235537495491953904

5572257514581235745059065

60 318605466175963076M 782

6672668915705277210374 2%

593506062161286

789076297658

716067266500626o6205

64506283594755865565

39 583 86438 588 86659 589 864

318327331

53 592 86433 593 84530 590 & 329 589 & O55 588 855

44 593 ‘ 85341 600 85543 596 84868 59 ~ 84255 593 837

55 594 83367 593 82955 588 82057 587 81452 587 810

45 589 807 393592 804 403

: ! 596 804 41136 596 8 0 1 41635 601 798 427

62208 5891862017 5996162138 M 25063015 6117963643 60109

65023 6217066552 6379964929 I 6A 07167639 6303668369 6-4630

339346347344346

357369

: ;371

37738 1379383387

1960 ]

19611962>19631964

6 9 6 2 8 6 5 7 7 8 i 545870459 65746 520070614 667021 494471833 67762 468773091 6 9 3 0 5 , 4 5 2 3

7 4 4 5 5 71088 436175770 72895 3,97977347 74372 3844 ‘78737 75920 381780734 77902 36Q6

19651966196719681969

197019711972 >

1973J1974

197519161977197811979

82771 7867884387 7936787034 8215389429 (33:91949

34633394348434703515

7521575972786698159483279

82438854218873492 66195477

95938’37 03096125

965599651696265958869553795460

9548595582956569596596, 18096,164,

96 544 96803 9714897487 9759797033

97428 973139674696981968406 4 5 8

96309963286 2 3 0961289654896310

961439625496180957639567095682

49 J34 7 9 7 ,59 Q 2 79 I

604 789: 438 78856 1 3 7 8 7

433434439447457

85 1 2 779 ‘77 616 775

623 777 1 632 719 58 637 778,

1 638 774 639 7 7 0

! 640 166,

93775 85846 340896158 88752 333199009 92017 3283

102251 048 3387M 9 6 2 98824 3347

4634734845005439

198019811982

1 0 6 9 4 0 1 9 3 0 3 3364108670 CKI 397 3368110204 9 5 2 6 3401

1 5521526

1980dn 106546 99872 3313

106637 99963 3387106394 99677 4 1 2106552 99204 3318106892 98922 3385106832 98769 3309

636363697575

640 777639 778637 775637 774639 716637 775

5161 5513514515514

FebM a r’Apr,M a yJune

JulyA u gSeptO c tNcv,D e c

198:JanFebMarA p !M a y

170 03C ~ 2099170217 2114170419 2121170624 2121170814 2119171 Oc 7 2124

98816 33319 8 8 2 9 3 2 4 7S$ 104 34489 9 3 2 7 3 3 6 299567 338799650 3486

787775757573

638 775637 773637 773638 773638 773636 770

515515514516516516

171229 2125171400 ‘ 2121171581 2128171770 2129171956 2127172172 2 , 1 3 1

172385 2139172559 21601 7 2 7 5 8 2 1 6 5172966 2 1 5 8173155 2158173330 2164

S9 964 3420lrll 143 3340ICWI 504 335610 I 006 3519100968 3371100393 3360

75 i74731727 57 4

639 773639 772640 773642 774642 ‘ 774638 767

51 8520521523524521June

)UlyAugSeptOctNovOH

19732l a .FebM a rAP{M a yJune

JulyA u gSepfO c tN o vOec

100748,100 7C410010410035510022999677

3320,33963358337433893219

7274

; :8386

6381 768638 769635 767638 767639 768637 767

521521517522523520

86 636 76588 638 76690 638 76593 639 76694 642 77095 6-40 765

173495 2159173657 2168173843 2175 /174020 2176174201 2175174364 2173

996889969599597994849999499681

995889968399543991769913699093

3tltUtll

337933673367335634463371

3445

9346 ~9 6 6 99881

10256103841 0 4 6 6

10828 ~1093111315Ii 5761190612036

521523523524527527

174544 2180174707 21961 7 4 8 8 9 2 1 9 8175069 2188175238 2180175380 ~ 2182

98 ~ 641 765766

1:? ;; 768,105 M 1 761107 642 768108 642 766

529529528527528530

342’33363341334663 4)1

‘Not seasonally adjusted‘Cwlllan labor force as percent of clvdlan non!‘Not strictly comparable with earlier data due to populatl I adjustments as follows” Begmrung 1953, mtroduct!on of 1950

census data added about 600,000 to population and about )0,000 to labor force, total employment, and agricultural employ-ment Begumn 1960, Incluslon of Alaska and Hawall added about 500,000 to population, about 300,000 to labor torce,

8and about 240, 00 to nonagricultural employment Begmnmg 1962, mtroduct!on of 1960 census data reduced populahonby about 50,000 and labor force and employment by about 200,000 Begmnmg 1972, Introduction of 1970 census dataadded about 800,000 to cwdlan nonmstltullonal population and about 333,000 to labor force and employment A subse-quent adjustment based on 1970 census m March 1973 added 60,000 to Iatwr force and 10 employment Beglnrung 1978,changes m samplmg and estlmat!on procedures mlroduced mto the household survey added about 250,000 fo labor forceand to employment Unemployment levels and rates were not significantly affected

II DoDulatlon

SOURCE Oeparfment of Labor Bureau of Labor Statlstlcs

24

Table 4.–Wage and Salary Workers in Nonagricultural Establishments, 1929-82(thousands of persons; monthly data seasonally adjusted)

‘ — T Manufacturing

Mmrsg

1,087744854W;

992925892836862955994930901929898; ; :

792822;3;

732712672650635634:3;

613606619623609628642;3;

779813851958

1,0271,1321,122

1,1021,1131,124

978985

1,1371,1641,1801,1921,1951,2021,206

1,2011,2031,1971,1821,1521,1241,1001,0861,0751,0581,0511,036

Finance,insu-rance,arrdreal

estate——.——

1,4941,2801,4471,4851,5251,5091,4811,4611,4811,6751,7281,8001,8281,8881,9562,0352,1112,2002,2!382,3892,4382,4812,549

2,6292,6882,754;,:8;

2;9773,0583,1853,3373,512

3,6453,7723,9084,0464,1484,1654,2714,4674,7244,975

5,1605,3015,350

5,2525,2645,2705,2865,2955,302

5,3115,3195,3285,3255,3245,331

5,3265,3265,3365,3355,3425,352

5,3595,3605,3675,3575,3625,376

—

Services

;,gg;

3:502

3,6653,9054,0664,1304,1454,2224,6975,0255,1815,240

5,3575,5475,6995,8355,9696,2406,4976,708;,;;;

7,3787,6207,9828,2778,6609,0369,498

10,04510,56711,169

11,54811,79712,27612,85713,441y:;

15:30316,25217,112

17,89018,59219,000

18,35218,38218,41418,48018,51718,556

18,61518,65418,70718,77318,81518,834

18,83118,86718,90418,92918,96318,988

19,04219,04819,08419c07419,12519,143

GovernmentTotalwageand

salaryworkers

rranspor-t~~jn

publlcuthtles

Year ormonth Non.

jurablegoods—..—

5,5645,6226,2256,4586,5186,4726,4506,9627,1597,2566,9537,1477,3047,2847,4387,1857,3417,4117,3217,1167,3037,3377,2567,3737,3807,4587,6567,9308,0078,1558,272;,;:;

8:1028,2628,1527,6357,9208,0868,2318,2808,098~,;::

8,0518,0518,0548,0748,0958,0888,1138,0838,0838,0388,0027,952

7,8957,8797,8297,794;,;;;

7,680;,:;;

7:6597,6287,601

Federal

533565905996

1,3402,2132,9052,9282,8082,2541,8921,8631,9081,9282,3022,4202,3052,1882,1872,2092,2172,1912,2332,2702,2792,3402,3582,3482,3782,5642,7192,7372,7582,7312,6962,6842,6632,7242,7482,7332,7272,7532,7732,8662,7722,733

2,7982,7892,7802,7742,7762,7772,7752,7692,7642,7572,7492,756

2,1412,7372,7362,7302,7282,7392,7372,7392,7342,7232,7262,728

—

Construc-tion State

andlocal

DurablegoodsTotal

——— — -t19291 9 3 31939.

31,32423,69930,603

10,7027,397

10,278

10,98513,19215,28017,60217,32815,52414,70315,545y:

15,24116,39316,63217,54916,31416,88217,24317,17415,94516,675

16,79616,32616,85316,99517,27418,06219,214;;,4:[

20:167

19,367y:;

20:15420,07718,32318,99719,68220,50521,040

20,285;;JU;

20,17120,14820,19720,27520,33220,334

20,37920,31120,26720,09719,90319,676

19,51719,45419,31919,16919,11518,930

18,81318,67218,57218,32518,18318,134

3,9162,6722,936

3,0383,2743,4603,6473,8293,9064,0614,1664,1894,001

4,0344,2264,2484,2904,0844,1414,2444,2413,9764,011

4,0043,9033,9063,9033,9514,0364,1584,2684,3184,442

4,5154,4764,5414,6564,7254,5424,5824,7134,9235,136

5,146y;

5,1395,1455,1535,1635,1585,162

5,1685,1685,1815,1625,1505,128

5,1255,1155,1005,0945,1015,078

5,044:,:;;

5;0074,9944,979

1,;;;

1,165

1,3111,8142,1981,5871,1081,1471,6832,0092,1982,194

2,3642,6372,6682,6592,6462,8393,0392,9622,8173,004

2,9262,8592,9483,0103,0973,2323,3173,2483,3503,575

3,5883,7043,8894,097$:;:

3;5763,8514,2294,463

4,3464,1763,912

4,3154,2404,2674,2814,2234,185

4,1754,1464,1244,1014,0714,026

3,9663,9743,9343,9383,9883,940

3,9273,8993,8833,8563,8483,818

2,5322,6013,090

3,2063,3203,2703,175:,::;

3;3413,5823,7873,948

4,0984,0874,1884,3404,563;,;;;

5:3995,6485,850

6,0836,3156,5506,8687,2487,6968,2208,6729,1029,437

9,82310,18510,64911,06811,44611,93712,138;;,:;:

13:174

13,37513,25313,051

13,40013,41013,37113,35413,30213,243

13,18913,12513,14013,16013,15913,161

13,12313,11313,12313,12213,12513,093

12,89812,93313,02913,019~yl;

4,715

5,3636,9688,823

11,0841:,:;:

7;7428,3858,3267,489

8,0949,0899,349

10,1109,1299,5419,8339,8558,8299,373

9,4599,0709,4809,6169,816

10,40511,28211,43911,62611,895

11,20810,63611,04911,89111,92510,68811,07711,59712,27412,760

12,18712,1 1?11,114

12,12012,097p:

12:23712,246

12,26612,22812,18412,05911,90111,724

11,62211,57511,49011,37511,33211,203

11,13310,99310,90010,66610,55510,533

1 9 4 01941194219431 9 4 419451 9 4 6 ;1 9 4 71 9 4 81949

32,36136,53940,10642,43441,86440.37441,65243,85744,86643,754

19501951 .“1952195319541955. ‘1956195719581959

19601 9 6 119621963196419651966p%: .“

1969

45,19747,81948,79350,20248,99050,64152,36952,85351,32453,268

54,18953,99955!54956,65358,28360,76563,90165,80367,89770,384

19701 9 7 11972197319741 9 7 519761 9 7 719781979 i

70,88071,21473,67576,79078,26576,94579,38282,47J86,69789,823

198019811982 p

1981Jan ,FebMar .’AriayJuneJulyAugSeptOctNovDec

1982;;:. ‘

Mar .“

WYJuneJ u l y

Aug $SeptOctNOVPDec P

90,40691,10589,619

90,90990,91391,01491,09991,13191,286

91,39691,32291.36391,22490,99690,642

89,53589,31289,26788,86088,68488,518

—.Source Department of Labor, Bureau of Labor Statlstlcs

25

employment* and unemployment, and in laborshortages.

Labor shortages exist where a sufficient numberof particular types of people are unavailable forwork at prevailing wages. Concern has been ex-pressed by people in industry and in governmentabout the economic effects of shortages in highlyskilled craft and technical occupations, from ma-chinists to certain types of engineers.** Allegedshortages have been cited as a motivation for in-

● For example, according to BLS many college graduates duringthe 1970’s took jobs not requiring college degrees.

● *The extent of current and possibIe future labor shortages thatmay affect the development or diffusion of programmable automa-tion is unclear. Among the reasons that shortages are hard to measureare the following: 1) Federal programs do not collect occupationalshortage statistics (due to cost and data reliability problems),2) available data do not accurately capture employee mobility withinand between occupations, 3) occupational classifications among firmsand Federal statistical programs are inconsistent, and 4) employerand union surveys tend to be statistically unreliable. A recent analysisby BLS found after evaluating data from several sources that amachinist shortage could be neither established nor disproved. 15

15Neal H. Ro~ntha], “shortages of Machinists: An Evaluationof the Information,” Monthly Labor Review, July 1982.

WORKING ENVIRONMENTIntroduction

Programmable automation may change notonly the numbers and types of people workingin manufacturing, but also the circumstances ofwork—what may be called the working environ-ment. The ways in which programmable automa-tion is applied will determine how it affects theworking environment. This discussion of thepotential implications of programmable automa-tion for the working environment will addresssome of the issues concerning worker safety andhealth, human factors, job content, and structureof work.

Expressions of concern about the effects of tech-nology on the conditions of work have increasedin the United States over the past two decades.For a long time it was assumed by managementthat the benefits of more efficient productionachieved through the introduction of new technol-ogies far outweighed any negative effects on thework force. In other words, the assumption was

vestments in automation, and also in retraining.While retraining can ease shortages by increas-ing the supply of skilled workers, raising wagesis another method of stimulating supply, althoughemployers are often unwilling or unable to do this.Note that, for skills that take years to develop,instituting training programs (or raising wages)will not eliminate a shortage immediately.

A satisfactory analysis of labor supply issuesassociated with programmable automation shouldaddress such issues as contrasts in the composi-tion of the U.S. labor force with that of othercountries producing and using programmableautomation, and the extent to which the produc-tion and use of programmable automation are in-fluenced by labor shortages. Such issues are fun-damental to the identification of components ofthe U.S. labor force that may be particularlyhelped or harmed by the spread of programmableautomation, and the determination as to whetheranticipated changes in the U.S. labor force arelikely to cushion or exacerbate impacts that mightarise from programmable automation.

that people could always adapt in some way tothe requirements imposed by the technology.l6

As in other countries, concerns about work-place conditions contributed to the growth of thelabor movement in the United States. Since themid-1960’s, changing social and economic envi-ronments, characterized by an emerging aware-ness of individual rights and well-being, increasedworker dissatisfaction, and declining productivity,have increased the importance of the working en-vironment to both management and government,as well as labor. Workplace issues in manufac-turing are currently being addressed in a numberof ways, such as: 1) emphasis on human factorsin the design of manufacturing equipment; 2) in-novations in the structure of work; 3) increasedcooperation between management and labor insolving workplace problems; and 4) a variety of

1eJoe] A. Fadem, “Automation and Work Design in the UnitedStates,” in Memational Comparative Study on Automation andWork lki.gn, International Labour Office, Geneva, January 1982,p. 25.

98-951 0 - 83 - 5

26

experiments in worker participation (such as qual- 5. These developments have met with varyingity control circles and quality of working life pro- degrees of success and commitment from manage-grams) intended to give workers greater input into ment and labor. Nevertheless, they are part of thedecisions directly affecting their jobs. See table backdrop to the spread of programmable automa-

Table 5.—Labor-Management Committees on Industrial Relations Issues, Safety, and Productivity by Industry(agreements covering 1,000 workers or more, Jan. 1, 1980)

Labor-management committees on—

Industrial relationsAll agreements issues a Safety b Productivity c

Industry Agreements Workers Agreements Workers Agreements Workers Agreements Workers

75079

8113111174215361514

G88418183

11211

9

800

1662808112

1233166

327

2

i o 2 i 1 5 0234,200

21,80028,850

207,90017,10023,10065,00031,60061,70025,50068,85023,10093,600

460,60097,000

242,150323,750957,100

27,650

14,600

456&650

169,050469,550620,000210,700

23,900405,200148,300323,450

1,195,000

3.500

All industries. . . . . . . . . . . . 1.550 6 5 9 , 8 0 0 - 6 0 2 4 5 , 4 0 0 - ‘ -

Manufacturing. . . . . . . . . . . . .Food, kindred products . . . . .Tobacco manufacturing. . . . .Textile mill products . . . . . . .Apparel . . . . . . . . . . . . . . . . . . .Lumber, wood products. . . . .Furniture, fixtures. . . . . . . . . .Paper, allied products . . . . . .Printing and publishing . . . . .Chemicals . . . . . . . . . . . . . . . .Petroleum refining . . . . . . . . .Rubber and plastics ., . . . . . .Leather products . . . . . . . . . .Stone, clay, and glass . . . . . .Primary metals . . . . . . . . . . . .Fabricated metals. . . . . . . . . .Nonelectrical machinery . . . .Electrical machinery. . . . . . . .Transportation equipment . . .Instruments . . . . . . . . . . . . . . .Miscellaneous

manufacturing . . . . . . . . . . .

Nonmanufacturing . . . . . . . . .Mining, crude petroleum,

and naturai gas . . . . . . . . . .Transportation . . . . . . . . . . . .Communications. . . . . . . . . . .Utilities, electric and gas . . .Wholesaie trade . . . . . . . . . . .Retail trade . . . . . . . . . . . . . . .Hotels and restaurants . . . . .Services . . . . . . . . . . . . . . . . . .Construction . . . . . . . . . . . . . .Miscellaneous

nonmanufacturing. . . . . . . .aA labor-management committee on industrial relations issues is a joint committee which studies Issues, fOr OXW@% subcontracting, seniority, and wa9e incentives,away from the deadlines of bargaining and makes recommendations to the negotiators. It also may be referred to as a “prebargaining” or “continuous bargaining”committee. It should not be confused with labor-management committees which meat periodically to discuss and resolve grievances and in-plant problems.

bA Iabor.management safety committee is a joint committ~ which meets periodically to discuss safety problems, to work out solutions, and to itTIpbfTlOIIt SafOty

396

———

41111

—4

—1724331

—21

1* 15025,500

———4,8501,0001,1001,0001,200—

29,250—1,000

40,1503,200

10,3508,200

20,0001,350

214222

—53

—

6,0009,000

45,0004,3003,9502,200—

22,1004,700

—

572

35—

78

183

211014

2267625484268

4

6

159

13223735

110

34

—

2,867,8501,835,550

140,400—1,2001,0009,9507,400

27,65010,80030,85018,90068,8503,200

66,550429,70066,150

141,800130,300656,150

16,700

8,000

1,032,300

161,200289,400316,050108,050

1,05019,05010,0008,800

118,700

81 1,091,350

585

———

11121

—2

——33

32

—7

—

—23

845,30069,700———1,0001,0001,2009,1002,000

—16,450——

316,8505,0502,100

—420,850

—

246,050

312

12

———

23

—

10,100208,350

1,5504,900

———3,650

17,500

—

programs in the plant.C A Iabor.management committee on productivity iS a joint committee which meets periodically to discuss in-plant production problems and to work Out methods O f

improving the quantity and quality of production.dExcludeg railroads and airlines.

NOTE: Nonadditlve.SOURCE: U.S. Department of Labor, “Characteristics of Major Collective Bargaining Agreements, January 1, 19S0.”

27

tion, and will influence how these technologiesare implemented and how they affect the overallconditions of work.

Occupational Safety and Health

Occupational safety and health issues may beclearer than others associated with programmableautomation. For example, the application of ro-bots to painting and welding tasks is widelyacknowledged as a measure that reduces workerexposure to occupational hazards by removingworkers from the hazards. However, the use ofrobots and other forms of programmable automa-tion may give rise to workplace hazards that arenew and perhaps unanticipated.

The hazards associated with programmableautomation are likely to be similar to those asso-ciated with industrial machinery, video displayterminals (VDTs), and other types of equipment.With the introduction of programmable automa-tion, there may be a shift of occupational safetyand health concerns in manufacturing away fromthose directly involving machinery toward VDT-related issues. VDTs will become more numerousin manufacturing, and one possible outcome ofthe spread of programmable automation is an in-crease in the percentage of manufacturing workersusing VDTs and a decrease in the percentageoperating machinery. The eyestrain, stress, andback, neck, and shoulder problems recently docu-mented by the National Institute for OccupationalSafety and Health among workers who use VDTsfor extended periods of time may become a prob-lem for those using computer-aided design andmanufacturing (CAD/CAM) systems.17

Unlike many older manufacturing technologies,programmable automation technologies are beingdeveloped in an era of greater awareness of occu-pational safety and health issues. Part of that con-text includes a body of Occupational Safety andHealth Administration (OSHA) standards as wellas a sophisticated set of nongovernmental techni-cal standards. The applicability of current OSHAstandards to the use of programmable automa-tion will depend on the type of industry or nature

“’’Health Hazards for Office Workers: An Overview of Problemsand Solutions in Occupational Health in the Office, ” WorkingWomen Educational Fund, 1981, p. 22.

of the operation involved. It is unclear whetheror not programmable automation may give riseto a need for further OSHA standards.

Human Factors

Programmable automation may change theway job performance is evaluated in manufactur-ing. The computer and communications capabil-ities of programmable automation permit the re-cording and monitoring in remote locations ofmany aspects of equipment and system utilization,such as the number of operations performed perminute or per hour. Such monitoring would pro-vide management with more information thanindividual piece counts conducted at the end ofa day or week and other traditional measures ofperformance. Although sophisticated monitoringfunctions are not a necessary feature of program-mable automation products, their possible usemay reduce worker discretion in performing tasksand raise levels of stress among workers. Suchresults have been observed where office automa-tion has been implemented with sophisticatedmonitoring features, such as tabulation of key-stroke-per-minute rates.

18 On the other hand, ifprogrammable automation requires fewer workersper machine, it may reduce the amount of directpersonal supervision required.

Many of the effects, both physical and psycho-logical, of programmable automation on peoplein the workplace will depend on the care andthought that go into the basic design of automatedequipment and systems, and on whether the de-signers are concerned about human factors issues.Consideration of human factors involves first ana-lyzing the roles people will play in a working envi-ronment using programmable automation, andthen “examining such human factors engineeringissues as design, procedurization, and protec-tion.” 19 Design engineers who do not work on theshop floor or in other manufacturing settings maynot be sufficiently sensitive to the physiological

‘sJudith Gregory, Testimony for 9 to 5, National Association ofWorking Women, Hearings, House Subcommittee on Education andLabor of the Committee on Education and Labor, June 23, 1982,p. 10.

I~H. MCI]vaine parsons and Greg P. Kearsley, “Human Factorsand Robotics: Current Status and Future Prospects, ” Human Re-sources Research Organization, Alexandria, Va., prepared for U.S.Army Human Engineering Laboratory, October 1981, p. 13.

28

and psychological needs of the user. Whether ornot the user is involved in the design process maydetermine to what extent the human needs ofmanufacturing personnel will be translated intoequipment and systems designs.20

Although worker involvement in the designprocess would seem logical on the surface, it mayalso present a dilemma for manufacturing employ-ees. While on the one hand their participationcould improve the consideration of human fac-tors, it could also facilitate the design of equip-ment that may eliminate jobs. This dilemma mayinhibit the full participation of many workers insuch activities as quality control circles and qual-ity of working life programs.

Job Content

Programmable automation may affect job con-tent in a number of ways and its impact on skillrequirements is likely to be highly variable. Bydesign, automated equipment and systems mayalter the skills required for certain aspects of theproduction process, but the implications for spe-cific jobs (e.g., in terms of the number and vari-ety of tasks comprised) depend on how program-mable automation is implemented. The impactsof programmable automation on skill levels areuncertain. While some jobs clearly will require ahigher level of skill, others may require a lesserlevel, largely because much of the process-controldecisionmaking may be incorporated into com-puter-controlled equipment and systems. It isunclear at this time whether the effects on skilllevels are inherent in the programmable automa-tion technology, or the extent to which innovativeuse provides a choice. Whether programmableautomation will provide jobs that are more stimu-lating and satisfying overall than those in tradi-tional manufacturing environments is uncertain.However, it is unlikely that all programmableautomation jobs will provide more challenge,variety, and responsibility-nor does everyone re-quire it.21 There will probably always be monoto-

2oFadem, op. Cit., p. 51.21Enc Tfist, ‘The Evolution of Socio-Technical Systems: A COI’I-

ceptual Framework and an Action Research Program, ” OccasionalPaper No. 2, Ontario Quality of Working Life Center, June 1981,p. 32.

nous jobs, and many workers accept this in returnfor such other benefits as fair wages and jobsecurity .22

Depending on how tasks are organized, pro-grammable automation may allow an increase inthe variety of tasks a worker performs.

Therais also a close relation between the man-ufacturing technology chosen and the organiza-tion of work. However, technology is not thesingle determinant, so there is no specific organi-zation corresponding to the use of a CAD/CAMsystem. Organizational philosophy has a pre-dominant role, for example if one believes incomplementary specialization of skills or inoverlapping skills. The CAD/CAM may be aloyal servant to any work organization, providedthat those who design and adapt the system knowwhat they want.23

A restructuring of work in which both technicaland human considerations are given equal treat-ment could offset the negative effects of chang-ing skill requirements that may arise where oldpatterns of work organization persist.

Programmable automation may lead to chang-ing roles and responsibilities at all levels, affect-ing both the nature of jobs and the distributionof power. The difficulties of reorganizing com-panies are well recognized. For example, changein the hierarchical structure (and thus control)brought about by the introduction of new tech-nology may meet with resistance from those whomight lose some authority .24 Consultants andtrade and professional associations concernedwith programmable automation have devotedmuch attention to the management challenges ofsuccessful use of programmable automation overthe past few years. Clearly, management plan-ning, practices, and policies will be key factorsin how the introduction, implementation, andoperation of programmable automation affectsthe overall working environment.

~ZSar A. ~Vitan and Clifford M. Johnson, t%?comi T.~O14?~~5 onWork (Kalamazoo, Mich.: W. E, Upjohn Institute for EmploymentResearch, 1982), p. 212.

23’The Promotion of Robotics and CAD/CAM in Sweden,” reportfrom the Computers and Electronics Commission, Ministry of Indus-try (Sweden), 1981.

ZtRo~rt Schrank, Ten Thousand Working Days (Cambridge,Mass.: The MIT Press, 1978), p. 221.

Chapter 3

Education, Training,and Retraining

Chapter 3

Education, Training, and Retraining

INTRODUCTIONAt various stages in U.S. history, changes in

workplace operations and procedures in all sec-tors of the economy have resulted in changes ineducation, training, and retraining requirementsfor those employed or preparing for employment.Changes in instructional requirements for manu-facturing-related work have been particularlydramatic. In some instances, they have been soextensive and widespread that they have triggeredchanges in the structures of institutions andorganizations engaged in the delivery of educa-tion, training, and retraining services or theemergence of new instructional providers. For ex-ample, new production techniques introducedduring the Industrial Revolution had much to dowith the creation of a system of free public educa-tion, since large-scale production and continuedindustrial expansion required a literate work forcecapable of functioning on production lines, super-vising manufacturing operations, keeping admin-istrative records, and performing other functions.During this period, both industry and the labormovement became involved in the design and im-plementation of instructional programs to addressshort-term and special needs they felt could notbe met either within a system of general instruc-tion or through the public and private vocationalprograms that were emerging.1

Another era of substantial industrial changeoccurred in the 1960’s, when the aerospace/de-fense industry underwent tremendous expansionand mechanization as a result of a concern overnational defense and a national commitment tomanned space exploration. Under provisions ofthe National Defense Education Act,2 an exam-ple of legislation that led to the establishment ofnational policy for certain forms of occupational

IFor additional information on the history of education and train-ing in the United States, see Mormationa] Technology and Its Im-pact on American Education (Washington, D. C.: U.S. Congress,Office of Technology Assessment, November 1982), OTA-CIT-187.

‘National Defense Education Act of 1958 (Public Law 85-864, Sept.2, 1958); National Defense Education Act Amendments, 1964 (PublicLaw 88-665).

instruction, U.S. educational institutions werecharged with coordinating efforts to preparethousands of individuals for careers in science,engineering, and related fields. Training in thesefields was considered necessary in order todevelop the expertise and the technological baseessential to creating a strong system of nationaldefense and to meeting the challenges of space.Rapid technological change in aerospace/defenseand other industries affected by these nationwideefforts required the continued involvement ofbusiness and labor in specialized instruction, toensure that skill levels advanced at the same rateas applications of new machinery. However,when national priorities changed, this cross-sectorcommitment to linking advances in technologywith the upgrading of skills within aerospace/de-fense and related industries disappeared. Sincethat time, the development of the human resourcefor these industries has been approached in moreparochial ways by business, labor, educators, andgovernment.

In these and other instances of changing educa-tion and training requirements, three factors haveimpeded the development of a coordinated, flex-ible system for occupational instruction in theUnited States. First, the absence of long-range,public and private projections of skill require-ments, particularly those that highlight changesin skill levels and in core skill requirements withinoccupations, has hindered the development anddelivery of instructional programs before indus-trial demand reaches critical proportions. * For ex-ample, the Electronic Industries Association re-ports that within the software technology field,certain highly specialized skills possessed by elec-trical engineers and computer scientists are inter-changeable. Although not captured in more for-

● There are those who would argue that establishment of a coor-dinated, flexible system for occupational instruction in which pro-jections of national as well as regional demand are taken into ac-count is not the best approach and that actual rather than projecteddemand within local labor markets should determine public and pr-ivate sector human resource development activities.

31

32

ma] projections, this interchangeability will affectrecruitment strategies and, in turn, supply withinboth occupational groups. The net result may bethat current and projected shortages of computerscience graduates may also be seen in electricalengineering and yet may not be reflected in for-mal occupational projections. * Second, a historyof responding to changing industrial skills require-ments as crises arise has perpetuated a fragmentedapproach to education and training. Little or noplanning or coordination of efforts takes placeamong traditional educators, business, labor, gov-ernment, and others. Third, rapid technologicalchange has placed great strain on educators asthey attempt to adapt instruction to the require-ments of new technology, while at the same timethey address other changes in instructional needs.

The application of programmable automationin manufacturing operations has the potential totrigger widespread changes in education and train-ing requirements. Robots and other forms of pro-grammable equipment and systems may changethe organization of the manufacturing process, thecharacter of the production line, the occupationalmix, and the human-machine relationship. Theutilization of programmable automation, depend-ing on its impact on employment levels withinspecific occupations, may also necessitate theretraining of individuals for occupations in othersectors.

This section of the technical memorandum ex-amines the changing role of education, training,and retraining in the United States; describes howindustry and labor engage in instructional servicesdelivery; presents some current views held by rep-resentatives of industry, labor, and the educa-tional community concerning changes in instruc-tional requirements and providers; and outlinesselected critical issues for those engaged in instruc-tional design and delivery, in light of possiblewidespread use of programmable automation.This picture of education, training, and retrain-

“Alternatively, it could be argued that electrical engineers, givenan adequate supply, could be recruited to fill computer science jobsand eliminate projected shortages for that occupational group, butnot in ways that could be foreseen by current methods of formalprojection. In addition, recruitment for computer science jobs fromthe ranks of individuals with liberal arts training such as music andforeign languages, a practice that has proven quite successful, is notcaptured in formal projections of demand for those disciplines.

ing, when viewed in light of present and futuretrends in the labor market, should facilitate theidentification of new opportunities, problems, andissues in education and training policy.**

Changing Role of Education,Training, and Retraining

Formal instruction has always been viewed asan important part of the human developmentprocess in the United States. Since the colonialperiod, a variety of institutions and organizationshave been established to deliver education and/ortraining services to the general public or to specialsegments of the population. Some of these institu-tions and organizations consider the provision ofinstructional services their primary mission;others, such as corporations and labor unions,view education and/or training as one of a num-ber of activities in which they are engaged. Therecent OTA study, Informational Technology andIts Impact on American Education, found thattoday, instructional services are available froman even wider variety of sources, including elec-tronic-based services delivered directly to thehome.3

As U.S. economic and social conditions havechanged over the years, the role of education andtraining in the lives of all citizens has changed aswell. Formal instruction was once viewed as a lux-ury that was unavailable to a large percentage ofthe population. Then, after a system for publiceducation was established, the role of instructionbecame the initial preparation of young peopleto assume responsibilities as productive numbersof society. In the 1980’s, instruction has come to

● *A number of investigations are now underway in the privatesector that could considerably improve the understanding of chang-ing education, training, and retraining requirements in general andrequirements related to the utilization of programmable automa-tion in manufacturing in particular. One such effort is a survey ofeducation and training representatives in 1,000 corporations, con-ducted by TrainingMagazine, designed to identify current instruc-tional needs and in-house program content. A second survey, ini-tiated by the Society of Manufacturing Engineers (SME), has beendesigned to elicit from a sample of the SME membership, as wellas from selected educators, views on instructional requirements thatmay stem from the application of computer-aided design (CAD) andcomputer-aided manufacturing (CAM). Knowledge derived fromthese investigations and others in progress will be incorporated intothe final report.

31nformational Technology and Its Impact on American Educa-tion, op. cit.

33

be seen as a lifelong process that enables in-dividuals of all ages to cope with economic andsocial change.

Questions concerning who receives instruction,who determines the content, who provides the in-struction, what modes of delivery are utilized,how much instruction costs, and who pays forit have received considerable discussion through-out modem U.S. history and have served to shapenational education and training policy. Educationand training for work and who should providethem have been controversial subjects since theIndustrial Revolution.

Participation in Instructional Programs

Due to a variety of forces, including the ac-celerating rate of technological change and growthin foreign trade, some workers, especially thosein manufacturing environments, are finding thattheir skills are not adequate either to continue toperform their current jobs or, if they are displaced,to secure new jobs. Others may find that they areoverskilled for their positions, due to the introduc-tion of computer-based equipment and systemsas well as other workplace changes. Changes inskill requirements and skill levels are affecting allmanufacturing occupations, from the productionline worker to the professional engineer. Inresponse, many individuals are seeking additionaltraining in order to keep pace with technologicaland economic change, although it is unclear fromavailable data which occupational groups theymay represent and whether they are predominant-ly white- or blue-collar. The relative quality ofthe instruction, and therefore its usefulness, is alsodifficult to determine from the available data. TheCurrent Population Survey’s Special Survey ofParticipation in Adult Education revealed thatover 21 million persons 17 years old and over,or some 13 percent of the adult U.S. population,participated in adult education programs in 1981.An analysis of enrollments in 37,381 courses re-vealed that approximately 60 percent had par-ticipated for job-related reasons. In 9,260 cases,courses were provided by employers; in another12,287 cases, employers paid enrollment fees forcourses delivered outside the company. Profes-sional and technical workers comprised the largestpercentage of those enrolled—some 30 percent

98-951 0 - 83 - 6

(there is no distinction made in the survey as towhether respondents are salaried or hourly em-ployees). About 54 percent of the participantswere under 35 years of age; 12 percent were over55.4

Industry and Labor asInstructional Providers

Since the mid-19th century, both business andthe labor movement have contributed to or par-ticipated in the design and delivery of instructionalprograms. Formal, in-house instruction is morecommon in larger business and labor organi-zations. At present business and labor both spon-sor a variety of employee education and trainingactivities, such as secondary-level remedialcourses, traditional apprenticeships, and postsec-ondary degree and nondegree programs.

The American Society for Training and Devel-opment estimates that U.S. industry now spendsapproximately $40 billion annually on educationand training programs for employees. The esti-mate excludes instructor fees or other adminis-trative costs, such as equipment and enrolleetravel expenses. Although rather dated, the resultsof a study of corporate-based training and educa-tion conducted by the Conference Board in 1974-75 indicated that in 1973-74, 75 percent of the 610firms responding provided some in-house courses,89 percent had tuition aid or refund programs,and 74 percent sponsored the enrollment ofselected employees, usually managers and profes-sionals, in courses offered outside the companyduring working hours. * The Conference Boardestimated that in firms with 500 or more employ-ees, with a combined employee base of 32 million,about 3.7 million, or 11 percent, were enrolledin in-house courses during working hours andanother 2 percent (or 700,000) were enrolled dur-ing nonworking hours. Participation was morecommon for salaried than for hourly employees.The survey also showed that firms with less than1,000 employees relied more on hiring trained in-dividuals and on informal, on-the-job training.’

“’Participation in Adult Education, May 1981, ” National Centerfor Education Statistics Early Release, June 1982.

*Roughly 1.3 million employees in responding companies weretaking advantage of tuition assistance programs.

‘Seymour Lusterrnan, Education in Industry (New York: The Con-ference Board, Inc., 1977), pp. 11-12.

34

This finding is consistent with other evidence thatfirms with up to 500 employees depend on infor-mal, on-the-job training for all types of staff, sinceformal instruction programs are often too expen-sive for businesses of this size. b

Labor unions and labor organizations are alsoactive sponsors and providers of employee educa-tion and training. Historically, unions have pro-moted liberal arts education in addition to morenarrowly focused occupational education. Laborunions and labor organizations have been a strongforce in shaping the popular view of educationas the key to social and economic advancement.7

Like some companies, unions sponsor 2- and4-year degree programs at community colleges,colleges and universities, as well as single coursesin labor studies. Unions also cosponsor appren-ticeship programs with industry, providing spe-cialized training in a skilled trade, craft or occupa-tion at the worksite, and on-the-job instruction.

According to Bureau of Labor Statistics (BLS)figures, at the close of 1979 there were 323,866persons enrolled in apprenticeship programs. Un-published BLS estimates of apprenticeship enroll-ments are 320,000 persons in 1980, 316,000 in1981, and 287,000 in 1982. While the drop in en-rollment reflects reduced public and private fund-ing levels, unions report there is no evidence tosuggest that interest in apprenticeship has de-clined. Reductions in U.S. Employment andTraining Administration apprenticeship andpreapprenticeship grants to individual laborunions, labor organizations such as the AFL-CIO’sHuman Resources Development Institute, andcommunity-based organizations such as the Na-tional Urban League, have diminished recruitmentfor and enrollment in apprenticeships. Deteriorat-ing economic conditions within industries pro-viding apprenticeship opportunities may beanother factor in declining enrollments.

Manufacturing-Related Instruction

Management and labor have been the majorproviders of employee instruction beyond initial

bInterview with Jerome Pelaquin, Chairman, Technical and SkillsTraining Special Interest Group, American Society for Training andDevelopment, July 1982.

‘Paul E. Barton, Workhk Transitions: The Adult Learning Con-nection (New York: McGraw-Hill, 1982), ch. 7.

occupational preparation, since public sponsor-ship of training and retraining of noneconomicallydisadvantaged adults has been limited. The extentto which individuals working in manufacturingparticipate in education, training, and retrainingprograms offered by industry, labor unions, orother private or public providers is unknown.However, within corporate-based instruction asa whole, fewer courses are designed specificallyfor production line workers (excluding apprentice-ship) than any other occupational group. Train-ing industry representatives suggest that corporateinstructional efforts have not previously concen-trated on technical training other than in programsprovided for engineering or data processing per-sonnel. * This situation may change as U.S. manu-facturing firms become familiar with new designand production technologies.

Another reason that technical and skills train-ing do not receive more emphasis in corporate-based instructional programs may be their relativecomplexity, which requires demonstration of skillas well as knowledge transfer. This implies thattechnical courses and programs must emphasizehands-on practice and include a performance testto ensure mastery of skills.8 Technical and skillstraining also require instructors with an under-standing of current manufacturing processes aswell as in-depth knowledge of the subject mat-ter. These requirements result in high costs forestablishing and maintaining an instructional pro-gram. In nonunion facilities, work-related instruc-tion may depend on the advantages companiessee in providing continuing education experiencesin-house or through tuition reimbursement plans.

Different views exist as to how much emphasisunions place on instruction beyond apprentice-ship. Some suggest that such instruction hasheretofore not been a great concern of unionmembers who therefore have addressed it in labor-management agreements in only a general way,through tuition reimbursement provisions. How-

“Technical training is the term commonly used to describe work-related instruction for individuals who perform technical proceduresor who work in environments where technologies have been applied.

‘Stanley J. Holden, “Business, Industry and Labor: Linkages Be-tween Training and Employment,” in -lob Training for Youth RobertE. Taylor, Howard Rosen, Frank C. Pratzner (eds.) (Columbus,Ohio: The National Center for Research in Vocational Education,The Ohio State University, 1982), p. 360.

35

ever, since the late 1960’s, layoffs in auto, steel, future emphasis on education and training benefitsand other industries have given rise to agreements in labor-management agreements.9

featuring education and training benefits. One in-dicator of expanded worklife learning may be ‘Barton, op. cit., pp. 125-126.

TECHNOLOGY AND MANUFACTURING: CHANGES ININSTRUCTIONAL REQUIREMENTS

Discussions of the impacts of automation oneducation and training for the American workerare hardly new. The National Commission onTechnology, Automation, and Economic Prog-ress (hereafter “Automation Commission”), in its1966 report, noted shifts in skill requirements oc-curring during that decade.l0 The report cites rap-idly increasing employment levels of the highlyskilled, as manifested in a technical work forcethat had grown from 6.6 percent of the total in1947 to 12.2 percent of the total in 1964, and sig-nificant shifts during the same period from manualto white-collar work. The Commission reportnotes the trend toward more formal schooling,particularly higher education, as well as the grow-ing education gap between the skilled and the un-skilled. The Commission observed:

The encouragement of an adaptable labor forcefostered through education and training is secondin importance only to the provision of adequateemployment opportunities in the facilitation ofadjustment to technological and other change . . .We wish to emphasize at the onset that we regardthe goals of education as far transcending eco-nomic objectives. These goals go beyond eco-nomic progress to the development of individualsas persons and as responsible citizens. A cleardivision of education into its “economic” and“noneconomic” aspects is impossible . . . Fromthe purely economic point of view, education hasthree principal effects: I) it can increase the ver-satility y and adaptability of people with respect tochange; 2) it can open up increasing opportunityto persons who might otherwise have difficultyin finding and holding employment; and 3) it canincrease the productivity of workers at any levelof skill or ability. Though education is much morethan a means of economic progress, it is a decisive

IOTw~o]oa and the Amerjcan E20nomy: Report of the NationalCommission on Technology, Automation and E20nomic Progress,vol. 1, February 1966.

factor in the economic advancement of any coun-try.11

Commission members acknowledged growth incorporate-based employee instructional programs,and they considered widespread basic skills defi-ciencies an impediment to future economicgrowth. Among their recommendations were:

●

●

●

●

●

provision of quality compensatory educationto all who need it;improvement of “quality and quantity” ofprimary and secondary education, especial-ly in economically depressed areas, in orderto achieve equity of access and equity ofopportunity;universal high school graduation;deferral of vocational training until after highschool, to ensure that individuals receive ageneral education to prepare them for subse-quent occupational education and to instillan appreciation for education as a continu-ing process” . . . indispensable for continuedadaptability in a changing world . . . ;“availability of education, training, andretraining to individuals throughout theirlives.

The Commission also proposed a nationwide sys-tem of public education lasting 14 years withdirect links between high school curricula andthose of community colleges and technical schoolsdesigned to prepare individuals for technical andparaprofessional careers.l2

Education, Training, andEconomic Growth

The concerns of the Automation Commissionhave reemerged in various forms. While the

“Ibid.121bid, pp. 4s-47.

36

Automation Commission’s report focused on therole of education and training as a complementto technological change in stimulating nationaleconomic development, more recent studies focuson education and training and new technology asfactors in regional, State, and local economicgrowth. A 1982 study published by the Northeast-Midwest Institute cites basic skills deficiencies asa critical problem already depressing economicgrowth rates in the Northeast and Midwest andthreatening U.S. participation in internationalmarkets. That study recommends a unified policyfor training, retraining, and skills upgrading forall workers.13

Other observations on the relationship betweeneducation, training, and economic growth arebeing made on the State level, as public andprivate groups explore the relationship of humanresource development and continued economicadvancement in their respective geographic areas.A Connecticut Business and Industry Associationstudy has found that an appropriately trainedwork force is the strongest influence on locationdecisions of advanced technology companies andis critical to expanding that State’s electroniceconomy. That study recommends: 1) divertingresources within Connecticut education institu-tions to programs that graduate individuals qual-ified to enter high technology industries, as wellas 2) publicizing existing, in-State continuingeducation programs for working, corporate-basedprofessionals.14 Another study, conducted for theNew York State Science and Technology Founda-tion, found that universities could participate inState economic development through cooperativeuniversity/industrial education programs, coop-erative university/industry research and develop-ment programs and improved responsiveness tounique industry needs.15

In a series of recent papers on higher educationand technological innovation, the New England

IJPat Choate, ~e~OO@ the American Work Force TOwarda ~a-tional Training Strategy (Washington, D. C.: Northeast-Midwest In-stitute, July 1982).

14An~ Wingate and H. Craig Leroy, E@h Techology ~dustriesand Future Jobs in Comecticut (Business and Industry Association,December 1981), p. 10.

15 Special Report V: The Higher Education System in New Yorkand Its Potential Role in Economic Llwelopment, prepared for NewYork State Science and Technology Foundation by BattelIe-Columbus Division, April 1982.

Board of Higher Education responded to chargesof unresponsiveness made by business leaders andothers concerned about changes in occupationalsupply and demand by redefining the problem asa need on the part of educators for:

. . . hard numbers on the regional supply oftrained personnel, and correspondingly, projec-tions of demand based on reasonably firm busi-ness plans . . . , a clearly assumed responsibility yfor the regular collection of such statistics and forthe underwriting of expenses associated with con-tinuing projects of this nature, (and) an organiza-tional structure whose mission is to gain consen-sus from leaders of the business, education andgovernmental communities on the regionalneeds . . . and on the appropriate goals andstrategies by which they can be attained. Plansfor implementation would, ideally, include a cleardemarcation between short- and long-termissues .16

Technological Literacy

It is possible that the United States is enteringan era in which the potential for mechanizationin the factory and the office will dramatically alterwork force skill requirements. This will requireemployees and individuals preparing to enter thejob market to enhance skills and/or to developnew ones. The OTA study, Informational Tech-nology and Its Impact on American Education,found that in order to function as citizens in aninformation-based society that is driven in largepart by technological innovation, individualsmust have knowledge of the computer as a toolfor managing and providing access to massiveamounts of information. This need to understandthe applications of computer technology has re-sulted in a modified definition of basic literacythat includes familiarity with the computer.“Technological literacy” is now a common termused to describe a level of understanding of tech-nology in its various forms that goes beyond afamiliarity with the computer. Experts suggest thattechnological literacy will soon be required of allmembers of the work force, as broader and moreextensive applications of information technologyare made in offices and plants. Widespread tech-

1“’Engineering and Technological Education in New England—Part II: Alternatives for the Eighties,” issues in Planning and Policy-making, December 1981.

37

nological literacy may be hard to achieve, how-ever, since about one-fifth of the U.S. populationhas yet to master the basic skills of reading, writ-ing, and arithmetic.17

Industry representatives have expressed grow-ing disillusionment with the lack of employabili-ty skills in entry-level workers with educationalpreparation through the graduate level. They de-fine employability as an individual’s understand-ing of the basic rules of the workplace, includingthe need to report for work, to arrive on time,to stay with a job for a reasonable period, andto demonstrate competence in the basic skills. Thishas led many companies to increase their involve-ment in education on the local and even nationallevels and to establish more in-house corporateeducation and training systems.18 Labor unionsand labor organizations have also been vocal intheir concerns about basic skills deficiencies in

“’’Ahead: A Nation of Illiterates?” U.S. News and WorldReport,May 17, 1982; National Center for Education Statistics.

18Jam= Campbe]], “Emp]oyers Expect the Best, ” reprinted fromVocational Education Journal of the American Vocational Associa-tion, vol. 55, No. 8, October 1981.

NEW TYPES OF INSTRUCTIONALAs a result of research performed to date, OTA

has identified several instructional programs toprepare individuals to function in computer-auto-mated manufacturing environments. It is too earlyto say whether or not the establishment of theseprograms constitutes the beginning of a trend, orto make qualitative evaluations, but the existenceof these programs does indicate that some busi-ness, labor, and government representatives areaware of a skills gap in manufacturing firms whereprogrammable automation has already been ap-plied. It is important to note, however, that theseprograms are scattered; by no means do they con-stitute a coordinated attempt by the public andprivate sectors to address the problem of a poten-tial widespread skills gap. At this stage in the in-vestigation, it appears that the evolution of theseand similar programs is occurring in traditional,uneven fashion, and that the capacities of educa-tional institutions and other instructional pro-viders would fall severely short of the potential

those seeking apprenticeships. One union educa-tional representative has found weak communica-tions and reasoning skills common among traineestoday. Many union locals establish close work-ing relationships with school districts to improvebasic skills, while national and international labororganizations address these problems by work-ing with national education groups.

Although many elementary and secondaryschools, both public and private, are placingrenewed emphasis on basic skills development,and many adult education programs offer re-medial courses in math, reading, and writing,these programs reach only some of the individualswho need this type of instruction. In addition,public school systems are hampered in modify-ing and strengthening curricula as a result of lowerlevels of Federal funding and reduced State andlocal tax revenues in many areas. These condi-tions complicate the process of developing strongbasic skills and technological literacy among thosepreparing for entry into, and those already in, thework force.

PROGRAMSdemand for skills development and upgrading thatmay be associated with the widespread adoptionof programmable automation.

Secondary-Level Programs

Since the establishment of a public educationsystem, local school districts have attempted todevelop secondary-level programs that achievetwo distinct ends: 1) the preparation of some in-dividuals for direct entry into the work force im-mediately after graduation; and 2) the prepara-tion of others planning to enter college who re-quire a strong foundation of knowledge on whichto base more advanced instruction. Since lifelonglearning is likely to become necessary for all mem-bers of the labor force, these objectives are be-coming blurred. For example, in some highschools serving areas where programmable auto-mation is now being produced or used in manu-facturing, there are indications that students not

38

going on to college are receiving more attentionthan in the past, and that career exploration forhigh technology careers is recognized as impor-tant for all students, regardless of their postgrad-uation plans.

The State of Michigan, due to its economic de-pendence on auto and truck manufacturing, hasbeen hit hard by massive layoffs over the past fewyears. With high unemployment among manufac-turing workers in the region, some local highschools in southeastern Michigan have been look-ing for new career opportunities for which theycan begin to prepare their students.

Several Michigan school systems, includingOakland County, have added introductory robot-ics courses to their curricula. These courses givestudents an opportunity to learn first-hand aboutrobotics technology and to explore career oppor-tunities within this manufacturing-related field.Students learn to operate simple, tabletop, elec-tric robots, which are provided to the schoolsystems by local robot manufacturers, or buildtheir own robots. In some cases, courses includesite visits to local auto manufacturing plants toobserve robot applications in welding and paint-ing. There are no prerequisites for juniors andseniors who wish to enroll. It is important to notethat these courses do not purport to developentry-level job skills in students, but are offeredsimply as an opportunity to develop some meas-ure of career awareness in high technology. Also,there are no formal placement services providedand at present no links to more advanced roboticstechnology instruction.

An experiment is underway in Oakland Coun-ty, where segments of a successful summer highschool robotics course have been incorporatedinto the curricula of several regional vocational-technical centers. Courses offered through thesecenters are open to high school students as wellas adults who wish to explore interests and careeroptions in the field of robotics.

Retraining for Skilled andSemiskilled Occupations

Unions and others concerned about the poten-tial social impacts of the use of programmableautomation have been active in promoting the

need for retraining programs for skilled and semi-skilled occupations. They regard education andtraining as tools for strengthening the job securityof and alternative job opportunities for their mem-bers. Through collective bargaining and othermeans, unions are looking for ways to influencewho is trained and what is taught in retrainingprograms. In particular, unions are looking atways to have more control over in-plant train-ing to upgrade skills and to modify standard tui-tion refund programs to provide members withmore opportunities to participate in education andtraining programs outside the workplace .19 Thisposition is in keeping with provisions included inselected agreements of the 1960’s, when earlierforms of manufacturing automation were appliedin steel, electronics, and aerospace firms.20

The United Auto Workers (UAW) and the In-ternational Association of Machinists (IAM) areamong the most active unions in promoting tech-nology-related education, training, and retrain-ing opportunities for their respective member-ships. Within 1982 agreements UAW reached withFord Motor Co., General Motors, and Interna-tional Harvester, there are provisions for train-ing and retraining programs for current employeesas well as those laid off. In addition, each con-tract calls for the establishment of a joint union-management employee development and trainingcommittee through which special instructional as-sistance will be provided to members who are dis-placed by new technologies, new techniques ofproduction and “ . . . shifts in customer pref-erence. ” Employees—both skilled and semiskilled—are covered under other provisions of the agree-ments. They are eligible to participate in upgradetraining designed to sharpen job skills and to pro-vide updates on the state of the art of technologybeing utilized in their plants.21 22 23

“’’Retraining: The Need for Flexibility,” in Silicon, Satellites andRobots: The Impacts of Technological Change on the Workplace(Washington, D. C.: Department for Professional Employees, AFL-CIO, 1979), pp. 44-45!

20S= Rwent C’o]]=tive Bargaining and Technological change

(Washington, D. C.: U.S. Department of Labor, Bureau of LaborStatistics, March 1964), BLS Report No. 266.

‘]’’Programs Set Up for Training and Retraining,” UAW-GM Re-port, March 1982, p. 11.

22’’T’raining, Retraining Plan to Cover Ford Workers,” UAW-FordReport, February 1982, p. 7.

Z3CJA w-~temationa] Mw-vester: Highlights of the New Agreement-1982-1984 (Detroit, Mich.: UAW Agricultural Implement De-partment, April 1982).

39

The Ford agreement called for the establishmentof a National Development and Training Center,where staff on loan from the union and the com-pany will promote training, retraining, and otherskills development opportunities for current anddisplaced workers. * Two projects were launchedby the Center in August 1982: a National Voca-tional Retraining Assistance Plan, which providesprepaid financial assistance of up to $1,000 peryear to workers on layoff who wish to undertakeself-directed, formal education or retraining; andTargeted Vocational Retraining Projects, highlyspecialized retraining activities designed todevelop skills for use in new or existing occupa-tions in which there are documented worker short-ages. The Vocational Retraining Projects wouldbe limited to geographic areas where establishededucational institutions and vocational trainingprograms are not already providing such instruc-tional opportunities. The Center also hopes tostimulate similar, publicly funded efforts in areasof the country where Ford workers are on layoffand might be eligible to participate.24

IAM initiated in the 1950’s an annual electronicsindustry conference, known since 1968 as the Elec-tronics and New Technology Conference, duringwhich national staff and representatives of IAMunion locals discuss issues that arise from the useof manufacturing technologies. In 1960, IAMbegan the practice of preparing a manual of modelcontract language that included provisions for usein dealing with in-plant technological change.1AM model contract language on training benefitscalls for instruction during working hours at com-pany expense and at prevailing wage rates. It alsostates that senior employees should have firstclaim on training opportunities and suggests thatmanagement should be required to train employ-ees for jobs not necessarily associated with newtechnology, in cases where “ . . . either the newtechnology requires substantially fewer workersor present employees are not capable of successful

● The Center, temporarily located at Ford World Headquartersin Dearborn, Mich., will move to its permanent headquarters atHenry Ford Community College (Dearborn) in 1983.

“’’National Vocational Retraining Assistance Plan and Other Proj-ects for Certain Employees on Layoff, ” UAW-%d Emp)oyee Development and Training Program Bulletin, Aug. 20, 1982.

retraining. “25 At an IAM-sponsored Scientists andEngineers Conference, held in the union’s PlacidHarbor Training Center in June 1982, membersexpressed concern that training and retraining pro-visions in contracts address instructional pro-cedures as well as content.

Retraining the Displaced

A comprehensive review of documentation rep-resenting over 20 years of plant-closing experiencerevealed that retraining programs are of greaterbenefit to displaced workers who are younger,have slightly more formal education, and haveachieved some level of financial security. Evenamong displaced individuals who possessed thesecharacteristics, only about 15 percent participatedin retraining, due to inadequate financial assist-ance during the training period.26 These findingssuggest that some new approaches to retrainingthe displaced should be developed that increasethe utility of instruction and its availability toworkers of all ages, with varying amounts of for-mal education and different degrees of financialsecurity.

Although the public perception is that industryis one of the chief sources of sponsorship for re-training of displaced workers, in the past it hassponsored few retraining efforts. In some cases,the communities surrounding plants lacked alter-native career opportunities for which instructioncould be provided; in others, workers expectedto be called back to their old jobs and resisted tak-ing advantage of instructional and placement op-portunities; in still other cases, economic condi-tions that led a company to close a plant madethe cost of retraining prohibitive. Althoughretraining activities authorized under the Com-prehensive Employment and Training Act (CETA)and the Trade Readjustment Assistance Act (TRA)

ZSL=lie E. Nu]ty, “Ca~ Studies of IAM Local Experiences Withthe Introduction of New Technologies, ” in Labor and Technology:Union Responses to Changing Environments (University Park, Pa.:Department of Labor Studies, Pennsylvania State University, 1982),pp. 115-139,

26J eame p. Gordus, paul Jarley, and Louis A. Ferman, r~n~ C]os-ings and Economic Dislocation (Kalamazoo, Mich.: W. E. UpjohnInstitute for Employment Research, 1981).

40

have been criticized, these Federal programs repre-sent the majority of resources that have been uti-lized to prepare displaced workers for new ca-reers. *

Two recent examples of retraining effortsfunded under CETA illustrate the potential forretraining some of the displaced for new, tech-nology-related occupations. The first, a pilot proj-ect made possible through a $300,000 Departmentof Labor discretionary grant to UAW, is designedto retrain 400 displaced auto workers for occupa-tions in demand within the aerospace/defense in-dustry. The first phase of the project, an assess-ment of the potential for skills transfer from jobsperformed within the auto industry to the newpositions within aerospace/defense, has alreadybeen completed. Other products of the grant in-clude two retraining programs, which will be de-veloped by combining components of existing re-training packages. 27 Although the project does not

train individuals solely for technology-relatedpositions, a UAW spokesperson indicated thatmany of the new aerospace jobs involve workingwith automated equipment and therefore relatedskills requirements will be addressed in the retrain-ing packages to be developed. Implementation ofthe training process now awaits Federal fundingor sponsorship by the aerospace/defense industry.A second CETA-funded project, initiated by theWarren County, Mich., prime sponsor, is a 40-week robotic technician program, which qualifies18 displaced auto workers, machinists, and otherswho completed the course to assume new careerswithin the auto industry, in local robotics firms,or in other companies using robots.**

*The recently enacted Job Training Partnership Act, which re-places CETA, authorizes the expenditure of Federal funds for em-ployment and training of dislocated workers. CETA will operateduring fiscal year 1983 at a $2.8 billion funding level, while pro-grams authorized under the Training Partnership Act are established(tiployment and Training Reporter, Nov. 19, 1982).

“’’Assessment of the Reemployment Opportunities for Unem-ployed Automobile Workers in the Aerospace/Defense Industry,”proposal prepared for the U.S. Department of Labor by the UnitedAuto Workers, 1982.

● *As stipulated in Public Law 95-524, CETA amendments of 1978,a prime sponsor may be” . . . a State; a unit of general local gover-nment which has a population of 100,000 or more . . . ; a consor-tium of units of general local government . . . ; . . . program grant-ees serving rural areas having a high level of unemployment . . . ;and any unit of general local government previously designated asa prime sponsor under the provisions of this Act . . . , regardlessof population decline. ”

Instruction for Technician= LevelOccupations

Although technicians emerged as an occupa-tional group within the field of engineering in the1920’s, the availability and application of tech-nology in manufacturing has increased the de-mand for and the popularity of this occupation.Technicians who are trained in the use of com-puter-aided drafting systems are now in great de-mand within aerospace and other industries.28

Technician instruction is typically a 2-yearassociate degree program, although other, moreconcentrated approaches to program delivery arebecoming more common, such as the one initiatedin Warren County, Mich. The electromechanicaltechnician curriculum, which combines two for-merly distinct engineering specialties, is viewedby some educators and industry representativesas an excellent foundation for careers that requireknowledge of programmable automation.

Community colleges in various areas of thecountry are currently offering electromechanicaltechnician programs, sometimes called roboticstechnician programs by the institutions in orderto capitalize on general public awareness of thisform of programmable automation. The State ofGeorgia began offering an electromechanical cur-riculum in its community colleges in 1982. Severalcommunity colleges in Michigan have offered elec-tromechanical programs for the past few years.In general, curricula are designed to prepareenrollees to perform installation, maintenance,repair, and programing functions. At present,however, no standardized performance criteriaexist for electromechanical technicians, so the con-tent and emphasis of these programs vary con-siderably.

Engineering Education

The utilization of programmable automationhas had an observable effect on initial and con-tinuing education for engineers. CAD, which en-ables faster design and analysis, is now commonin the aerospace and auto industries. Selectedengineering schools are working with industry to

“’’The Engineering of a Revolution: Computer Now Designer’sTool,” The Atlanta Constitution, Oct. 24, 1982.

41

add CAD instruction to their curricula. BoeingCommercial Airplane Co. in Seattle, Wash., hasestablished at the request of local universities (e.g.,Washington State) CAD laboratories adjacent toengineering school campuses. These labs providestudents with opportunities to work with Boeingaircraft data bases when they are not being utilizedby Boeing personnel. The program is voluntary,but students receive university credit for par-ticipating.

On the national level, a 1981 grant from theNational Science Foundation’s Directorate for

Science and Engineering Education (now theOffice for Science and Engineering Education)established the College CAD/CAM Consortiumas a nonprofit group dedicated to the developmentof CAD/CAM curriculum and the improvementof CAD/CAM instruction. Twelve engineeringschools, including Carnegie-Mellon Universityand Rensselaer Polytechnic Institute, were found-ing members.29

29?7w College CAD/CM Consortium: An Overview (Cluwlottes-ville, Va.: University of Virginia, School of Engineering and Ap-plied Science, January 1981).

SURVEY OF CURRENT VIEWS OF EDUCATION, TRAINING,AND RETRAINING REQUIREMENTS

There has been little or no information gatheredon how representatives of the key groups involvedin or affected by the manufacturing automationprocess—producers of the equipment and sys-tems; users of the equipment and systems; andvarious groups in the work force—view the poten-tial retooling of the operations with which theyare associated. In addition, no national readingshave been taken of current views held by thesegroups on education, training, and retraining re-quirements associated with the use of program-mable automation. In order to supplement avail-able information of this type, and in so doing geta better sense of the climate in which automationis occurring, OTA commissioned structured tele-phone interviews with a sample of representativesof firms within the electric and electronics equip-ment, industrial machinery, and transportationequipment industries (industries in which firmsare especially likely to use programmable automa-tion). OTA also contacted producers of program-mable automation equipment and systems, as wellas educators and others familiar with the instruc-tional design process. A total of 506 interviewswere completed in July and August 1982. * In thissection of the technical memorandum, a summaryof selected survey findings is presented. A descrip-

*The term users refers to firms applying programmable automa-tion; the term producers refers to firms producing programmableautomation; and the term others refers to educators and others in-volved with education and training.

tion of the survey methodology and sample sizeare included in appendix A.

Education and Training:Users and Producers

The survey found that 40 percent of the repre-sentative manufacturing plants contacted utilizedsome form of programmable automation, and ofthis number, only 22 percent sponsored or con-ducted education and training for new technol-ogy. Among plants currently not offering educa-tion and training of this type, only 18 percent in-dicated any plans to implement programs in thefuture. Low benefits relative to costs was by farmost commonly cited by user firm representativesas a barrier to the establishment of instructionalprograms for new technology. The low levels ofcurrent and anticipated direct involvement in ed-ucation and training for new technology is par-ticularly notable in light of the nearly unanimousview expressed by users, producers, and othersthat the users should bear the costs for new tech-nology instruction. This seems to indicate thatwhile users may be willing to pay for instructiondelivered by vendors, educational institutions,and others, few are planning to establish theirown, in-house programs. Another possible inter-pretation of the low levels of in-house instructionamong users might be that changes brought aboutthrough the utilization of programmable automa-

98-951 0 - 83 - 7

42

tion thus far have not been sufficient to warrantthe establishment of formal instructional systems.

Instruction Available Through Producers

In contrast to the low proportion of users whosponsored and conducted education and trainingfor new technology, a very high proportion ofproducers (93 percent) provide such instructionfor their customers. Manufacturers within the in-dustry groups polled appeared to depend on pro-ducers for design and delivery of new technol-ogy-related instruction. Results indicate that ven-dors or producers of programmable automationequipment were more heavily used for instructionthan were training industry/management consult-ants, traditional educational institutions, proprie-tary schools, unions, and government-sponsoredinstructional programs such as CETA.

The nature and scope of instruction currentlyoffered by producers, however, seems to be quitelimited. Over 80 percent provide only singlecourses, and few provide series of courses. Fur-thermore, only about one-third of the producersfelt that vendors or manufacturers of computer-automated equipment and systems were currentlyready to provide the necessary education andtraining. One can speculate that the producerswho work closely with new technology under-stand the education and training implications ofimplementing their technology, but are currentlyonly providing part of what they consider is re-quired. Producers may be providing limited serv-ices for a variety of other reasons, including costfactors, customer demand, and their views of theresponsibilities of other institutions (particularlyusers) in providing additional training.

Occupational Coverage andContent Coverage

Both users and producers reported generallybroad occupational coverage in the instructionalprograms on new technology that they provided,although there was considerable variation in theextent to which occupations were covered. Themajority of both users and producers sponsoredor conducted programs for various types of shop-floor staff (e.g., assemblers, handlers, loaders,equipment operators), repair and maintenance

staff, engineers, programmers, and supervisors ormanagers. Apparently, the impact of program-mable automation on a wide array of occupationsis recognized by industry.

Broad occupational coverage was not accom-panied by breadth in instructional content. Theprimary content of current education and train-ing programs appears to reflect traditional topicsaddressed in technical training; e.g., machineoperation, safety procedures, and maintenance.Current instructional programs focus least on thebasic skills—reading, writing, and arithmetic—and basic physical science. The survey results sug-gest that manufacturers assume that these needsshould be met in ways other than in instructionalprograms they devise.

Government Role in NewTechnology Instruction

Survey results show a lack of receptivity togovernment involvement in instruction for newtechnology by both users and producers. As notedearlier, government-sponsored instructional sys-tems such as CETA were generally considered notready to provide such training and were not ex-pected to become ready within 10 years. Whenasked about possible sources of funding for educa-tion and training for new technology, only abouthalf of the respondents in both groups indicatedthat Federal or State and local government fund-ing was desirable, while funding from all othersources, particularly private sector user industriesand foundations, was endorsed by at least threequarters of the respondents. In contrast, a greatmajority of the others (the group that includededucators and Federal and State officials) endorsedgovernment as a funding source. *

● It is not clear whether these responses reflect popular politicalviews, attitudes toward government intervention in general, or actualpreferences for private control of instruction for high technology.In any case, it is unlikely that respondents had in mind all formsof Federal, State, or local support (e.g., funding of colleges, uni-versities, and research efforts), although it is not possible to deter-mine this from the present data. Nevertheless, the consistency ofresponses in the user and producer groups may provide some guid-ance for determining the nature of the government role in instruc-tion for computer-automated manufacturing. It is likely that indirector less visible forms of government intervention would be more ac-ceptable to industry than more direct forms of intervention, suchas the provision of education and training services (e.g., CETA pro-grams) or direct subsidies to industry for worker retraining.

43

SELECTED CRITICAL ISSUES FOR INSTRUCTIONAL PROGRAMINGCurrent views of representatives from industry,

labor, the educational community and govern-ment are consistent with other indicators discussedearlier in this technical memorandum in suggestingthat training and retraining requirements for pro-grammable automation are, at this point, poorlydefined. Even within specific geographic areas,programs initiated to address changing instruc-tional requirements do not in the aggregate repre-sent a coordinated approach to defining instruc-tional needs associated with new industrial proc-esses. While it is too soon to know how wide-spread applications of programmable automationwill be, there is little evidence that any sector—including private industry-is seriously consider-ing the long-range implications of possible wide-

spread use on occupational skills requirements andcurrent instructional capacities.

There are a number of pressing issues facingthose who operate instructional systems, in theevent that widespread utilization of program-mable automation occurs. Among them are:

1.

2.

3.

4.

how and by whom the need for technologi-cal literacy will be addressed,types of short-range and long-term counsel-ing and instructional systems,initiation of appropriate curriculum designprocesses, andfunding sources for curriculum design andimplementation.

Appendixes

Appendix A

Survey Methodology

Introduction and Overview

In support of the Automation and the Workplacestudy undertaken by the congressional Office of Tech-nology Assessment, Westat conducted a survey toidentify education and training requirements inherentin the use of programmable automation in manufactur-ing settings. The survey describes current levels of uti-lization of programmable automation, as well as exist-ing instructional opportunities focused on this formof technology, and elicits various opinions related tocurrent and anticipated education and training needsresulting from applications of computer-based automa-tion.

Survey data were collected by the Westat TelephoneResearch Center on three samples: 1) users, 2) pro-ducers of computer-automated technologies in manu-facturing, and 3) a diverse sample of knowledgeableothers. The surveys used structured instruments devel-oped for the study. Data were collected by telephoneinterviews over a 2-week period in August 1982.

Methodology

This section briefly describes the methodology usedfor the survey. The first part describes sampling pro-cedures, the second describes data collection instru-ments and methodology, and the third describes dataanalysis procedures.

Sample

Three groups were contacted for this study: usersand producers of computer-automated equipment andsystems, and others, a diverse group of individuals in-volved in instruction for employees in computer-auto-mated manufacturing environments or for individualspreparing for careers in such settings. Formal samplingprocedures were used only for the user group, and fora subset of the others. Details of the procedures aredescribed below.

USERS

Sampling Frame. —The sample of users was com-posed of manufacturing establishments from industriesidentified as currently using or likely to use computer-programmable equipment and systems in the near fu-ture. An establishment was defined as an individuallocation of company. This location might constitute

a division, subsidiary, plant, branch, or the entirecompany.

Three major manufacturing industries were repre-sented: transportation equipment manufacturing, elec-tric and electronic equipment manufacturing, and in-dustrial and metalworking machinery manufacturing.For each of these three major industries, specific stand-ard industrial classifications (SIC) were selected basedon two criteria: 1) proportion of total employees inindustry accounted for by establishments within theSIC code; and 2) likelihood of establishments withinthe SIC code using computer-automated technology.SIC codes meeting the second criteria were selectedbased on judgments of project staff, as well as OTAAutomation Study Advisory Panel members. The se-lected SIC codes for the transportation equipment in-dustry account for 76 percent of the total workersemployed in the industry; SIC codes in electric andelectronic equipment manufacturing account for 59percent of the total employees; and the SIC codes inindustrial and metalworking machinery account for41 percent.

The data source for constructing the frames for thethree user samples was National Business Lists (NBL),a firm which compiles a national list of most types ofestablishments, including manufacturing and commer-cial. The NBL lists rely heavily on the Dun & Brad-street directory of establishments, supplemented byNBL’s own sources.

Sampling Methodology.—A probability sample ofusers in the three industries was selected from the NBLlists using a two-stage sampling approach. This sam-pling procedure involved stratification by size and re-gional location, and included as selections with certain-ty a small number of establishments known to usecomputer-automated equipment for manufacturing.These were included to assure a minimum of currentusers within the sample to provide an adequate basisfor analysis of this subgroup.

The first step in a two-stage sampling procedure en-tailed compiling a list of approximately 5,000 estab-lishments from the NBL master file in the three majorindustry groups specified earlier. The purpose of “over-sampling” establishments at this initial stage was toobtain a sufficiently large sample for examining thesize distribution of establishment by SIC group forsubsequent use in deriving appropriate (and more near-ly optimal) sampling rates. Since larger establishmentsaccount for a larger share of the work force while ac-

47

48

counting for a smaller share of the total number ofestablishments, selection of the initial sample fromNBL was stratified by establishment size. The sizestrata used by the initial sampling were:

Small: 1-99 employeesMedium: 100-499 employeesLarge: 500 or more employees

Furthermore, to take these size differences into accountin the sampling, large establishments were sampled ata higher rate than small ones. Therefore, the initialsample consisted of all of the large establishments inNBL for each of the three industries, one-half of themedium-sized establishments and one-tenth of thesmall establishments. Since the listings of establish-ments in the NBL file were geographically sorted with-in each of the three size classes, and the samples wereselected systematically (using a random start), themethod of sample selection simplicity included strat-ification by geographic region. These proportionsyielded 5,128 total establishments in the initial sample.

In the second stage of the sampling procedure, the5,128 establishments were further stratified by industrytype, establishment size, and regional location. Regionallocation was defined by the four regions delineated bythe U.S. Census Bureau (i.e., Northeast, North Cen-tral, West, and South). In addition to the three majorsize strata defined above, the “small” size class wasfurther subdivided into two classes for sampling (1-20and 20-99). This more detailed stratification by sizepermitted a more nearly optional allocation of the sam-ple cases to the various strata. This stratificationyielded 48 different cells in which establishments wereplaced for sampling.

To determine the appropriate sampling rates to se-lect the second-stage sample, three options were con-sidered:

1. The sample could be allocated to each cell in pro-portion to the total number of establishments inthat cell.

2. The samples could be allocated to each cell in pro-portion to the total employment in that cell.

3. The samples could be allocated to each cell in pro-portion to some function of employment, saysquare root of employment.

The implication of the first option was to have largenumbers of small establishments and few large estab-lishments since most manufacturing establishmentshave fewer than 500 employees. This would be desir-able for estimation of counts of establishments, butwould not be sufficient for estimation of magnitudevariables such as employment.

The implication of the second option was to havelarge numbers of large establishments and very fewsmall establishments, since manufacturing establish-

ments of 500 or more employees account for most ofthe work force. This would be approximately optimumfor estimation of magnitude variables (in particular,those correlated with employment), but would be lessefficient for estimates of the numbers of establish-ments.

The final option, which combines the first and sec-ond options by sampling with probabilities propor-tionate to the square root of employment, distributesthe numbers of establishments somewhat more evenlyacross cells. This last option was selected because itprovided a better basis for making comparisons be-tween the different size classes, in addition to beingreasonably efficient for estimating both magnitude andcount variables.

Sampling Methodology .—Establishments from theuser sampling frame were screened to eliminate fromthe user samples those not meeting the following cri-teria:

1. establishments must be performing manufactur-ing functions at the location contacted (purely ad-ministrative facilities were dropped); and

2. establishments must be able to identify an indi-vidual within the first three referrals during thephone call who can answer selected key ques-tions. (Those unable to do so were treated as non-responses. )

Selection of the second-stage sample, therefore, wasbased on the assumption that there would be extensivedropouts due to ineligibility and nonresponse. Fromthe initial sample of 5,000 user establishments, 200establishments were drawn from each user industrygroup for a total of 600 sample establishments. Thesewere allocated to the various size strata in proportionto the aggregate measure of size based on the squareroot of employment. This sample included 18 estab-lishments that were selected with certainty in additionto the 600 selected establishments. The sample alloca-tion of the noncertainties in each user industry groupby size class is shown in table A-1.

Detailed records were kept of establishments failingto meet these criteria as well as refusals and nonre-sponses. Table A-2 shows the distribution of the ini-tially sampled cases and the final number of completedinterviews by size strata and region.

PRODUCERS

Sampling Methodology. —The producer group wascomposed of companies who manufacture and/or sellprogrammable equipment to U.S. manufacturing in-dustries. The compilation of a list of producer com-panies was no simple task, since no such lists werereadily available. An intensive search to identify com-

49

Table A-l. -Sample Allocation for User Groups

Number of MOS a based on squareSize class (employment) establishments in NBL root of employment Sample allocation

Transportation equipment1-19 . . . . . . . . . . . . . . . . . 2,070 5,430 3020-99 . . . . . . . . . . . . . . . . 1,030 7,050 40100-499 . . . . . . . . . . . . . . 596 9,020 50500+ . . . . . . . . . . . . . . . . 287 14,969 80

Total. . . . . . . . . . . . . . . 3,983 36,469 200

Electrical and electronic machinery1-19 . . . . . . . . . . . . . . . . . 4,520 11,440 4020-99 . . . . . . . . . . . . . . . . 2,090 14,510 50100-499 . . . . . . . . . . . . . . 1,244 18,742 60500+ . . . . . . . . . . . . . . . . 363 14,240 50

Total. . . . . . . . . . . . . . . 8,217 58,932 200

Machinery manufacturers1-19 . . . . . . . . . . . . . . . . . 12,620 30,900 6020-99 . . . . . . . . . . . . . . . . 3,590 22,450 50100-499 . . . . . . . . . . . . . . 1,296 19,092 50500+ . . . . . . . . . . . . . . . . 318 12,005 40

Total . . . . . . . . . . . . . . . 17,824 84,447 200

aMea~ure@f.~ize. F~~a~ivensizeclagg, theaggregate MOSwascomputed a9S ‘} Ei, where Eiistheaverage employment

slzeof all establishments inthe SICmouP and size class based on 1979 CO@YBUS~rW-3SS Patterns, and where the summa-tion extends over all establishments-in the NBL frame.

SOURCE:Westat.

Table A-2.—Stratification of User Establishmentsand Costs of Initially Sampled Cases and

Completed Interviewsa

Northeast North Central South West

Transportation region1-19 . . . . . . . . . . . 4 : 220-99 . . . . . . . . . . 8 : 4100-499 . . . . . . . . 8 : 3500 or more. . . . 12 : 6

Electrical/electronic region1-19 . . . . . . . . . . . 12 : 320-99 . . . . . . . . . . 18 : 9100-499 . . . . . . . . 22 : 11500 or more. . . . 16 : 9

Machinery region1-19 . . . . . . . . . . . 12 : 520-99 . . . . . . . . . . 10:5100-499 . . . . . . . . 10:55 0 0 0 r m o r e . . . . 8 : 7

8 : 114 : 826 : 1642 : 24

8 : 310: 414 : 712 : 8

28 : 1328: 926 : 1020: 12

6 : 1 1 2 : 46 : 3 1 2 : 48:8 8:2

16:11 10:6

8 : 4 1 2 : 48 : 4 1 4 : 8

1 2 : 4 1 2 : 71 0 : 6 1 2 : 6

1 0 : 4 1 0 : 31 0 : 8 6 : 21 0 : 3 4 : 18:6 4:1

aThe number on the left in each cell shows the inltlai Sample count, and thenumber on the right shows the number of completed interviews. The numberof completed interviews shown in this table does not include the nine completedusers which were sampled with cerlainty, since these were not preassigned toa size class.

SOURCE: Westat.

panics who were involved in manufacturing or sellingcomputer-automated equipment was conducted byOTA, with assistance from Westat & Hadron, a sub-contractor. The list used in this study was constructedfrom a variety of sources, including rosters of exhibi-tors at conventions on computer-automated manufac-

turing, lists from organizations such as the RoboticsInstitute of America, trade publications citing com-panies involved with such products, and personal con-tacts with relevant companies. The final list consistedof 203 producers, and is considered to be a fairly goodapproximation of the actual universe of companiesproducing computer-automated equipment for man-ufacturing in the United States.

Producers were contacted in random order until 101companies had completed interviews.

OTHERS

Sampling Frame.—The others group was composedof individuals who have had experience in designingand/or delivering and/or evaluating formal instruc-tion for employees operating in computer-automatedor conventional manufacturing environments. Theseothers were selected because of their pertinent exper-tise and/or because they represented institutions (e.g.,unions) whose opinions are important to consider informulating policy in this area. A list of 280 others wascompiled by OTA. The list was composed of six sub-groups:

●

●

Traditional educational institutions (e.g., collegesand universities, community colleges, technicalschools);proprietary educational institutions (private, prof-itseeking, trade and technical schools that operateon the secondary and postsecondary level);

98-951 0 - 83 - 8

——

50

● labor unions and labor organizations;● training industry representatives (individual con-

sultants and training firm representatives);● State and local agency representatives (e.g., voca-

tional education and economic development agen-cies); and

. miscellaneous others (e.g., Federal Governmentand trade association representatives, individualscholars, and experts).

Sampling Methodology. —Representatives of tradi-tional educational institutions were randomly sampled,while attempts were made to contact all those in theother five subgroups. The initial goal of obtaining 25interviews from the traditional education subgroupand 75 from the remaining subgroups had to be re-vised, due to nonresponse rates among the five othersubgroups. Actual portions are presented below.

Final Samples and Response Rates

A total of 506 interviews were completed for thestudy. There were 303 users (105 in transportationequipment, 98 in electric and electronic equipment,and 100 in industrial and metalworking machinery),101 producers, and 102 others. In the others sample,there were 34 traditional educators, 11 educators fromproprietary educational institutions, 13 union repre-sentatives, 2 representatives of the training industry,17 representatives of State and local agencies, and 25“others.”

The response rates obtained (defined as the numberof completed interviews plus refusals) were 82 percentoverall, 76 percent for the users, 89 percent for the pro-ducers, and 95 percent for the others. The completion

rates (defined as the number of completed interviewsdivided by all completes plus all incomplete) weresomewhat lower, due to unknowledgeable, unavail-able, or nonlocatable respondents. Table A-3 sum-marizes the final completion status of the telephonesurveys conducted with further explanations of variouscompletion statuses in table A-4.

Data Collection Instruments and Methodology

SURVEY INSTRUMENTS

Three telephone survey instruments (for users, pro-ducers, and others) were developed for the study. Theinstruments were closed-ended in format—i.e., re-sponse options were provided for most of the ques-tions. A core set of questions was asked three groups,along with additional questions designed specificallyfor each group. The instruments were designed to re-quire approximately 15 to 20 minutes per interview.

In general, the instruments were designed to obtaininformation about the extent and nature of the involve-ment of the respondents with programmable automa-tion technology, their involvement with education andtraining focused on the application of various formsof programmable automation in manufacturing set-tings, and their opinions about a variety of issues re-lated to such instruction. In addition, questions onbasic background characteristics (e.g., size of the workforce) were also included in the instruments.

Table A-5 presents the major topics covered by thethree survey instruments. The greatest number of ques-tions were asked of the users, although most topicswere covered in the three instruments. One major dif-

Table A-3.—Final Response and Completion, Statuses of Telephone Surveys

Users

Status codes Group 1 Group 2 Group 3 All users Producers Others Total survey sample

Complete . . . . . . . . . . . . . . . . . . . . . . .Admin Hdqtrs. . . . . . . . . . . . . . . . . . . .Not working . . . . . . . . . . . . . . . . . . . . .No answer . . . . . . . . . . . . . . . . . . . . . .No new tech. . . . . . . . . . . . . . . . . . . . .No E&T . . . . . . . . . . . . . . . . . . . . . . . . .Duplicate . . . . . . . . . . . . . . . . . . . . . . .Final refusal . . . . . . . . . . . . . . . . . . . . .Not available . . . . . . . . . . . . . . . . . . . .Not locatable . . . . . . . . . . . . . . . . . . . .E&T knowledge . . . . . . . . . . . . . . . . . .Response ratesa . . . . . . . . . . . . . . . . .Completion ratesb . . . . . . . . . . . . . . . .

1053211

1——

13111——77 ”/071 “/0

9821

911——

2281321

1780/o61 0/0

100 30322 7511 3112 24— —

—74”!0620/o

—3

943735

176%650/o

1011

11—39—

71223

9—890/o70”!0

102————

375

145

—95%81 0/0

50676422439

317

1117449

1820/o68 ”/0

aResponse rate =

No. completes x 100(No. completes) + (No. final refusals)

bCompletion rate = No. completes x 100

(No. completes) + (No. noncompletes)SOURCE: Westat.

51

Table A-4.—Definitions of Completion, Ineligible,and Nonresponse Status Codes

/. CompletionA. Complete(C)-completed entire interview. A complete

means all pertinent questions have been answered.

//. Inteligibility (The following categories of respondents werescreened out of the survey.)A.

B.

c.

D.

E.

F.

Admin hdqtrs (l) —user establishment is an ad-ministrative headquarters which does not perform amanufacturing function.Not working (NW) —phone number is not in serviceand, after calling directory assistance, there is no newlisting for that facility.No answer (NA)—there is no answer after three at-tempts at different times on different days.No new tech (N L) —producers only; if producerestablishment is not manufacturing or selling newtechnology included in the survey.No E&T (S2)—others only; if respondent representsa traditional educational institution, proprietary educa-tional institution, or a training firm, which does nothave an education and training program.Duplicate (OA)—duplicate respondent.

///. Noncompletion (The following categories of respondentsare included in computation of a completion rate.)A.

B.

c.

Final refusal (R B) —respondent refuses the interviewor breaks off interview.Not avai/ab/e (0)—respondent was not available dur-ing field period.Not locatable (Sl)—appropriate respondent was notlocated after three referrals or the respondent was notknowledgeable about new technology for his/herestablishment.

SOURCE: Westat.

ference should be noted in the questions on educationand training in the three instruments. Users were askedabout the education and training provided to their ownemployees. In contrast, producers and others wereasked about the instruction they provide to customersor clientele, not their own staff.

The instruments were developed in several stages byWestat & Hadron staff, in collaboration with OTA.At the start of the reseach effort, 12 in-depth interviewswere conducted, in person and on the telephone byWestat & Hadron staff, to identify important issuesand develop possible questions and response optionsfor the instrument, A topic guide was developed forthese interviews.

Based on these preliminary interviews, a list of draftquestions and response options was submitted toOTA. Comments from OTA staff and further inter-views by Westat & Hadron staff were used to refineand shorten the questionnaires.

The instruments were pretested by the Westat Tele-phone Center on a small number of respondents, andfurther minor changes resulted from the pretests.

Data Analysis

SAMPLE WEIGHTS

Since disproportionate sampling procedures wereemployed in drawing the user samples—i.e., differentsampling ratios were used for the different strata—

Table A-5.—Topics in Survey Instruments

User Producer Other

Background. Year founded . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . X● Gross sales . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . X● Work force or clientele characteristics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . XComputer automation● Use, production, or sale of new technologies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . X● Extent of computerization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . X● Computerized integration of equipment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . XEducation and training (E&T)• Presence of general E&T, and E&T for new technology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . X● Priority given to setting up E&T . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . X● Barriers to setting up new technology instruction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . X. Work force/clientele percentage who received or will need E&T . . . . . . . . . . . . . . . . . . . . . . . . . . X• Number of instructors for new technology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . X● Forms of instruction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . X● Sources for designing/delivering instruction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . X● Target occupational groups . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . X• Skill and knowledge areas covered. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . X• Policies and opportunities on E&T outside the company. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . XOpinions● Current and future readiness of institutions to provide instruction . . . . . . . . . . . . . . . . . . . . . . . . X● Options for institutional collaboration on E&T . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . X● Sources of funding for E&T . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . X

xxx

x

xxx

xx

xx

xxx

xxx

xx

SOURCE: Westat.

52

mechanisms to equalize the differential probabilitiesof selection attached to establishments from the dif-ferent strata were required. Such weighting adjust-ments are necessary in cases where generalizations aremade from the sample to a larger sampling frame oruniverse, and to take into account nonresponse.

Weights were applied only to the user sample andthe others. Weights were not necessary for the produc-ers, since they did not constitute a sample from a largeruniverse of such firms. Users were given weights suchthat the sample represented an estimated 24,142 ac-tive and eligible establishments in the NBL frame—and the other sample was weighted to represent the280 others in the original sampling list. The estimated24,142 active and eligible user establishments was ob-tained by summing up the weights of the respondingestablishments in the sample and compares with about30,000 establishments in the designated SIC groups inthe NBL frame.

The sampling weights for the OTA samples (userand other groups) were computed from the formula:where

Whi = The weight for establishment i instratum h (for a particular group).

nh = The total number of establishments (inthe frame) in stratum h.

Nh = The number of establishments in stratumh that were finally sampled.

n ‘h = The number of eligible and respondingestablishments in stratum h.

n “h = The number of eligible but nonrespondingestablishments in stratum h.

The factor, (n ‘h + n“h)/n ‘h, in the above expressionrepresents an upward adjustment for total question-naire nonresponse. The weight for any given establish-ment depends on the stratum (and group) from whichthe establishment was sampled, but is uniform for allresponding establishments in a particular stratum.Weights for the establishments selected with certain-ty would be 1.0 if there were no nonresponding cases,and otherwise exceed 1.0 by a factor representing anadjustment for nonresponse.

Table A-6.—Sampling Weights for Estimation

Sampling stratum

Respondent group Certainty < 20 employees 20-99 employees 100-499 employees 500+ employees

User 1 (transportation) . . . . . . . . . . . . . . 2.5 155.2 42.8 18.4 4.8User 2 (electric and electronic). . . . . . . 4.0 248.4 74.5 36.1 10,0User 3 (industrial and metalworking) . . 1.0 419.0 136.6 52.2 10.3Others . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.2 Noncertainties a: 4.7aA~ lndlCat~ In ‘(oth~~~)r ~eCt\On In text, all trgdltlonal educato~ except for educational Institutions were Included in the 9~Ple with certaintY. Educational ‘institutionswere sampled at a fixed rate of about 1 prior to adjustments for nonresponse.

SOURCE: Westat.

Appendix B

Industrial Relations

Overview

The activities, institutions, and circumstances of in-dustrial, or labor-management, relations influence theimplementation of new technology and its conse-quences within firms and industries. In particular, theycontribute to employment patterns and workplaceconditions that might not arise with technology changeand market forces alone. Therefore, an understandingof industrial relations is necessary for understandingnot only how programmable automation may affectcompany and industry employment and wage levels;but also how job content, promotion paths, and work-place conditions may change with programmable auto-mation; and why employees and management in dif-ferent companies and industries may have different ex-periences with technological change.

Despite the important role of industrial relations inthe U.S. economy, the analysis of industrial relationstends to be relatively imprecise and experiential. Asone participant in the OTA Labor Markets and Indus-trial Relations Workshop put it, there seem to be more“ad hoc-cries” than true theories for explaining indus-trial relations phenomena. Further complicating anevaluation of industrial relations issues are the differ-ences in approach taken by different analysts. For ex-ample, most labor economists and so-called industrialrelationists tend to regard workers and managers ashaving opposing interests, with workers striving tominimize work effort and maximize compensation,and managers striving to minimize cost and maximizeproduction. Most organizational behaviorists and or-ganizational development specialists tend, by contrast,to regard workers and managers as sharing basicallysimilar interests that stem from their association withthe same organizations. The former group tends tofocus on the setting of wages and other “economic”issues, while the latter group tends to focus on jobsatisfaction and performance, supervisory relation-ships, and job design. *

A final, but critical, factor complicating attemptsat precise analysis of industrial relations issues is thefact that rhetoric that tends to exaggerate conflict be-tween labor and management can obscure the actualcircumstances of industrial relations, particularly inunionized settings. According to some observers, rhe-

IPeter Feuille and Hoyt N. Wheeler, ‘Will the Real Industrial Conflict pleaseStand Up?” in U.S. industrial Relations 19s@19&.): A Critical Assessment,Jack Steiber, et al. (eds. ) (Madison, Wis.: Industrial Relations Research Asso-ciation 1981).

torical hostility between organized labor and manage-ment has been especially high during the last few years:

. . . (W)e are witnessing a continuation of this recenthigh level of rhetorical hostility between labor and man-agement compared to the situation that prevailed dur-ing most of the 1950-80 period. In addition, . . . theone-sidedness of our (and the traditional) definition ofconflict as worker action shows a tendency to obfuscatethe reality of conflict between managers and workers,for it leads us to reject aggressive action by manage-merit. z

This rhetoric, amplified by the news media in the con-text of deteriorating economic conditions, may biaspublic opinion against organized labor, despite the lackof objective analysis of the contributions of both laborand management activities to current economic condi-tions.

The popular, and even the research, view of indus-trial relations tends to focus on unionized settings,since unions (and employee associations that functionsimilarly) serve to focus and articulate the concernsof workers both at the workplace and in the communi-ty, although only a portion of U.S. companies andworkers are unionized. The union-nonunion distinc-tion is misleading, however, because labor-manage-ment relations fall into a spectrum that includes inter-mediate arrangements containing greater and lessernumbers of pure union-like and nonunion-like attri-butes. The principal difference between the union andthe nonunion setting is that in the nonunion setting,management typically imposes job descriptions, wagelevels, working conditions, and technological changeunilaterally, while in the union setting, many of theterms of the workplace are jointly set by labor andmanagement through a negotiation process. Therefore,the role or conduct of labor is as important as that ofmanagement in the unionized setting.

Unions are of particular, but not exclusive, interestto a study of the impacts of programmable automa-tion because workers in many of the occupations andindustries where programmable automation is ex-pected to have the greatest impacts are especially likelyto belong to unions. Unions whose members will beexposed to programmable automation include thoserepresenting workers in metalworking manufacturingindustries, such as the United Auto Workers, the In-ternational Association of Machinists (IAM), the In-ternational Brotherhood of Electrical Workers, andothers that are listed in a paper by W. Cooke, appen-

21bid.

53

54

dix C. Although the median size for national unionsis around 25,000 members, several unions represent-ing manufacturing workers are among the largest, withmemberships between 100,000 and 1.5 million.3 Seetable B-1. While unions may influence the adoptionof programmable automation and its impacts on theirmembers, the adoption of programmable automationmay in turn affect the strengths and abilities of unionsinsofar as job content, numbers of different types ofworkers, wage levels, and job satisfaction levelschange. How unions change as programmable automa-tion is adopted has implications for both the spreadof automation and the characteristics of industrialrelations.

In addition to unions, and to the various entities thatinfluence labor-management relations in unionized set-tings (e.g., the National Labor Relations Board, theFederal Mediation and Conciliation Service, arbitra-tors, and the courts), there are other institutions thatshape industrial relations in both unionized and non-unionized settings and that may influence the adop-tion of programmable automation and its impacts.These include labor-management committees (insti-

1

IJ. S. Bureau of Labor Statistics, Directory of National Unions and Emp-loyee Associations, 1980.

tuted in both unionized and nonunionized settings),and government regulatory agencies such as the Occu-pational Safety and Health Administration and theEqual Employment Opportunity Commission.

The remainder of this section will provide a briefdescription of the collective bargaining process andoutline some of the issues facing labor organizationsand management in the context of the spread of pro-grammable automation. Union and management atti-tudes and practices regarding education and trainingand working environment issues are addressed else-where in this report. Industrial relations in nonunionsettings is not addressed in this technical memoran-dum.

Legal/Regulatory Framework

The central feature of labor-management relationsin the unionized setting is collective bargaining, theprocess of negotiating the terms and conditions ofwork that will be codified in a contract that may ap-ply for a period of 1 to 3 or more years. Guidelinesfor collective bargaining governing the processes ofunionization and selection of worker representatives,procedures for bargaining and resolving disputes, and

Table B-1 .—National Unions and Employee Associations Reporting 100,000 Members or More, 1978a

Members MembersOrganizationb (in thousands) Organizationb (in thousands)

Unions:Teamsters (Ind.) ... , . . . . . . . . . . . . . . . . . . . 1,924Automobile Workers (Ind.). . . . . . . . . . . . . . . 1,499Steelworkers . . . . . . . . . . . . . . . . . . . . . . . . . . 1,286State, County. . . . . . . . . . . . . . . . . . . . . . . . . . 1,020Electrical (IBEW) . . . . . . . . . . . . . . . . . . . . . . . 1,012Machinists . . . . . . . . . . . . . . . . . . . . . . . . . . . . 921Carpenters . . . . . . . . . . . . . . . . . . . . . . . . . . . . 769Retail Clerks . . . . . . . . . . . . . . . . . . . . . . . . . . 736Service Employees . . . . . . . . . . . . . . . . . . . . . 625Laborers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 610Communications Workers . . . . . . . . . . . . . . . 508Clothing and Textile Workers . . . . . . . . . . . . 501Meat Cutters . . . . . . . . . . . . . . . . . . . . . . . . . .Teachers. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 500Operating Engineers . . . . . . . . . . . . . . . . . . . 412Hotel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 404Ladies’ Garment . . . . . . . . . . . . . . . . . . . . . . . 348Plumbers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 337Musicians. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 330Mine Workers (Ind.) . . . . . . . . . . . . . . . . . . . . 308Paperworkers. . . . . . . . . . . . . . . . . . . . . . . . . . 284Government (AFGE) . . . . . . . . . . . . . . . . . . . . 266Electrical (IUE). . . . . . . . . . . . . . . . . . . . . . . . . 255Postal Workers . . . . . . . . . . . . . . . . . . . . . . . . 246Letter Carriers . . . . . . . . . . . . . . . . . . . . . . . . . 227

Unions:—ContinuedGovernment (NAGE) (Ind.) . . . . . . . . . . . . . . . 200Railway Clerks. . . . . . . . . . . . . . . . . . . . . . . . . 200Rubber . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 200Retail, Wholesale . . . . . . . . . . . . . . . . . . . . . . 198Painters . . . . . . . . . . . . . . . . . . . . , . . . . . . . . . 190Oil, Chemical . . . . . . . . . . . . . . . . . . . . . . . . . . 180Fire Fighters . . . . . . . . . . . . . . . . . . . . . . . . . . 176Transportation Union . . . . . . . . . . . . . . . . . . . 176Iron Workers . . . . . . . . . . . . . . . . . . . . . . . . . . 175Bakery, Confectionery, Tobacco . . . . . . . . . 167Electrical (UE) (Ind.) . . . . . . . . . . . . . . . . . . . . 166Sheet Metal . . . . . . . . . . . . . . . . . . . . . . . . . . . 159Transit Union. . . . . . . . . . . . . . . . . . . . . . . . . . 154Boilermakers . . . . . . . . . . . . . . . . . . . . . . . . . . 146Transport Workers . . . . . . . . . . . . . . . . . . . . . 130Printing and Graphic . . . . . . . . . . . . . . . . . . . 120Maintenance of Way . . . . . . . . . . . . . . . . . . . 119Woodworkers. . . . . . . . . . . . . . . . . . . . . . . . . . 118Office . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105

Associations:National Education Association. . . . . . . . . . 1,696Nurses Association . . . . . . . . . . . . . . . . . . . . 187Classified School Employees . . . . . . . . . . . . 150Police . . . . . . . . . . . . . . . . . . . . ., . . . . . . . . . . 140California . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105

aBmw on ~epofl~ t. the Bureau. All Unlong not identiflad as (Ind.) are affiliated with the AFL-CIO.bFor ~erger~ and changes since 1978, see app. ‘

SOURCE: U.S. Department of Labor, “Directory of National Unions and Employee Associations, 1979.”

55

the sanctioning of unfair labor practices on the partof both management and labor, are found in severalpieces of Federal legislation: 1) the National LaborRelations Act (Wagner Act/NLRA) of 1935, which es-tablished the National Labor Relations Board (NLRB)for labor practices rulemaking, investigation, anddispute-adjudication; and 2) its amendments promul-gated in 1947 (Taft-Hartley Act) and 1959 (Landrum-Griffin Act).4 The statutory framework for collectivebargaining has remained unchanged since 1959, al-though attempts at legal reform were made unsuccess-fully in the late 1970’s.

Labor contracts can have enormous influence onhow programmable automation affects existing andfuture workers in unionized firms. What kind of influ-ence they have depends on what is included in the con-tracts, how the contracts are administered, and howNLRB, arbitrators, and courts interpret provisions sub-ject to dispute.

The NLRA established that “wages, hours, and otherterms and conditions of employment” constitute man-datory bargaining material. NLRB has interpreted thisprovision to mean that labor and management maynegotiate over issues in two categories, one categoryof issues for which bargaining is mandatory, and onecategory of issues for which bargaining is permissiblebut not mandatory. NLRB and court rulings on theadoption of (conventional) automation through the1970’s generally imposed a requirement to bargain asto the effects of automation on workers, but not onthe decision of whether and when to introduce auto-mation.

Automation and the Law

Past NLRB and court rulings have generally treatedthe decision to automate as protected by “managerialrights” established in labor contracts. The breadth ofthe managerial rights protection depends on the lan-guage of the contract and its interpretation, given man-agement’s other obligations. Managerial rights havebeen construed to apply (in the absence of proven anti-union conduct) to the control of the production proc-ess, including the making of changes in property,plant, and equipment associated with production. Al-though changes in property, plant, and equipment canaffect the terms and conditions of employment, andcan, especially in the long term, lead to reductions incompany employment levels, NLRB rulings to date im-ply that employers need not bargain where new tech-nology “does not deprive employees of jobs, work op-portunities, or otherwise cause a real change in work-ing conditions” immediately. s Similarly, arbitration

’29 U. SC. sec. 151-167 (1964).‘Joseph Manners, “New Technology and the Law, ” notes for remarks pre-

sented at IAM Electronics and New Technology Conference, Sept. 21, 1982.

rulings regarding the interpretation of existing con-tracts suggest that management is accorded broad dis-cretion for implementing new technology, alteringwork rules, and reallocating work between employeesin the bargaining unit and others as a result of tech-nological change, in the absence of specific contractlanguage governing such changes.’

Both the language of NLRA and past rulings ofNLRB and the courts leave unanswered many ques-tions regarding the scope and timing of bargaining towhich an employer is obligated regarding the adop-tion of new technology in general and programmableautomation in particular.7 Consequently, in the ab-sence of new legislation, the development of clearerstandards for collective bargaining regarding program-mable automation would appear to await the passageof time and the development of precedents throughNLRB and court rulings. The development of prece-dent, in turn, will depend in part on the changingmembership of the NLRB which is comprised of presi-dential appointees serving 5-year terms.8 Additionaldiscussion of the role of NLRB may be found in a paperby W. Cooke, appendix C.

Contract Language

Existing contracts vary greatly in the degree to whichthey can influence the adoption of programmableautomation or its effects. The substantive focus ofmost labor contracts has historically been on such mat-ters as wages and hours, work rules and labor grades,and procedures for grievance resolution. Indeed, agovernment survey of labor contracts covering at least1,000 workers that were in effect at the beginning of1980 indicates concern over only one issue directlyrelevant to the adoption of programmable automation—advance notice of technological change. See tableB-2. The general lack of specificity of past contractswith respect to technological change suggests that mostunionized workers are preoccupied with the so-calledbread and butter” issues of wages and hours and thatthey may accept management’s responsibility to makeand implement decisions necessary to keep the com-pany financially healthy and competitive—except, per-haps, where those decisions can be clearly linked tothreats to job security. The infrequency of specificlanguage regarding technological change may alsoreflect a lack of appreciation on the part of workersof how technological change may affect employment

6Doris B. McLaughlin, ‘The Impact of Labor Unions on the Rate and Direc-tion of Technological Innovation,” report prepared for the National ScienceFoundation (Detroit, Mich.: Wayne State University, Institute of Labor andIndustrial Relations, 1979); and Manners, op. cit.

7“Notes: Automation and Collective Bargaining,” Harvard Law Review,84, 1971.

‘Robert S. Greenberger, “Reagan NLRB Tilts Toward Management, ” WallStreet Journal, Aug. 2, 1982.

56

Table B.2.—Major Collective Bargaining Agreements Advance Notice Provisions by Industry(agreements covering 1,000 workers or more, January 1, 1980)

Requiring advance notice

Plant shutdownAll agreements Total Layoff or relocation Technological change

Industry Agreements Workers Agreements Workers Agreements Workers Agreements Workers Agreements Workers

’75079

8113111174215361514113588418163

11211

9

1662808112

1233188

3272

6,593,8003,025,150

234,20021,80026,850

207,90017,10023,10085,00031,80061,70025,50068,85023,10093,600

460,80097,000

242,150323,750957,10027,85014,800

3,588,850189,050489,550620,000210,70023,900

405,200148,300323,450

1,195,0003.500

796499

4587

114

111425

912

528483585817611

4297

6246353

882122621—

3,689,1002,202,350

159,90021,80021,000

118,0006,700

13,10034,90030,80036,85015,50052,5009,750

63,050193,60067,150

212,100259,850810,15027,850

6,2001,488,750

148,200128,800492,450155,800

18,250304,45051,050

115,80072,354)

—

Ail Industries. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.550 682 2 , 9 8 6 , 7 0 0 lg 1,201,650Manufacturing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108 504.950 713,950

Food, kindred products . . . . . . . . . . . . . . . . . . . .Tobacco manufacturing. . . . . . . . . . . . . . . . . . . .Textile mill products . . . . . . . . . . . . . . . . . . . . . .Apparel. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Lumber, wood products . . . . . . . . . . . . . . . . . . .Furniture, fixtures. . . . . . . . . . . . . . . . . . . . . . . . .Paper, allied products . . . . . . . . . . . . . . . . . . . . .Printing and publishing . . . . . . . . . . . . . . . . . . . .Chemicals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Petroleum refining . . . . . . . . . . . . . . . . . . . . . . . .Rubber and plastics. . . . . . . . . . . . . . . . . . . . . . .Leather products . . . . . . . . . . . . . . . . . . . . . . . . .Stone, clay, and glass . . . . . . . . . . . . . . . . . . . . .Primary metals . . . . . . . . . . . . . . . . . . . . . . . . . . .Fabricated metals. . . . . . . . . . . . . . . . . . . . . . . . .Nonelctrician machinery . . . . . . . . . . . . . . . . . . .Electrical machinery . . . . . . . . . . . . . . . . . . . . . .Transportation equipment . . . . . . . . . . . . . . . . . .Instruments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Miscellaneous manufacturing . . . . . . . . . . . . . .

Nonmanufacturing . . . . . . . . . . . . . . . . . . . . . . . . . .Mining, crude petroleum, and natural gas . . . .

aTransportation . . . . . . . . . . . . . . . . . . . . . . . . . . .Communications. . . . . . . . . . . . . . . . . . . . . . . . . .Utilities, electric, and gas . . . . . . . . . . . . . . . . . .Wholesale trade . . . . . . . . . . . . . . . . . . . . . . . . . .Retail trade . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Hotels and restaurants . . . . . . . . . . . . . . . . . . . .Services . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Construction . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Miscellaneous nonmanufacturing . . . . . . . . . . .

NOTE: Nonedditive.aExcludes railroeds and airlines.

SOURCE: U.S. Department of Laboc ”Characteristics of Major Collective Bargaining Agreements, Januatyl, 1960:’ May 1961.

and working conditions, and/or an inability of unionsto negotiate successfully for such language.

Overall, the scope of labor contracts began to ex-pand beyond traditional provisions in the 1960’s inresponse to technological change, growth in foreigncompetition, and growth in the practice of subcon-tracting work to both domestic, and particularly for-eign, firms. Clauses in the following areas, which maybe relevant to the adoption of programmable automa-tion, have become more common during the past twodecades:

● Job and Wage Security. Retraining (for whom,who pays); layoff, transfer, and relocation proce-dures; “red-circling’’( maintenance) of wages ofpersons transferred to lower paying jobs; sever-ance payments; early retirement.

• Technology Change. Advance notice; consulta-tion; establishment of labor-management advi-sory committees.

In 1966 the Automation Commission endorsed thepractice of advance notice of technological change asa measure that the private sector could take to facilitateadjustments in the labor market, together with explicit

43136

52249

101223

893

214332845974114

2514

166150

584102219—

1,756,75080,00016,1002,3006,2006,700

10,80017,95028,40034,45013,00020,2506,300

67,750128,55079,750

21O,900I188,300807,20027,650

6,2001,229,950

133,20056,950

463,650135,350

11,200226,50032,25008,15054,700

—

14 89;9006 11,8001 5,0002 18,2001 l,5002 2,3009 15,2503 3,3002 3,8006 11,8004 34,0501 1,100

15 51,1507 75,8506 11,800

11 50,15010 99,700

5 13,5503 5,550

— —42 204,250— —10 78.950

3 16,2502 3,7501 1,550

17 48,0001 1,0005 31,5003 21,250

— —

12—68

——

99512273

—7

:——61

35792

4339

——

32;650—

19,80096,800

——

17,10024,S00

9,0001,700

23,4503,450

17,3508,800—

12,6509,000

437,400——

487,700140,000

23,65067,45036,450

5,500148,95021,75045,950

——

advance planning by companies for attrition and otherinternal work force adjustments. g A comparison ofcontract scope in the mid-1960’s and the early 1980’sis provided in a paper by M. Roberts, appendix C.

Additional areas for labor contract change in con-nection with programmable automation include modi-fication of work hours (currently included in some con-tracts as a means of adapting to periods of slack busi-ness), specific triggers for reopening negotiationsbefore contracts formally expire, procedures for reclas-sifying workers, definition of and assignment of workto the bargaining unit, and involvement of labor rep-resentatives in planning, design, and purchase deci-sions for automated systems. Whether, when, and howlabor contracts accommodate the adoption of automa-tion will depend on many factors, such as the dura-tion of the current concessionary bargaining trend andthe weight given to technological change relative toother concerns by both labor and management. IAM,for example, appears to attach great weight to tech-

‘Techno]o~ and the AmerJc“ an lkonomy, report of the U.S. National Com-mission on Technology, Automation, and Fconomic Progress, February 1980.

57

nological change, especially automation, as a bargain-ing issue; it has included technological change provi-sions in model contract language it has developed sincethe 1960’s. In 1982, two IAM locals engaged in long-term strikes over proposed work-rule changes associ-ated with programmable automation.l0

A key question with regard to the impacts of pro-grammable automation on industrial relations amongunionized firms is whether the collective bargainingframework is adequate for meeting needs of both laborand management with respect to programmable auto-mation. At this time, there does not appear to be em-pirical data suitable for evaluating how programmableautomation may affect industrial relations, and viceversa. Participants in the OTA Labor Markets and In-dustrial Relations Workshop appeared to agree thatcollective bargaining can accommodate new needsassociated with programmable automation, althoughsome participants maintained that the resiliency of col-lective bargaining depends in part on how the relativebargaining power of unions and management changesin response to new technology and to other factors.A discussion of relative bargaining power is providedin a paper by W. Cooke, appendix C.

1OMari]yn chase, “wOrh Rule Changes Sought, ” American Metal Market/Metalworking News, Oct. 25, 1982.

Institutional Change

The overall bargaining power of unions relative tomanagement and the overall role played by unions inthe transition to new manufacturing technologies, in-cluding programmable automation, depend on the ex-tent of union representation and on the response ofunions to specific aspects of programmable automa-tion (and other new technologies). Factors influencingunion representation and union responses to new tech-nology are outlined below.

Union representation is largely a function of numeri-cal strength. Changes in the numerical strength of thelabor movement as a whole are widely acknowledged.Although membership in labor organizations hasgrown, the proportion of the labor force that isorganized and the rate of growth of union member-ship have both declined during recent decades, andunions have been less successful in arranging and win-ning elections. Moreover, unions have become less suc-cessful in overcoming recertification efforts in the pastfew years. See figure B-1.

Factors Influencing Union Representation

The erosion of overall union representation has beenattributed to many factors, including changes in em-

Figure B-1 .—Change in Union Representation Over Time

Chart 1. Membership of national unions, 1930-78 a

20

18

16

14

12

10

8

6

4

2

o1930 35 40 45 50 55 60 65 70 757880

Chart 3. Union membership as a percent of total labor forceand of employees in nonagricultural establishments,1930-78a

4UI Percent of employees i n

nonagricultural establishments

28 - -

24 -

20 “labor force

1 6

8 “

1930 35 40 45 50 55 60 65 70 75788

aExciude9 Can-ian membership but includes rnernbers in other areas outside the United States. Members of AFL-CIO directly affiliated IOGd UniOnS are alSO included.Members of single-firm and local unaffiliated unions are excluded. For the years 194&52, midpoints of membership estimates w-hich were expressed as ranges were used.

SOURCE. U.S Department of Labor, “Directory of National Unions and Employee Associations, 1979. ”

58

ployer practices (as a factor enhancing employer effec-tiveness in avoiding unionization), relocation of pro-duction, structural change in the economy, and prolif-eration of new parties to industrial relations activities.It is uncertain, however, whether the overall economicstrength of unions has declined commensurately.

Modern personnel practices may diminish the incen-tive of workers to organize where management pro-vides grievance procedures, complaint channels, com-pany information, fair compensation, and other serv-ices or benefits that unions have been instrumental inlaunching at unionized firms. Personnel practices haveimproved as a result of growth in government regula-tion of employment conditions, growth in businessschool training of managers, increased attention ofbusiness school curricula to human resource manage-ment, and other factors.11 One industrial relationsanalyst relates change in employer practices to thespread among managers of the view that “unions ex-ist as a reflection of management failures, ” althoughhe notes that such generalizations tend to be unmer-ited, reflecting doctrine rather than analysis of specificsituations. 12 A review of the industrial relations litera-ture shows that this characterization appears to be ac-cepted by many academic observers of industrial rela-tions trends.

The shift in location of production from unionizedto nonunionized regions in the United States, and fromthe United States to other countries, has also dimin-ished the union presence in the workplace. Locationalshifts occur for many reasons, most related to costs,and in some cases including a desire by managementto evade unions.13 Where locational shifts involveplant closings, unions can gain political supportthrough community opposition to closings. * On theother hand, management develops political support(though not necessarily at the local level) by relatinglocational and other decisions to business strategy formaintaining competitiveness. Although “competitive-ness” has become a battle cry in rhetorical wars be-tween unions and employers, the true extent of the ef-fect of unions on industrial competitiveness, and thesoundness of that rationale for relocating productionfacilities away from unionized areas, are uncertain.

Another important factor in observed erosion ofunion representation is structural change in the econ-omy. In brief, growth in service industry relative tomanufacturing employment, and growth of public sec-

1lD, ~~ Mills, ‘~an=ment Performance, ” and Fred K. Foulk=, ‘largeNon Unionized Employers, “ in U.S. Mustrial Relations 19s01980: A Criti-cal Assessment, Jack Steiber, et al. (eds. ) (Madison, Wis.: Industrial Rela-tions Research Association, 1981).

‘zIbid.I’Ibid.● Point debated in 1982 OTA Labor Markets and Industrial Relations Work-

shop.

tor relative to private sector employment have in-creased the proportion of employment opportunitiesin occupations and industries with traditionally lim-ited union representation. See figure B-2. Moreover,growth in electronics and other so-called “high tech”industries which have little union representation rela-tive to traditional manufacturing has also reduced theproportion of employment in unionized industries (al-though unionized, traditional manufacturing industriesemploy more people than high-tech industries). * Thecontinuation of these divisions between predominantlyunion and nonunion industries and sectors is uncer-tain.

Finally, several new parties have entered the indus-trial relations arena in the past two to three decades.First, the use of consultants who specialize in person-nel management and in combating unions and the es-tablishment of labor-management committees havegrown among both unionized and nonunionizedfirms. ** Although the legality of labor-managementcommittees in unionized firms has been questioned (aspossibly unfair employer interference in the bargain-ing process), and although some unions regard com-mittees as conflicting with the bargaining process,many committees have been established through col-lective bargaining, and legal problems are beingresolved. * * * The long-term impacts of labor-manage-

● A BLS analysis conducted for the Joint Economic Committee notes thathigh-tech industries account for 4.6 percent of total wage and salary employ-ment. By contrast, all manufacturing industry wage and salary jobs com-prise about 22 percent of the total.

‘ ● This point was raised at the OTA Labor Market and Industrial Rela-tions Workshop and in a roundtable reported in Fortune magazine, Sept. 20,1982.

● **Point discussed at the OTA Labor Market and Industrial RelationsWorkshop.

Figure B-2.—Job Growth for Major OccupationalCategories Under Alternative Economic Projections,

1978-90

20High

trend I

15

:

1 0

5

0- 1

White-collar Blue-collar Service Farmworkersworkers workers workers

SOURCE: Bureau of Labor Statistics, “Monthly Labor Review,” August 1981.

59

ment committees on union-management relations areunclear, since existing committees differ in focus (e.g.,training, quality control) and scope, and since the cur-rent increase in interest in committees seems linked atleast in part to current economic conditions and im-port levels.

Second, new regulations and regulatory bodies be-gan to influence labor-management relations in theareas of occupational safety and health protection andequal opportunity in hiring and promotion in the1960’s and 1970’s, beginning with the 1969 Coal MineHealth and Safety Act and continuing with the 1970Occupational Health and Safety Act and the 1972Equal Employment Opportunity Act.14 New regula-tions served to force changes in union practices, includ-ing contract modification. Some observers believe thatgovernment regulation of hiring, promotion, and occu-pational health and safety practices may have under-mined the value of collective bargaining in those areas,by establishing new complaint mechanisms for work-ers outside the traditional industrial relations frame-work, and placing an emphasis on concerns of the indi-vidual worker rather than the bargaining unit .15 Occu-pational health and safety regulations, in particular,may also affect unions by promoting technologychange in general and automation in particular. And,

l~pub]ic Law 91.173, public L.aW 91-596, and Public Law 92-261, respec-tively.

IJphy]]ls A, Wallis and James W. Driscoll, “Social Issues in Cokctive Bar-gaining, ” in V, S. industrial Relations 19s01980: A Gitical Assessment, JackSteiber, et al. (eds, ) (Madison, Wis.: Industrial Relations Research Associa-tion, 1981).

as noted earlier, regulations motivated improvementin personnel management.

Programmable automation may present opportuni-ties or liabilities for labor organizations. How labororganizations are affected by programmable automa-tion depends on how the equipment and systems aredeveloped and implemented, and on where they areused. To develop an understanding of how program-mable automation may affect labor organizations, avariety of issues should be addressed, such as theaspects of programmable automation design and im-plementation that may be fundamental to union (andother labor) responses to programmable automation,the degree to which workers consider programmableautomation design and implementation characteristicsto be inevitable or negotiable, and, in particular, theimpact of programmable automation on the organiz-ing base for unions.

While unions are perceived as representing primarilyproduction workers, the application of programmableautomation to all aspects of the manufacturing proc-ess, including nonproduction activities such as draft-ing and inventory control, may broaden the base ofworkers interested in organizing. Already, scientific/engineering and clerical unions have been formed,serving constituencies which may be especially vulner-able to technological change in the future. Whethernonproduction workers do organize at higher rates,and if they do, whether they join unions dominatedby production workers or separate labor organiza-tions, may be important factors in determining howlabor organizations influence the spread of program-mable automation and moderate its impacts.

Appendix C

Papers Prepared for Workshop

The following papers were prepared as backgroundmaterials for the OTA Labor Markets and IndustrialRelations Workshop which was held July 27, 1982. 3.Their content and conclusions are the sole responsibil-ity of the authors and do not necessarily reflect the 4.views of OTA.

1. “The Economics of Technical Progress: LaborIssues Arising From the Spread of Programmable 5.Automation Technologies,” by Eileen Appelbaum. 6.

2. “Assessing the Future Impacts on Employment of

Technological Change: An Input-Output Ap-proach,” by Faye Duchin.‘The Effect of Technical Change on Labor,” byLouis Jadobson and Robert Levy.“Programmable Automation: Its Effect on theScientific-Engineering Labor Market,” by WilliamN. Cooke.“Technology and Labor,” by Markley Roberts.“Labor-Management Relations in an Era of Pro-grammable Automation,” by William’N. Cooke.

THE ECONOMICS OF TECHNICAL PROGRESS: LABOR ISSUESARISING FROM THE SPREAD OF PROGRAMMABLEAUTOMATION TECHNOLOGIESby Eileen Appelbaum*Department of EconomicsTemple University

July 27, 1982

One of the specific objectives of this workshop isto place alternative analytic methods which may beused to make inferences about programmable automa-tion technologies and labor market issues in perspec-tive. A second objective is to specify the informationrequirements associated with the formulation of ap-propriate policies and legislative initiatives. The twoobjectives are related in that the choice of variablesfor study as well as the specification of the behavioralrelations among them is dictated by the system of anal-ysis that is utilized. My remarks are directed to thesecond set of questions which this panel will be con-sidering, subsumed under the heading, “To what ex-tent are the production and use of programmable auto-mation likely to result in unemployment of currentworkers and job displacement?” My view is that anadequate system of analysis is indispensable to anevaluation of the questions raised in this section andto the design of studies capable of providing reliableinformation.

The first section of this paper is a critique of thestandard economic analysis of technology and employ-

● The author is a member of the Working Group on Reindustrializationat the University of Massachusetts (Amherst) and participates in its Subgroupon Microelectronics and Work Process. She is indebted to her colleagues forvaluable discussions, though she is solely responsible for the content of thispaper and any errors in it.

ment. This view probably exercises some influence inthe policy community and elsewhere. Its conclusionsare remarkably sanguine. They may or may not proveto be true; but the analysis itself is faulty and the con-clusions are not supported by it. I will keep my com-ments brief, and will limit them to demonstrating themost critical weaknesses of this approach. In the nextand most important section I will proceed to the con-structive task of developing the economics of technicalprogress to encompass the issues of interest to thispanel. The analysis presented in this section will, Ibelieve, make clear the existence of important relation-ships that are not otherwise obvious, and will providequalitatively new insights in the analysis of theemployment issues. The final section will indicate thekinds of studies that will best serve to increase ourknowledge of the effects of this latest round of automa-tion on workers, jobs, and employment.

Critique of the Standard EconomicAnalysis of Technology

The analytical methodology employed by econo-mists to analyze the effects of the introduction ofrobots on wages and employment of workers in theaffected industries and upon employment generally

60

61

tends to trivialize what is a complex question, and of-fers little policy guidance. This does not appear tocreate any difficulties, mainly because the analysisleads economists to the conclusion that, apart froma short-term need for retraining, workers face no prob-lems. Thus, standard analysis suggests that robotics:1) will have a positive impact on wage levels, 2) willprobably tend to reduce rather than increase unem-ployment in the long run, and 3) will stimulate totalemployment even in the industry introducing therobots.1 The analysis on which these conclusions arebased can only be characterized as glib and superficial.

In analyzing the wage and employment effects ofintroducing robots, a typical approach taken by econ-omists is to present a hypothetical situation. Onetypical presentation compares the case in which awidget factory adds one additional worker to itsexisting production line and increases output by 4 unitsa day to the case in which the firm buys several robotsand hires one additional worker to oversee them, thusincreasing output by 10 units a day. At $20 per widget,the author concludes that the value of a worker’s mar-ginal product has increased from $80 to $200 and thatthe firm, which previously would have been unableto pay any worker more than $80 will now be forcedby competition to pay this new worker something nearthe $200 figure. This type of example, though com-mon, is peculiar; and the argument behind it is logical-ly incorrect.

It is strange that in this example the robots do notreplace any previously employed workers (welders,painters, machinists, etc.), but are simply added to anexisting production process. It’s an unusual technol-ogy, indeed, that simply adds two or three robots toan existing process without materially affecting thatprocess. As no workers have been replaced, utiliza-tion of the robots simply increases employment by oneskilled worker because the robot requires maintenanceand oversight. Economists making this argumentsometimes conclude that the worker hired will be askilled worker. Now, robot-fixer may be a more skilledjob than painter or welder, but robot-minder is not.With only two or three robots, the firm is likely tocontract out for robot maintenance, increasing em-ployment in the contracting firm by considerably lessthan one skilled worker. The worker hired by the firmusing only two or three robots is more likely to be arelatively unskilled adjunct to the robot. By makingthe robot overseer both robot-minder and robot-fixer,this approach manages to evade entirely the issues in-

‘See, for example, Richard K. Vedder, “Robotics and the Economy, ” a staffstudy prepared for the Joint Economic Committee, Congress of the UnitedStates, Mar. 26, 1982 (Washington, D. C.: U.S. Government Printing Office,1982).

volved in the fact that some skills will be downgradedeven as others are increased.

Moreover, the argument is logically incorrect. Ingoing from the first case to the second, an increase ofcapital (two or three robots) as well as an increase oflabor (one robot-fixer) has been slipped into the ex-ample. Capital and labor have both increased, andconsequently it is not possible to speak of an increasein the marginal product of labor. Nor can the firm af-ford to use the total increase in productivity to increasewages. What this argument appears to ignore is thatwith an increase in wages for the newly hired workerfrom $80 to $200, the rate of profit on capital hasdecreased. The amount of capital has increased by twoor three robots, but all the increase in output has goneto pay the wages of the robot-fixer. Consequently,gross profits are unchanged. As a result, the ratio ofnet profits to capital (the rate of profit on capital) hasdecreased. An even more serious consequence is that,with gross profits unchanged, the firm will be unableto amortize the new robots. How will it replace themas they wear out’?

Economists sometimes use oligopolistic industrieslike auto and steel as illustrative examples in consider-ing what will happen to the price of, and demand for,output. In these examples, price and quantity sold ofsteel and cars are determined by impersonal marketforces. Auto and steel firms exercise no control overmarket price and, in particular, they do not view priceas a strategic variable which they can manipulate torestore profitability after the introduction of robots.Nor are steel or auto workers capable of bargainingfor higher money wages as their productivity increases.Thus, the introduction of robots reduces productioncosts, increases supply, and ultimately causes pricesto decline and the quantity demanded to increase. Theproblems with this analysis when applied to industrieslike steel and autos are myriad. If firms exercise marketpower to increase profit, or workers bargain for wageincreases, or the new technology requires greater in-vestment, or the new technology is likely to becomeobsolete in a short time as robots become more sophis-ticated and hence has higher amortization costs, thenthe decline in production costs will be damped, sup-ply need not shift very much, prices need not fall verymuch, and the quantity demanded need not be muchaffected. Real wages and/or profits will increase anddemand may very well be affected, but simple text-book models of supply and demand are not sufficientfor analyzing industries that are highly unionized,capital intensive, and in which firms are able to exer-cise market power. An alternative analysis which posesvery different questions and possibilities, is a prereq-uisite to research.

62

On one point standard economic analysis is correct:there is no stopping the introduction of robots. Evenif it should turn out that employment is adversely af-fected and the rate of profit on capitaJ is approximatelythe same after the introduction of robots as it wasbefore, firms cannot avoid trouble by never adoptingthe new technology. So long as some firms (in thiscase, Japanese or European firms) are going to in-troduce the new technique, U.S. firms will have to dolikewise or find they are unable to compete in worldmarkets. In the absence of competition, firms havesometimes delayed the introduction of cost savingtechniques to protect the value of existing capital. Inthe presence of competition, delay risks loss of salesand, hence, of jobs. U.S. firms will ultimately haveto adopt robotics and related technologies. If they doso later rather than sooner, they will lose whateveradvantages accrue to having been among the leaders.

The Economics of Technical Progress:Labor Issues Arising From the Spread ofProgrammable Automation Technologies*

Programmable automation technologies are intro-duced in order to reduce manufacturing cost per unitof revenue. That is, as a result of robotics, firms ex-pect their net receipts from a given outlay to be higherthan they would have been otherwise. A successful in-novation reduces costs—saving labor cost, or savingcapital cost, or expanding the resource base and thussaving resource cost. An innovation may be worthmaking even if it raises one element of cost providedthat it reduces other elements more. Thus, program-mable automation technologies will be introducedbecause they reduce labor costs by increasing outputper man-hour or by allowing the substitution ofcheaper for higher paid labor, or because they reduceamortization charges by embodying major advancesin technique in new plant and equipment, thus reduc-ing the risk that the capital stock will become pre-maturely obsolete and extending its probable usefullife. Such changes in technique maybe cost saving evenif they increase capital costs. The effect of the newtechnologies on wages and employment will vary,however, depending on whether they are capital sav-ing or capital using. Capital costs are increased, andthe new technique may be characterized as capitalusing, if the firm requires higher levels of investmentin order to achieve a given increase in capacity. It may

This argument owes much to Joan Robinson, “Notes on the Economicsof Technical Progress,” Z% Generalisation of the General Theory and OtherEssays (New York: St, Martin’s Press, 1979 (1952)). The importance of capitalsaving technology in economic development was suggested to me by Pro-fessor Thomas Hughes of the University of Pennsylvania.

be characterized as capital saving if less investment willachieve a given increase in capacity, and it may becharacterized as neutral if the amount of investmentrequired to achieve a given increase in capacity is thesame with the old and new techniques.

Whether they are capital using or capital saving,programmable automation technologies are labor sav-ing. A saving in labor cost may be due to an increasein output per man-hour with the type of labor (i.e.,the distribution of skill levels) unchanged, or it maybe due to the substitution of less skilled labor for moreskilled. It will, thus, be necessary to consider severalcases.

Wage and Employment Effects

Programmable automation technologies are likelyto spread throughout the manufacturing sector, thoughthey may have fewer applications in some industriesthan in others. Consider first what will happen if thesetechnologies are capital saving (or neutral) and, in ad-dition, are labor saving due to an increase in outputper man-hour and not due to the deskilling of work.If money wage rates are unchanged following the in-troduction of robots and related technologies, then theeffect of these innovations depends on what happensto the price of manufactured goods. If the price of theseproducts is unchanged, the net profit per unit of out-put and net profit per unit of capital will be increased.Real wages will be unchanged. If the prices of theseproducts decline, as a result perhaps of the strengthof foreign competition for the domestic market, thenreal wage rates will rise. The increase in real wageswill be greatest when competition is strong enough toforce prices down to the point at which net profit perunit of capital is the same after the change in technol-ogy as it was before the change. Lesser price decreasesimply an increase in the rate of profit and smaller in-creases in real wages. An alternative scenario is onein which money wage rates increase in line with pro-ductivity advances so that unit labor costs are con-stant while prices are constant or increase. Again, solong as competition is sufficient to assure that pricesincrease by less than wages, real wages will increase.The extent of the increase in real wages will dependon the strength of competition in product markets,which limits the increase in the rate of net profit oncapital following the introduction of the new tech-nique. Provided that competition in product marketsis not entirely absent, real wages of manufacturingworkers will increase following the spread of program-mable automation techniques if the proportion ofskilled manufacturing jobs does not decline.

The effect on employment in manufacturing de-pends, in the first instance, on what happens to effec-

63

tive demand for manufactured commodities and cap-ital goods. To maintain a given level of manufactur-ing employment, demand for manufactured goods (in-cluding robots) must increase in the same proportionas output per worker. If the new technology is capitalsaving (or neutral), then the increase in capital goodsrequired to maintain a given level of employment willbe less than proportionate (just proportionate) to theincrease in output forthcoming from that employment.Then the increase in demand (including export de-mand) for manufactured commodities will have to in-crease more rapidly than (at the same rate as) outputper head in order to maintain employment constant.If effective demand for manufactured products doesnot grow apace, technological unemployment in man-ufacturing will result. Keynesian demand managementpolicies may or may not be capable of increasing ef-fective demand sufficiently to prevent unemploymentin individual industries—it depends on the extent towhich the new technology increases output perworker. Given worldwide excess capacity in steel pro-duction, for example, reliance on demand managementalone is insufficient to restore the U.S. steel industry’semployment to its 1979 level, even after cost savingrobots are introduced.

A reduction in the workweek with no reduction inweekly income is an alternative means of maintain-ing employment in manufacturing. The question thatarises is whether this can be done without raising unitlabor costs or reducing real weekly wages. If the newtechnology is capital saving or neutral and doesn’treduce the proportion of skilled jobs, the answer is“yes.” With output per worker increasing and the ratioof investment to output constant or declining, it isalways feasible to reduce the workweek without a re-duction in weekly income or an increase in price. Theincrease in labor productivity in an industry may beshared between a slower increase in weekly incomeand a reduction in the workweek for workers in thatindustry. * A reduction in the workweek under these

● The following example may clarify this point. Consider a firm thatemploys 10 workers at 40 hours a week and produces 100 widgets a week.The workers are paid $1.00 an hour and the widgets are priced at $5.00 each.Each worker earns $40 per week. The total wage bill is $400 and total profitsare $100. Now suppose a capital saving advance in technology is introducedthat increases output per worker by 50 percent, and that wages also rise by50 percent to $1..50 an hour. Suppose that demand for widgets increases, butnot enough to maintain employment of all 10 workers. Suppose that demandfor widgets increases to 120 a week. Following the advance in technology,this output can be produced with only eight workers. The total wage billis now $480 (eight workers working 40 hours at $1.50 an hour). The wagebill has increased 20 percent. With no increase in price, total profits will in-crease to $120, also a 20-percent increase. The eight workers are employeda total of 320 hours a week. If a decision is made to maintain employmentby reducing the workweek, the 10 workers will each work a 32-hour week.At $1.50 an hour, each worker will earn $48 a week. The workweek hasdecreased, labor costs have not increased, real weekly wages have increased(though more slowly than they otherwise would have), and the increase inprofits is unaffected.

circumstances is always possible. There are at least tworeasons, however, why it may not be practical. First,a shorter workweek is only possible if prices of the in-dustry’s products do not decline. In the face of inter-national competition for domestic markets, it may notbe possible to meet this condition. Second, considerwhat happens when some manufacturing industries arebetter able than others to utilize the new technologies,and thus achieve much greater increases in productivi-ty. If the standard workweek in manufacturing is re-duced, the reduction is likely to apply to all manufac-turing industries, including those with slower produc-tivity gains. Labor costs and prices in industries whereproductivity improves less must rise as a result of thereduction in the workweek. If the products of such in-dustries are capital goods, they will raise the final priceof the outputs of the technically progressive industriesin which they are used, possibly making them lesscompetitive in world markets. In any event, the in-crease in the price of output of industries where thenew automation technologies are less applicable retardsthe rise in real wages of workers. Thus, a reductionin the weekly hours of manufacturing workers withno reduction in weekly income can most easily be un-dertaken if 1)programmable automation technologiesprove to be capital saving or neutral, 2) real wages ofworkers rise because money wages increase in line withproductivity increases while prices remain constant(and not because money wages remain constant whileprices decline), and 3) the technologies are widelyspread among manufacturing firms and not muchmore highly concentrated in some than in others.

If programmable automation technologies are cap-ital using rather than capital saving or neutral, thediscussion of the effect on real wages has to bemodified somewhat. Again, if money wages and pricesare both unchanged, net profit per unit of output willincrease. Whether any increase in the rate of profit oncapital and/or real wages is possible depends onwhether the increase in the amount of capital requiredper unit of output and in net profit per unit of outputare proportionate. That is, it depends on whether thesaving in labor cost is completely offset by the increasein capital costs. If it is not entirely offset, then pro-vided competition in product markets is not entirelyabsent, real wages of labor will rise. The increase,however, will be more moderate than if the technol-ogy were capital saving or neutral. The reason for thisis straightforward: If the new technologies are capitalusing, a larger share of gross output will be requiredfor amortization. The increase in net output availableto be divided between increases in real wages and inthe rate of profit is consequently smaller. For the samereason, the possibilities for trading off increases in realwages for a shorter workweek are more limited.

64b

Even when demand for manufactured goods growsin line with the growth of output per worker, a capitalsaving technology will not directly increase employ-ment in the capital goods industries. A capital usingtechnology, however, may have this effect. With in-vestment requirements higher, whether employmentwill increase in the capital goods industries dependson whether these industries have also introduced laborsaving technologies as they shifted from the produc-tion of ordinary to programmable automation technol-ogies.

An important motive for the decision of firms to in-troduce capital using techniques is that such techniquesmay substitute machinery for skill, reducing the skillcomponent of the task, and making possible the sub-stitution of lesser skilled machine-tenders for morehighly skilled craftsmen. * Such developments in tech-nology save labor costs by allowing firms to substitutecheap labor for more expensive, and bring us to thethird case which must be considered. This type of tech-nical change is, perhaps, more invidious than a changewhich saves labor by increasing output per workerwithout reducing skill requirements. One reason is thateven if aggregate demand policies could be devised tomaintain reasonable full aggregate employment levels,skilled manufacturing workers would be displaced byless skilled workers. If the wages of skilled workersdo not then decline substantially, the new techniquewill diffuse through the industry, effectively displac-ing the skilled workers. A substantial decline in wagesof skilled workers might check the spread of the newtechnique, allowing both techniques to operate side byside. This outcome, however, would probably requirea decline in the wages of skilled workers sufficient tonearly eliminate the wage differential between skilledand unskilled workers. Unlike changes in technologythat save labor costs by increasing output per worker,changes in technology that save labor costs by enablingfirms to substitute cheap, unskilled labor for morehighly skilled workers do not increase real wages.Whether the rate of profit on capital increases or not,net profit as a share of output and amortization willboth increase as a result of the introduction of a morecapital using technique. Given the overabundance ofworkers available for less skilled jobs in the U.S. econ-omy, and their lack of union organization and bargain-ing strength, it is unlikely that real wage rates of lessskilled workers will increase. Union organization of

● Such a development has occurred in the last decade in the baked goodsindustry. Bakers continued to be classified among craft workers although,increasingly, the baker’s task has become one of minding, measuring, andmixing machines, Cheap labor, often female, has been substituted for highlyskilled craft labor as mechanization increased and the job was deskilled.

these workers could change this, of course, by cap-turing for the less skilled employees some part of thecost saving associated with the elimination of morehighly skilled jobs. Otherwise, average wage rate inmanufacturing will be reduced and the rate of profitincreased. In any event, high wage jobs will have beeneliminated.

This raises a related issue. As we have just seen,firms in an industry can improve their cash flow posi-tion (net profit plus amortization) by adopting a tech-nique that is capital using but saves labor cost byallowing the substitution of less skilled labor for moreskilled. For this reason, they may choose such a tech-nique in preference to an alternative available tech-nology that saves labor cost by increasing output perworker without reducing skill levels since such tech-niques, in general, increase real wages. Such a choice,however, may have serious negative ramifications forfurther technological change 10 or 20 years in thefuture. These negative impacts may affect the com-petitive position of such firms vis-a-vis foreign com-petitors who have chosen not to reduce the propor-tion of skilled workers employed. A hypothetical ex-ample, which Peter Albin of CUNY suggests may notbe so far from the truth, may serve to illustrate thispoint. Suppose that when a U.S. firm puts in a roboticinstallation, it replaces as many as six mastermachinists with one programmer plus three entry-levelpeople whose function is to “load and unload,” keep-ing things lined up for the robots. In Japan, on theother hand, let us suppose that master machinists areretrained and prepared for positions as machinist/pro-grammers. The machinist’s job is transformed but notdowngraded, and the machinist is ready for futurechangeovers. The initial cash flow advantage is gainedby the American firms which have a less skilled andlower paid labor force. However, the man/machineconfiguration in U.S. firms is more permanent, lessflexible. In the absence of skilled master machinists,the opportunities for learning by doing are severelycurtailed. The Japanese in this example, because theyretain their master machinists, need to build less intothe machine, can design less immutable man/machineconfigurations, have enhanced opportunities for learn-ing by doing, and have increased opportunities forcontinuous technological change. Longrun competitiveadvantage would rest with the Japanese. This exam-ple suggests that the substitution of less skilled workersfor craft and highly skilled workers, as a means ofholding down costs and increasing profits in the nearfuture, may be myopic. In the longer period, it couldhave serious implications for international competi-tiveness and manufacturing employment.

65

Changes in Aggregate EmploymentOutside Manufacturing

Changes in technology that involve the deskillingof tasks have a deleterious effect on the employmentpossibilities of skilled manufacturing workers, whilethose that increase output per worker raise the specterof technological unemployment. Such unemploymentwill materialize if growth in demand lags behind pro-ductivity growth. In an open economy, in which im-ports supply some part of the domestic market, growthin demand at home and abroad for domestically pro-duced manufactured products will have to outstrip theproductivity gains just to maintain employment con-stant. Otherwise, as seems likely, employment in U.S.manufacturing will decline. The question, then, iswhether employment in nonmanufacturing sectors ofthe economy will expand and absorb displaced work-ers as well as new labor force entrants who wouldpreviously have found skilled and semiskilled manu-facturing jobs.

Economists are used to thinking of increases in pro-ductivity as increasing real income and spending, thusincreasing demand for services and other products, aswell as employment in the industries producing them.It happens, however, that it matters whether the in-crease in real income is an increase in real wage ratesor an increase in net profit. An increase in net profitresults in an increase in business saving, and sets upthe conditions for a cyclical decline in employment.Moreover, the marginal propensity to save and to con-sume imports is higher among those income groupsreceiving dividends than among wage earners, againwith negative implications for U.S. employment. Somepart of the increase in net profits will increase demandfor producer services (e.g., advertising, financial, ar-chitectural, or computer services), so that output ofthis sector will increase.

An increase in real wages, on the other hand, is like-ly to generate an increase in demand for the full rangeof consumer goods and services. Output will increase,but the effect on aggregate employment is more uncer-tain. Two cases need to be considered: the case inwhich technology in nonmanufacturing firms is essen-tially unchanged, and that in which labor saving pro-grammable automation technologies spread throughthe nonmanufacturing sectors as well.

Suppose that the application of programmable auto-mation technologies is confined to the manufacturingsector and does not spread to the two other broad sec-tors of the economy—other goods producing industries(agriculture, construction, and mining) and nongoodsproducing industries broadly construed as the servicesector. Price declines for some manufactured goods,as a result of the increase in output per worker, are

likely—in clocks, calculators, home computers, etc.Typically, however, real wages of manufacturingworkers and profits of manufacturing firms will in-crease as a result of increases in money wages in linewith productivity and prices that remain largely un-changed, perhaps creeping upward slightly. * Withprices largely unchanged, real wages of nonmanufac-turing workers are also unchanged. As manufactur-ing workers’ spend their higher real incomes, demandfor nonmanufactured goods will increase. The higherincomes (gross profits and total wage bill, though notwage rates) generated in these sectors again increasedemand for manufactured goods, other goods, andservices. Since existing technologies in the service sec-tor have a much higher ratio of labor to “output,”growth in services and consequently in service sectoremployment could possibly maintain aggregate em-ployment, increasing the share of service jobs.

The employment outcome is far less optimistic ifunionized workers in construction, mining, transpor-tation, and the public sector, concerned over thedecline in their relative real wage position, bargain forhigher money wages. The resulting rise in prices, es-pecially of housing, will slow the increase in real wagerates due to the advance in technology in manufac-turing, and hence will retard the increase in employ-ment generated by expenditures out of increased wageincome. Moreover, with prices rising the already lowreal wages of nonunionized workers in retail trade andhealth services, and of many of the clerical workersfound in the service sector, will decline, The resultingchange in relative output prices may stimulate demandfor these services, resulting in a disproportionate in-crease in such low paying jobs.

It is not realistic, of course, to assume that program-mable automation technologies will be limited to themanufacturing sector. In particular, the spread ofdistributed data processing in offices has already begunand should be well under way by 1990. Automationin manufacturing had progressed far even before theintroduction of robots and related technologies. Reli-able estimates of productivity growth as a result ofthese new techniques are not available, though theguess that is bandied around is that between 4 and 9percent of the work in manufacturing is suited to suchautomation. * * The effect on employment will be lessthan 4 to 9 percent since robot support staff—main-tenance and repair personnel, programmers, and ma-chine-tenders-will be required. Productivity might in-crease by as much as 6 percent. These are only guesses,

● Such was approximately the case in the United States from 19s0 to 1964.If there are price shocks from increases in resource prices (oil, food, etc.),a wage-price spiral will upset this pattern of real wage gains.

“h-i the Carnegie-Mellon survey, responding firms placed it at 8 percent.Like most of our information on the subject, this is little more than a guess.

66

of course, but that future productivity gains inmanufacturing will approximate current gains in atechnologically progressive sector like communica-tions, does not seem entirely unreasonable. Unlikemanufacturing, automation in offices is a qualitative-ly new experience. Mechanization, in which machinessupplement the input of workers, has occurred steadilywith the introduction of typewriters, electric type-writers, and word processors as well as other businessmachines. But office automation, in which machinestake over entire worker functions and drastically alterjob content for the remaining workers is a recentphenomenon. Gradual gains in clerical productivitysince 1948 may, within the decade, be replaced byrapid advances in output per worker. Again no reliableestimates of the potential spread of office automationhave been made for the United States. European ob-servors3 suggest that 30 to 40 percent of clerical workmay be suitable for automation, with the impact onemployment in the 20- to 30-percent range. Theseguesses (and I emphasize that they are only guesses)suggest an increase in labor productivity in the officeof about 25 percent. Even if these guesses are too highby a factor of 2 or 3, they suggest the severity of thetechnological unemployment problem that mayemerge if programmable automation technologiesspread through both manufacturing and nonmanufac-turing sectors. With so small a fraction of clericalworkers unionized and so large a percentage of themwomen, clerical workers are likely to reap but few ofthe gains of their higher productivity directly. Majoremployers of clerical workers—insurance companies,accounting firms, advertising agencies, computer serv-ice firms, law firms-may lower their rates, raising realincomes generally and stimulating demand for theirservices. Alternatively, net profits will increase if ratesare unchanged. In either case, the rapid growth thathas characterized clerical employment in the last dec-ade is likely to slow, further compounding the prob-lem of providing jobs for new labor force entrants.

Rapid Technological Change and Investment

One other aspect of the relationship between im-plementation of new technology and employmentshould be briefly noted. Older plant in manufactur-ing industries, auto and steel perhaps, may have beenrendered obsolete more quickly than anticipated. Cer-tainly the step up in the pace of automation has gen-erated a major increase in demand for office equip-ment. During the period of the changeover to pro-grammable automation and microprocessor-basedtechnologies, a period which may last 10 to 20 years

“The Job-Killers of Germany,” New Scientist, June 8, 1978.

while more versatile robots and software for office andother applications are developed, investment in thecapital goods required to replace manufacturing plantthat has had to be prematurely scrapped or to increasecapital-to-labor ratios in offices will generate employ-ment. High capital requirements as offices switch toelectronic equipment and manufacturers use com-puterized machinery implies employment growth overthe next two decades in industries supplying advancedcapital goods. If these goods are produced in U.S.plants, growth in manufacturing jobs in these indus-tries will, during this period, partially offset the lossof manufacturing jobs in industries where the newtechnologies are implemented. The full impact oftechnological unemployment will not be felt until 2000.The unknown here is the extent to which this market,both domestic and international, will be supplied byfirms located in the United States. The Japanese leadin computerized manufacturing is substantial and itsgreater experience in successfully producing and usingrobots gives it an advantage in future sales of thesemachines. The Yamazaki Machinery Works reportsmore than 300 serious inquiries from U.S. companiesalone.4 American firms are better situated in themarkets for electronic office equipment, home com-puters, electronic toys, but competition from Japan,West Germany, the Netherlands, and France is in-tense.5 The growth of U.S. multinational corporations(MNCS) is a further threat to the future of U.S.manufacturing jobs in these industries. The widelyscattered manufacturing operations of semiconductorfirms are a case in point.

Multinational Corporations and Employment

The existence of MNCS further complicates the anal-ysis of job loss due to changing technology. Whenfirms were national entities and not multinational con-cerns, it was increasing wages of skilled workers withina country that motivated firms to seek new techniquesthat save labor cost by substituting less skilled for moreskilled workers. The existence of MNCS with affiliatesin low wage areas means that such firms have an in-centive to invest in techniques that substitute lessskilled Third World workers for American workers solong as the wage differential between the two groupsof workers exceeds some minimum. When transpor-tation costs for bulky products manufactured in theThird World were high, even a substantial wage dif-ferential might not have induced U.S. firms to movetheir fabricating operations abroad. Progressive ad-vances in microminiaturization, lightweight materials,and transportation technology have changed that.

4New York Times, Dec. 13, 1981, sec. 3.5Business Week, Dec. 14, 1981, pp. 39-120 and Dec. 21, 1981, pp. 52-54.

67

The ability of the developing countries to competein the production and sale of manufactured productsto the United States is only partially the result of in-digenous economic development. It results as wellfrom the growth of U.S. multinationals. The uniqueadvantage of the MNC in comparison with nationalfirms is its transnational ability to combine labor andmaterial resources in the host country with technol-ogy and administrative capabilities developed in thehome country. Because labor costs in less developedcountries are so much lower than labor costs in theUnited States the possibility exists that increasingNorth-South trade will result in increasing specializa-tion in production within countries with the develop-ing countries specializing in the production of an in-creasing number of manufactured products. The de-bate over whether steel production, like black andwhite televisions, should be phased out in favor of im-ports emphasizes the immediacy of this concern. Leav-ing aside the issue of whether the United States shouldview industrial structure as an object of policy, for-mulating economic policies designed to achieve somestructural outcome, let us consider the implications foremployment as technology advances. The export ofmanufacturing jobs outside the United States meansthat productivity gains will be realized elsewhere. Howmuch of the increase in productivity will go to increasethe real wages of local workers in American-ownedplants overseas is uncertain. Even if the MNC capturesthe entire productivity increase, it may or may notreduce the price of the product in U.S. markets. A pricereduction would, of course, mean an increase in realincome in the United States, which would raise de-mand for output and increase employment in this way.The magnitude of the effect would depend on howmuch of the increase in productivity the MNC keepsin the form of retained earnings (to finance expansionof its foreign subsidiaries as well as, or in place of, itsAmerican operations) and how much it passed alongin the form of lower prices.

Service Sector Jobs and the Quality ofEmployment Opportunities

The analysis of the previous section focused on theeffects of programmable automation technologies onreal wage rates and on aggregate employment. Todiscuss the qualitative aspects of service sector employ-ment growth requires that we turn our attention tohow labor markets are currently structured. To begin,I want to consider the familiar thesis that investmentin human capital will enable the service sector to con-tinue expanding, so much so that it will easily absorb

the workers displaced from manufacturing.’ This hy-pothesis, relating economic growth to human capitalformation, is a simple extrapolation of the findings of“growth accounting.” In growth accounting, past eco-nomic growth is broken down into its componentsources—quantity and quality of labor inputs, quan-tity and quality of capital inputs, The great accom-plishment of growth accounting has been the iden-tification of the increasing educational attainments ofthe labor force as one of the quantitatively most signifi-cant sources of growth in output per worker between1929 and 1969.7 However, growth accounting providesno theory to explain the nature of the relationship be-tween the quality of labor inputs and economicgrowth. All that can be said on the basis of growthaccounting is that investment in human capital andeconomic growth both proceeded at an impressive rateduring the period following 1948. It is incorrect to in-fer that increased schooling caused economic growthfrom the association between educational attainmentand the growth in gross national product (GNP). Serv-ice sector employment has also grown substantiallyduring this period, though again cause cannot be in-ferred from correlation. In considering the ease withwhich displaced manufacturing workers can be ab-sorbed by a growing service sector, two points needto be addressed. The first relates to the sex-labelingof jobs in the U.S. economy, and the second to thequality of service sector jobs.

Much of the expansion in service sector employmentin the 1970’s was in jobs for which women are the pre-ferred labor force—retail sales including restaurantsand fast food establishments, and health services.These are jobs that typically employ women at wagesbelow the average in manufacturing, and that provideshort hours, few fringe benefits, and little opportuni-ty for advancement. Clerical employment was anotherimportant area of job growth. Here, again, women arethe preferred labor force. Though the jobs are morevaried in terms of responsibilities, opportunities, andwages paid, they are nevertheless among the poorestpaid and least prestigious of the white-collar occupa-tions. The shift to service sector jobs has largely beena shift away from jobs employing male workers andtoward jobs employing females. Despite the growth

bEli Ginzberg and George J. Vojta, ‘The Service Sector of the U.S. Econ-omy,” Scientific American 244, March 1981.

The development of growth accounting is best exemplified in the workof Edward F. Denisen, The Sources of fionomic Growth in the United Statesand the Alternatives Befbre Us (New York: Committee for Economic Develop-ment, 1%2); Why Growth Rates Differ: Postwar Experience in Nine WesternCountries (Washington, D. C.: The Brookings Institution, 1%7); Accountingfor Llnited States Economic Growth, 1929-69 (Washington, D. C.: The Brook-ings Institution, 1974); Accounting for Slower &onomic Growth: The UnitedStates in the 1980’s (Washington, D. C.: The Brookings Institution, 1979).

68

in employment, high unemployment rates for womenpersisted throughout the decade, suggesting there wasno shortage of female workers available to take suchjobs. Had manufacturing employment decreased bynearly a million jobs during the last decade instead ofhaving increased by that amount (measured from itspeak in 1969 to its peak in 1979), with most of thedisplaced workers men, it is unclear for which servicesector jobs they might have been retrained.

My second point is that the growth of service sec-tor jobs has meant an increase in both good and poorjobs. It is evident that the U.S. labor market issegmented into a primary labor market segment inwhich most of the “good” or “acceptable” jobs arelocated, and a secondary segment in which most ofthe “poor” jobs are to be found.8 A further distinctioncan be made between autonomous primary segmentjobs (professionals, like doctors, lawyers, and pro-fessors, or craftsmen like plumbers, electricians,carpenters) and subordinate primary sector jobs (mailcarriers, city transit drivers, steel workers, etc.). Thedifference between a subordinate primary sector joband a secondary sector job is sometimes based onworker skills and sometimes on nonskill-related jobcharacteristics. The better jobs are those in which theemployer values steadiness and low turnover and iswilling to reward tenure on the job with promotionsand higher pay, or in which the employee is repre-sented by a union that has won decent wages, a payscale that rewards seniority, and protection against ar-bitrary treatment. Many of the subordinate primarysector jobs available for male workers are to be foundin the goods producing sector of the economy, whereworkers have more union representation than in theservice sector. This is probably the basis for the casualobservation that changes in technology and the shiftto service sector employment is eroding the middle ofthe job distribution, making the labor market morestarkly two-tiered with good jobs at the top, poor jobsat the bottom, and shrinking opportunity in themiddle.

The growth of the service sector has provided bothgood jobs and poor ones. Producer services (adver-tising, architecture, law, management consulting, com-puter, financial) now account for 19 percent of GNP,and firms in this industry have been a major sourceof good jobs.9 This observation is confirmed by aBureau of Labor Statistics (BLS) study of the computerand data processing services industry (SIC 7372 andSIC 7374),10 This industry currently employs nearly

‘See Richard C. Edwards, Contested Terrain (New York: Basic Books,1979); and David Gordon, The Working Poor (Washington, D. C.: Councilon State Planning, 1980).

‘Ginzberg and Vojta, op. cit.‘“Department of Labor, Bureau of Labor Statistics, “Industry Wage Survey:

Computers and Data Processing Services, ” Bulletin 2028, March 1978.

350,000 workers and is one of the rapidly growingbusiness services.11 The BLS survey found that approx-imately 37 percent of the nonsupervisory employeesin this industry are highly paid professional ortechnical workers, many of them computer systemsanalysts, computer operators, and computer pro-grammers. At the same time, however, the survey foundthat office clerical employees account for another 32percent of employment in this industry. Two-fifths ofthe office clerical employees are key entry operatorsearning $135 to $175 (Class B) or $150 to $205 (ClassA) in 1978. If we calculate their hourly wage on thebasis of a 35-hour week, they earned between $3.86and $5.86 an hour. By comparison, average wages inmanufacturing in 1978 were $6.17 an hour. The pointis that service sector employment, like employment inthe U.S. economy generally, is two-tiered. The growthof this sector has meant an increase in professional,technical, and managerial jobs at the top; but it hasalso meant an increase in clerical, sales, and nonprofes-sional service jobs at the bottom. In particular, poor-ly paying jobs in health services have been a majorsource of employment growth in the 1970’s.

The number of good jobs has grown but, since 1970,it has failed to keep pace with the supply of collegegraduates. Employment prospects have deterioratedsubstantially since the 1950’s and 1960’s.12 The resulthas been a credentials inflation in which a collegedegree is now a prerequisite for many jobs which couldbe done or previously were done by workers with ahigh school education. It would be incorrect, therefore,to conclude that providing an educated work forcewould be sufficient to guarantee the growth of jobsrequiring such workers. With good jobs in the servicesector growing more slowly than the supply of col-lege educated labor, what are the prospects for workerswith less schooling who in earlier years would havefound employment in the manufacturing sector? Re-cent growth in employment opportunities in the bot-tom tier of the service sector suggests that, unlessclerical and service workers are displaced by technol-ogy, aggregate employment will continue to grow. Theavailable jobs, however, may not provide viable alter-natives for workers accustomed to decent wages anddue process—union benefits and protections charac-teristic of much subordinate primary sector employ-ment in manufacturing. Moreover, sex stereotypingof clerical and retail sales occupations operate to denyjobs to displaced male workers even were they to ac-cept the low wages and authority relations typical insuch jobs. Can we specify the jobs for which these

llfiplo~ent ad &m@s, November 1981, table B-2.IZRichard FrWman (cd.), The Overeducated American (New York:

Academic Press, Inc., 1976).

69

young workers should be trained? The future to whichthey should aspire?

Research Needs

It should be evident from the foregoing analysis thatwe need detailed, disaggregate data by industry on:

1.

2.

3.

4.

5.

6.

the kinds of work suitable for automation, withcurrently available robots and programmabletechnologies and with future generations ofrobots;the number of workers likely to be displaced, andthe slowdown in growth or decline in growth ofthe labor force;the demographic and educational characteristicsof the displaced workers;estimates of the average increase in average laborproductivity, derived from 1. and 2.;an estimate- of the cost saving to be realized asa result of introducing the new technology; andestimates of average labor requirements by oc-cupation, capital requirements by type of plantand equipment, and energy requirements per unitof output at a baseline point prior to the introduc-tion of the new technology, and at discrete in-tervals in the next 10 to 20 years.

Some of this information already exists in engineer-ing cost studies of technical improvements and in othertechnical studies done by businesses or consultingfirms, in reports of private research firms, and in casestudies of individual firms. Access to these studies issometimes limited and even the published material isnot always easy to obtain. A chief difficulty is thatit has not been organized into a coherent picture thatwould enable us to say with confidence what is knownand what remains to be learned about the impact ofprogrammable technologies. A survey and synthesisof the existing literature that identifies the gaps in whatis known and is followed up with case studies and sur-veys taken to fill the spaces seems essential. If sucha study is undertaken, it should address not only em-ployment issues but changes in the nature of work aswell. Changes in work process, management functions,and social relations within firms should all be studiedfor insights into changes in the quality of workinglife. ’3 In addition, we need to know how widely used

IJFO~ ~ d~talled exposition of the work process issues involved see EileenAppelbaum, Kenneth Flamm, and Leonard Rapping, “A Proposed State ofthe Art Survey of the Literature on the Microelectronic Revolution, ” mimeo,June 1981.

in manufacturing the programmable automation tech-nologies will be. For firms that do not directly employrobots and related technologies, we need to knowwhether they will utilize capital goods produced usingsuch technologies since this will have the effect ofreducing capital costs in such industries even withouta change in technique (provided, of course, that theprice of the capital goods declines).

Information obtained from such case studies canusefully be incorporated into an input-output frame-work in order to obtain estimates of the size of thelabor force, its demographic composition, and its oc-cupational distribution under alternative assumptionsabout how the technology develops.14 Final demandmight be treated exogenously so that the percentageincrease in various detailed input requirements perdollar of aggregate GNP could be calculated. Demandstructures that hold constant, respectively, the physicaland value composition of output might reasonablyprovide orders of magnitude for demand effects. Sincethe assumptions about the new technologies, or “sce-narios, ” are necessarily subject to substantial uncer-tainty, ranges of plausible technology changes oughtto be tested using the input-output framework to assessthe sensitivity to changes in the scenarios of the qual-itative shifts in demand for energy, labor, and capitalinputs.

Input-output analysis refers to changes in physicalquantities of required inputs per unit of output. Thedefinitions of capital using and capital saving utilizedin this paper depend on the dollar value of investmentper unit of capacity. I am familiar with one study thatexamines that concept empirically for the Canadianeconomy. l5 That study, which is concerned with un-employment and only touches on the technologyissues, uses measures that are aggregated for theeconomy as a whole. The results are not very usefulfor the purpose of studying technology, but themethodology is very straightforward. I have put myresearch assistant to work using this technique to lookat particular industries and time periods to determinewhether technology has been capital saving or capitalusing.

14A detailed d=ription of this methodology can be found in EileenApplbaum, Robert Costrell, Kenneth Flamm, and Leonard A. Rapping, “TheImpact of High Technology on the Structure of Factor Demands and on theProcess of Economic Growth,” mimeo, March 1981.

“Paul Davenport, “Capital-Using Technical Change and Long-PeriodUnemployment in Canada, 1947-1981,” journal of Post Keynesian E20nomics,forthcoming, September 1982.

70

ASSESSING THE FUTURE IMPACTS ON EMPLOYMENT OFTECHNOLOGICAL CHANGE: AN INPUT-OUTPUT APPROACH

by Faye DuchinInstitute for Economic AnalysisNew York University

July 27, 1982The OTA project proposal on Information Technol-

ogy, Automation, and the Workplace is an attemptto anticipate now the likely situation with respect toemployment of the U.S. working-age population inone or two decades. This paper describes a method-ology for quantifying the size and occupational mixof the employed labor force that will be required asa result of the progressive introduction into the varioussectors of the economy of new technologies, in par-ticular production and office automation. It makes fre-quent reference to a model currently being developedfor this purpose at the Institute for Economic Analysis(IEA) with support from the National Science Foun-dation, called the IEA model.

The input-output (I-O) approach is now used in, orrather as a component of, virtually all the well-known,large-scale models of the U.S. economy like that ofthe Bureau of Labor Statistics (BLS) Office of Eco-nomic Growth and the proprietary forecasting modelsof DRI, Chase Econometrics, Wharton, and manyothers. 1-0 is valued in these applications for its abili-ty to disaggregate overall economic activity into sec-toral detail and to ensure the consistency of differentassumptions made about separate parts of the econ-omy. In this paper I will concentrate on other aspectsof I-O that are generally not exploited, and I will tryto address the many detailed questions about meth-odology, assumptions, and data that have been putto me since I agreed to participate in this workshop.I will focus on the issues that can be most readilyanalyzed at the present time for policy purposes butI will also indicate the research areas into which theI-O framework can and should be extended to inte-grate issues related, for example, to education andtraining into the analysis of employment and techno-logical change.

Representing Technology andTechnological Change

The technology used in a particular sector can becharacterized by the mix and amount of each inputrequired to produce one unit of that sector’s output.Inputs include raw materials; various types of proc-essed goods and of services; different categories of

machines, tools, and other capital goods; and an as-sortment of labor skills. A technology is most concrete-ly associated with a single production process, for ex-ample use of the open hearth v. the basic oxygen fur-nace for steelmaking, each requiring a different mixof inputs. Since one technology displaces others onlygradually, there are generally several distinct processesin use in a given sector at any particular time and alsoseveral stages of production; the sector’s “technology”is the weighted average input structure of the variousprocesses in use. The technological structure of ann-sector economy in a given year can be described bythree matrices of coefficients: the A matrix (nxn) ofintermediate inputs (or interindustry transactions), theB matrix (nxn) of capital stock requirements, and theL matrix (mxn) of labor requirements, assuming m oc-cupational categories. All coefficients are expressed perunit of output.

This description is not of course assumed to be anadequate basis for actual production: it is like the listof ingredients for a recipe which does not include thedirections to the cook. Rather, technology is beingdescribed in terms of the demands placed by a sectoron the rest of the economy.

Technology defined in this way also reflects otherinputs not customarily associated with the choice ofproduction process: “overhead” in the form of legalor personnel services or the purchase of an executivejet. The discussion of scenarios below indicates howone proceeds to isolate the phenomena of interest with-in this broad interpretation of technology.

A change in the input structure of a given sector(where inputs are measured in constant physical orvalue units) may reflect any number of underlying fac-tors. In a comparison of statistically compiled, his-torical I-O tables, it will often be necessary to attemptto distinguish the impacts of different factors. On theother hand, one purpose of designing experiments—or scenarios as they are generally called—is to isolatethe changes of interest to assess their separate impacts.

Choice of Classification Scheme

The A, B, and L matrices describe the state of theeconomy at a given time as classified into n produc-

71

ing sectors and m occupations. The number and choiceof categories will depend on the purposes of a par-ticular investigation. (In many practical applications,the abilities to collect the required information and tohandle large matrices will also constrain the choice.)

Sectors can be described in terms of industries orcommodities: * if each industry produces one com-modity or service and no two industries produce thesame commodity or service, the two schemes are thesame. The confounding factors are the presence of:

1. secondary outputs that are marketed; and2. in-house operations (essentially “captive” produc-

tion and producer services) that may, or may not,be the primary output of some other sector. Theseare not reported as outputs.

The commodity scheme has the advantage that thecorresponding input structures are more readily inter-preted in terms of engineering technology although inthe presence of secondary products the “overhead”operations may be difficult to allocate to individualcommodities. The IEA model uses essentially an in-dustry classification for practical reasons: the laborand capital data and price deflators are much morereadily available in that form (for historical reasons).At the level of sectoral detail of the IEA model (the85 sectors are identified in Annex 1 along with the 54occupational categories), industry and commoditylargely coincide; the divergence between the twonecessarily increases with progressive product disag-gregation. Practically speaking, any scheme is a com-promise between the two,** and it presents no par-ticular conceptual problem to disaggregate some partof the economy more than others.

For present purposes, let us consider the case ofrobots which may be produced for an establishment’sown use, for sale, or some combination of the two,In a commodity scheme, robots would constitute a sec-tor which is the sole source of (domestically produced)robots for all other sectors even, say, the automobilesector, although the automobile industry might pro-duce most of its own robots. In an industry scheme,the input structure of every sector that produces somerobots would need to be modified to reflect the pro-duction of robots in addition to the creation of aseparate robotics industry. The former is easier to im-plement (and is the approach chosen for this sector inthe IEA model even though most other sectors are clas-sified by industry). Both representations will lead tothe same conclusions as to the effects of the use ofrobots on the economy as a whole and as to the total(direct plus indirect) labor requirements of each sec-

‘An industry in this usage is the set of establishments that produce a givenprincipal output: they may in addition produce (the same or different) second-ary outputs.

● ● The sector including mills producing two slightly different grades of papercould be considered an industry with a secondary product. Alternatively,it could be considered a single commodity.

tor. They will differ in a well-defined way in the com-puted allocation between direct and indirect require-ments of individual sectors.

If the automobile sector produces all of its ownrobots, the labor used in producing these robots (thevector L) is charged directly to the automobile sector(still L) will be counted as indirect. The vector L itselfcan easily be computed given either representation andallocated in any desired proportions between what arereported as the direct and indirect labor requirementsof the automobile sector. Eventually it may be impor-tant to distinguish between the average input struc-ture of General Motors robots and say, those producedby Unimation for other users. In this case the singleproducing sector would have to be disaggregated intoseveral, producing different types of robots.

Data

The principal sources of data for the IEA model havebeen official government series produced by the Bu-reau of Economic Analysis (BEA) of the U.S. Depart-ment of Commerce (input-output tables and capitalflow tables for 1963, 1967, and 1972) and by BLS (in-dustry-occupation matrices for 1960, 1970, and 1978and industry price deflators). The sources of these dataare given in Annex 2.* We have relied also on manualand automated search of the open business and tech-nical literatures, reports of private research organiza-tions, and some interviews and informal surveys.

There is no substitute in terms of scope and com-prehensiveness for the official data. At the same timemany of their limitations are well known and havebeen described in various places. For example, a greatdeal of effort has been expended at our Institute torender the data tables compatible with respect toassumptions and classifications.

The official I-O and capital flow tables can be con-sidered accounting type data in terms both of theirbasic sources and the methods used in compiling them.Relatively little use is made of technical sector expertsin evaluating the input structures (i. e., table columns);this is one objective of the IEA case studies describedbelow.

There is no concrete concept in the official data ofoutput units for the so-called service sectors; conse-quently many of the associated price deflators arelargely arbitrary. ** This situation makes it hazardousto analyze the changing position of the non-goods-

‘Historical data have been assembled because the model will also be usedto examine the period from about 1960 to the present: they are not used toextrapolate the structure of the economy in the future from a time seriesdescription of the past.

● ’It might be pointed out in a discussion of price deflators that for otherreasons the official deflator for computers implies no change in real price overthe past two decades. We have instead assumed an average annual decreaseof 10 percent which more accurately reflects the already steep increase in theiruse.

7 2

producing sectors, which broadly defined alreadyemploy about two-thirds of the labor force, based onthese tables. In attempting to make some progress inthis area, we have concentrated first on the educationsector: it is a large employer, it is the source of train-ing and retraining, and it is easier to improve thanmost of the other service sectors. We have redefinedthe unit of output and disaggregated public educationfrom health care and nonprofit organizations in thecapital flow tables using fragmentary information fromgovernment studies and the other types of sources in-dicated above.

The capital flow tables compiled by BEA would beeven more useful if sector experts brought more in-formation to bear on the present methods of allocatinga particular capital good among using industries. Theofficial data on capital stock requirements by using in-dustry distinguish only between plant and equipment.A detailed breakdown by type of capital based notonly on deduction from flow data, but including directsurvey, should be a high priority objective. We haveconstructed very tentative stock matrices for the IEAmodel based on the capital flow tables.

The IEA model will make use, through its case stud-ies, of the available business and technical literatureand the data and reports of private organizations,which contain a wealth of information but require agreat deal of effort to piece together into a coherentpicture. An example of this source of information isthe “composite forecast” prepared in electronic andhard-copy form by Predicasts, Inc.: an illustrativetable describing robots in the U.S. market is shownin Annex 2. Not surprisingly a number of vendors ofdetailed economic information either use or have ex-pressed active interest in using I-O both as an organiz-ing device and to improve the consistency of a greatquantity of numbers taken from diverse sources.

Information for describing, for example, a robot sec-tor, which had little economic significance in the yearof the most recent official tables (1972), was piecedtogether from other sources including the items refer-enced in the Predicasts forecast mentioned earlier andinformal interviews and written surveys by our ownstaff of knowledgeable individuals willing to coop-erate, for example the president and the manager ofpersonnel of a major producer of robots. This type ofsurvey is also the basis of the so-called ex-ante I-Otables compiled by the Battelle Memorial Institute ina more formal and standardized fashion. *

● The Battelle tables are based mainly on surveys of engineers. We havefor various reasons not been able to make use of these tables for ongoingInstitute work but consider them a valuable source of information-and, moreimportant, a valuable methodology-for use in future work.

Case StudiesAny model that uses an I-O module has at least in-

directly to be concerned with the representation oftechnology and technological changes, but typicallythis concern is indeed indirect. The historical data areusually accepted as given in the official sources. Thecoefficients describing the structure of the economy inthe future are generally estimated by a formal statis-tical procedure with little effort at interpreting thetechnical structure implied by it. In most cases wherean I-O module is used to disaggregate the projectionswhich have been made by an aggregate econometricmodel, technological change is not at the center ofattention.

The case studies being carried out at IEA vary inscope and depth but have a common purpose: to eva]-uate and improve the corresponding portions of theavailable historical I-O type data (i.e., A, B, and Lmatrices) and to project these data to the future on thebasis of alternative assumptions about structuralchange in the sector under study.

Two of the case studies carried out at IEA on ro-botics and education have already been mentioned;others include the chemical, iron and steel, automobile,textile, and health care sectors, and telecommunica-tions, office equipment, and computers are in progress.The robotics study, for example, includes a compila-tion of current, capital, and labor input vectors forthe U.S. robotics producing sector circa 1990. In ad-dition, it projects the level of robot purchases for 45likely using sectors. In each using sector, seven occupa-tions in which workers are likely to be displaced byrobots and two which will be required in greater num-ber because of robots, are identified. The percentagechange in labor input requirements (for each occupa-tion in each using sector) is projected under alternativeassumptions. The quantitative output of the case studyis the direct input to the computer programs.

The Dynamic Model

The basic objective of an I-O analysis of the impactsof technological change on employment is, in opera-tional terms, to quantify the levels of detailed laborinputs which will be required in a given year to satisfyany given level and composition of final demand, mak-ing alternative assumptions about production technol-ogies (i.e., using alternative A matrices). In an openstatic model, the level and composition of final de-mand, which includes private and public consumptionand investment and net exports, are exogenous; theyare implicitly assumed to be relatively independent ofthe technologies in use. Almost all empirical I-O

7 3

analyses have been carried out within an open staticframework.

In a fully closed dynamic model, at the other ex-treme, the different components of final demand areexplained endogenously. For example, trade flows inthe IEA World I-O Model are parameterized by meansof region- and commodity-specific import coefficientsand export shares, and the levels of imports and ex-ports are thus computed endogenously. (In the openmodel one can—at least in principle—using alternativeassumptions concerning the exogenously determinedexports and imports, approximate changes in the struc-ture of foreign trade that are taken care of automatical-ly (i.e., endogenously) in the closed model. )

The present version of the IEA model represents asignificant advance in the endogenous determinationof investment. * In experiments carried out with thismodel, investment will be governed by a B matrix as-sociated (in terms of technological considerations) witheach A matrix. Exogenous final demand will excludeprivate fixed capital formation which will instead becomputed endogenously to satisfy in level and com-position the production of a particular final bill ofgoods. This means of course that the labor requiredto produce these investment goods will also be ap-propriately computed,

Scenarios

A scenario is an experiment carried out within theframework of the model to address a specific question.The starting point is a description of the I-O structureof the economy (i. e., A, B, and L matrices) for eachyear in the time horizon under examination, and eachscenario consists of specific alternative assumptionsabout how that structure (i. e., the individual coeffi-cients in the A, B, and L matrices) might evolve. Thesize of the labor force and its composition by occupa-tion as implied by each scenario are then computed(along with many other variables) and can be com-pared across scenarios. It is standard practice to definea baseline scenario for purposes of comparison. In theIEA model the baseline scenario assumes that the pres-ent I-O structure of the economy remains unchangedin the future.**

Scenarios may differ in their assumptions about vir-tually any aspect of the economic structure. I will brief-

● A model in which investment is endogenous can be considered a dynamicmodel. Many models have been called dynamic; in the dynamic J-O model,capital flows and stocks are disaggregated by producing and by using sec-tors, and the framework requires not only intersectoral but also intertem-poral consistency in the production and disposition of the highly disaggregatecapital stock.

‘ ● “Present” here means the most recent year for which official accountingdata are available; 1-0 and capital flow tables for 1977 should be availablein the near future,

ly discuss some possible scenarios about the rate ofadoption of automation, the identification of the oc-cupations affected and quantification of the impact onindividual labor coefficients, and the level and com-position of final demand, These are the types of sce-narios that can be most readily computed at the pres-ent time. As the model is progressively “closed” withrespect to other components of final demand, a sce-nario approach will also be used to represent alter-native assumptions about the constraints under whichthe economy operates.

In order to represent the use of programmable auto-mation, it is first necessary to distinguish in physicalterms the types of automation equipment in question,by what combination of existing and new sectors itwill be produced, and the input structure for produc-tion. Then, depending on the case, it is necessary toidentify the sectors and/or the operations in which thisequipment may be used and the likely level of use. Theoccupations of workers that may use both the auto-mated equipment and that which it complements orsupersedes must be specified, along with other affectedcurrent and capital inputs of the using sector, on a perunit of output basis. All the assumptions must ofcourse be quantified. A standard set of assumptionsfor 1990 is produced by the associated case study and“spliced” into the A, B, and L matrices for that yearand for prior years as required: the actual splicing con-sists of course of replacing equipment and the cor-responding input flows, including labor inputs, perunit of output of a particular sector by new equipment(e.g., robots) and input flows, including labor, re-quired (also per unit of output) by the new technol-ogy. (Note that the incremental output—and labor—required in prior years to produce the capital in placein May 1990 will automatically be computed. )

The “standard” assumptions referred to in the pre-ceding paragraph reflect the combination of what weconsider to be the moderate projections of experts deal-ing with different aspects of, say, robotics. Alternativescenarios incorporate more extreme views, eitherhypothetical or actually expressed by other analysts.(In later stages of our research, the portions of the casestudies dealing with the future will rely increasinglyon technical factfinding; this work will make possiblethe elaboration of more detailed, technical scenarios. )

Sample robotics scenarios for which preliminarycomputations have been made include:

• What will be the level and occupational composi-tion of employment in 1990 implied by the stand-ard assumptions of the robotics case study com-pared to the baseline scenario (assuming in bothcases the same projected 1990 final demand billof goods and the same state of the economy in1980)?

7 4

● What if one additional mechanic is required forevery three robots instead of every six robots ineach using sector?

● What if each using sector acquires twice as manyrobots as under the standard assumptions?

Scenarios will also be developed for office automa-tion involving changes in the use of telecommunica-tions equipment, computers, and office equipment byclerical and certain categories of managerial and pro-fessional employees in all sectors. Production automa-tion, involving essentially computers and sensors (aswell as robots), includes process control, computer-aided design and manufacturing, inventory control,and scheduling. While the latter case study is morecomplicated, the investigation will follow the same ap-proach. Scenarios will specify values for the amountof use and the impacts on individual I-O coefficientsof different forms of automation, singly and in com-bination.

The following type of scenario is included to givea concrete idea of the range of questions that can beaddressed in the near future with the present modeland expanded effort on the case studies:

• The different case studies deal with the introduc-tion of specific types of automation equipmentinto specific operations. Assume that all aspects(that have been identified) of production andoffice automation proceed at a “moderate” paceover the next 10 or 20 years. How will the sizeof the employed labor force evolve? Which oc-cupations will experience the slowest growth (orgreatest decline)? What will be the demand forcomputer programmers?

Policy and Research IssuesFor policy purposes it is necessary to have a model

that can produce results quickly and inexpensively.This can be achieved if:

1. there is a methodology for preparing at least cer-tain types of scenarios quickly by human ana-lysts, and

2. the computer software is designed (among otherconsiderations) to process an entire scenario asa single input.

The section on scenarios describes the types of sce-narios for which these two considerations can be metmost readily. The software design issues are well un-derstood.

The case study approach has been developed to pro-vide the required format for scenarios, and improv-ing the scope and depth of the case studies should bea high priority. In fact, “micro” studies now sponsoredby the policy community for more general factfindingpurposes could be guided to include a section struc-turing the information content into a format thatwould permit its being used directly by a model suchas the one described.

In subsequent work on technological change, theIEA model can and should be “closed” with respectto trade, accomplished by integrating it into the WorldModel. This will make it possible to analyze a con-siderably expanded range of policy scenarios whereU.S. production is directly linked to its trading part-ners’ economic activities. In the open system, these im-pacts are instead approximated by hypothesizingchanges in U.S. imports and exports.

From a research point of view, there are many ave-nues of work that would improve the accuracy of theprojections of the basic model (without even touchingfor present purposes on the important area of com-putational research).

Additional basic research is required to make fur-ther conceptual advances toward a fully dynamic I-Omodel in addition to the critical need for better dataon capital flows and especially capital stocks, on adisaggregated basis.

The other important area for further work is the in-corporation of a demographic model into the dynamicI-O model. This would provide the basis for improv-ing and rendering consistent the projected demand foreducation and health care as well as other componentsof household consumption.

7 5

THE EFFECT OF TECHNICAL CHANGE ON LABOR

by Louis Jacobson and Robert LevyThe Public Research Institute ofthe Center for Naval Analyses

July 27, 1982

Summary

Technical change, which makes possible the produc-tion of more output from a given amount of resources,is a major contributor to increases in society’s well-being. The very process of change may, however, im-pose hardships on those who use old and no longerefficient methods or produce products that are nolonger wanted. The net effect of technical change onlabor is hard to predict. Workers, as consumers, gainfrom increases in productivity; they are able to buythings at lower prices. If they cannot adapt to new pro-duction methods (and lose their jobs as a result), theycan end up as net losers,

Not only is the effect of technical change on laborhard to predict, but technical change itself is hard todefine and measure. It cannot be measured simply asthe installation of new (and different) equipment. Norcan it be measured solely as the growth of productivi-ty. All three—technical change, new equipment, andproductivity growth—are related. A goal of this paperis to clarify the relationship.

It is important, at the start, to distinguish technicalchange resulting in the appearance of new productsfrom a change in production processes. To somedegree, this distinction is artificial; programmableautomation equipment is a new product of industriesproducing capital goods, but its use represents a newprocess in other industries. Both new products and newprocesses can affect labor. The development of a newproduct —passenger aircraft—and new processes—as-sembly-line methods of production of automobiles—led to increases in employment in those and a host ofrelated industries—home construction, leisure prod-ucts, etc. At the same time, workers in other indus-tries—horse-breeding, saddle-making, the rail pas-senger industry-were being squeezed out into otherjobs. Now, changes in production technology, hereand abroad, are leading to a reduction in employmentin auto production and autoworkers are having toadjust.

We analyzed the effects of technical change onworkers in two steps. The first step is to estimate theeffect of technical change on an industry’s employ-ment. We concentrate on the effect of new process in-novations since these are likely to have more concen-

trated impacts on employment. The next step is toestimate the effect on individual workers: How manyworkers are displaced during the adjustment process?How long does it take them to find new employment?How much does displacement affect their earnings?

The Industry Cost Model

To determine industry effects, we use an econo-metric model of cost in which cost is assumed to de-pend on the prices of factor inputs, the level of industryoutput, and the level of technology. This is the stand-ard analytic approach in economic studies of technicalchange. The rate of growth of total cost is composedof the weighted average of the rates of growth of in-put prices, a weighted rate of growth of industry out-put, and the rate of cost reduction due to technicalchange. Technical change is therefore only one deter-minant affecting industry cost; input price changes, forexample, may be more important.

We included five factor inputs—production labor,nonproduction labor, capital, energy, and materialsin the cost function and imposed no restrictions onhow they can substitute for each other. We assumedcost minimization and estimated how technical changehas affected the steel, auto, and aluminum industries.The model was estimated at the four-digit StandardIndustrial Classification (SIC) category, a finer levelof disaggregation than that used in most industrystudies.

One of the advantages of using this kind of modelis that with relatively few variables, much can belearned about factor substitution, input adjustment,and the effects of technical change. Data on inputquantities and prices, industry cost and real output,and the level of technology are almost all that isneeded. Data needed to measure capital and technol-ogy is often difficuIt to find, but the model clearlyspecifies what is needed for estimation.

Equations in the econometric system illustrate im-portant relationships. For example, equations that ex-press the dependence of each input’s share in total costson all input prices, output and technology maybe usedto describe how input-output coefficients, which aremeasured similarly to input shares, move over time.More important, the equation also describes why they

7 6

move. It explains changes in input-output coefficientsin terms of: 1) adoption of new techniques, 2) changesin input prices, and 3) changes in scale.

For labor, the equation describes how changes inlabor’s share (i.e., the ratio of payroll to total cost)depend on changes in the wage, other input prices, out-put, and the level of technology. The effect of anyvariable on labor’s share is obtained from the regres-sion estimates. For example, the effects of technicalchange, holding prices and output constant, is de-scribed as labor saving, using, or neutral if theparameter on technology is negative, positive, or zero.Labor-saving technical change means that as technol-ogy increases, the share of payroll in total costs goesdown. The estimated equation can then be used toderive the effect of technical change on the demandfor the quantity of labor.

Another equation in the system maybe used to ex-plain the relationship between the rates of technicalchange and productivity growth. Productivity growthis measured in terms of total factor productivity ratherthan labor productivity, which is only a partial meas-ure of input use.

The rate of productivity growth has two compo-nents. The first is the rate of technical change. The sec-ond depends on the relationship between changes inindustry cost and output changes. Most researchersassume that, in longrun equilibrium, industry cost andoutput change proportionately. This assumption,called “constant returns to scale, ” is often imposed onthe equations. If this assumption is correct, then thissecond component of productivity growth becomesequal to zero, making the rates of productivity growthand technical change equivalent. Thus, low rates oftechnical change would imply low rates of productivitygrowth.

We feel, however, that constant returns to scale istoo strong an assumption. Certain factor inputs, likecapital and nonproduction labor, maybe “fixed;” theycannot be adjusted quickly without incurring largecost. Because of these fixed inputs, costs do not changeas much as output. This means that studies that sim-ply impose constant returns to scale, thereby disre-garding the second component of productivity growth,will overstate the effects of technical change.

The Measurement of Industry Technology

Thus far, we have illustrated that technical changeis only one determinant of industry cost changes andone component of productivity growth. We have notyet related technical change to the introduction of newtechnologies.

To do this in an econometric model required meas-urement of the level of industry technology. The stand-

ard approach is to represent the level by a time trend.This is satisfactory if changes in technology unfoldregularly and gradually. It is unsatisfactory if newprocesses are introduced rapidly, that is, within a rel-atively short period of time. This distinction is impor-tant since sudden or unexpected shifts in productionprocesses and labor demand may make adjustment dif-ficult for the industry’s work force.

To be as precise as possible, we therefore con-structed direct measures of steel and auto technology.(A measure for aluminum could not be constructed.)For steel, the measure was based on the use of the basicoxygen furnace, which is important in the steelmak-ing process. Technological innovation in the auto in-dustry since World War II has proceeded under theterm “Detroit” automation. The term refers generallyto the substitution of machines for workers in actualproduction processes, such as welding. To quantify theconcept of automation, we measure the stock of trans-fer machines, the basic unit of Detroit automation.

Empirical Results of the Cost Model

In our empirical work on steel and autos, we com-pared the precise “direct” measure with the simplertime trend. The findings were about the same; the useof the direct measure added little precision to theestimates of the effect of technical change beyond whatwe estimated using the time trend. Regardless of howtechnology was measured in steel and autos, estimatesof the rate of technical change in the industry weresimilar. We found the average rate of technical changeto be virtually O percent in steel. It was just under 1percent in aluminum, where little new process innova-tion has been observed, and 1.50 and 2 percent in autos(the higher figure was for the time trend version).These findings illustrate the tenuous connection be-tween new process innovation and technical change.

The effects on input demand are similar across in-dustries when the time trend is used to represent tech-nology. We found that technical change was labor sav-ing and capital using. This meant that, holding inputprices and output constant, labor’s share was decreas-ing over time while capital’s share was increasing byabout the same amount. In autos, in the 1970’s, labordemand decreased by about 4 to 5 percent a year.

When direct measure was used, the results for steelwere about the same as when the time trend was used.For autos, there were some differences from the timetrend regression. There were smaller negative effectson labor and smaller positive effects on capital. Onetentative interpretation of this is that advances intechnology meant that newer capital was more pro-ductive than the capital it replaced and that it usedonly slightly less labor.

77

In summary, we found that the effects on employ-ment of introducing new technologies, whether meas-ured directly or by a time trend, were indeed negativebut occurred gradually. We did not find evidence ofabrupt changes in the demand for labor. Some of theempirical estimates differed according to how technol-ogy was measured, but the implications for employ-ment adjustment were essentially similar. Given thesesimilarities, the time and expense of creating directmeasures seem unnecessary. The exploration of otherissues may prove more fruitful in determining hownew technologies affect employment.

As an example, our findings point to an interestingimplication of the generally labor-saving and capital-using effects of technical change. When the price oflabor increases relative to capital, little short-termsubstitution of capital for labor takes place. Over thelong term, however, the use of advanced technologiesallows capital to be less labor intensive and so thequantity of labor decreases relative to the quantity ofcapital. This possible “induced innovation” has beendifficult to identify empirically. Generalizations of ourmodel may help quantify the link between new tech-nologies and their determinants.

The Effect of Employment Reductionson Workers

We have taken our analysis further than simply ex-amining the effects of technical change on aggregateemployment. We have also examined:

• the extent to which employment reductions canbe accommodated through attrition rather thandisplacement, and

• the earnings losses of displaced workers.

Displacement Findings

Our results are that attrition varies widely acrossindustries, from about 5 percent per year in high-wageindustries (steel) to about 65 percent in low-wage in-dustries. If high-wage, low-turnover industries, suchas steel and autos, had to reduce employment by 5 per-cent in a given year, about two-thirds of the reduc-tion could be accommodated by attrition under aver-age conditions. If employment had been increasingprior to the reduction, attrition would be higher, thatis, more than two-thirds would be attrition. If employ-ment had been falling, attrition would be a lower pro-portion. This is because attrition is primarily deter-mined by the tenure structure of the industry. Attri-tion, which is extremely high among recently hiredworkers, falls dramatically after a year or two, and

is extremely low thereafter until workers near retire-ment age. Recent hiring and layoff patterns cause largeswings in the proportion of the work force that is mostlikely to leave voluntarily, and this, in turn, causesswings in the attrition rate by as much as 2 percent.

Aside from their effect on the tenure distribution,general business conditions (the business cycle) havelittle effe6t on attrition. Although other studies showthat the quit rate is sensitive to changes in business con-ditions (and quits are the major element of attrition),those studies generally ignore the changes in the tenuredistribution over the business cycle (use imperfectmeasures of tenure structure). Instead, they rely on ag-gregate turnover statistics that show a fall in quits dur-ing recession simply because workers who would other-wise quit are laid off.

A 5 percent employment decline cannot be fully offsetby a 5 percent attrition rate. There will be some dis-placements because the employment change across firmsis not uniform. Even while total employment in the in-dustry is declining some firms will be expanding. Attri-tion in the firms that are expanding must be replaced byhiring and obviously cannot count against the net decline.The dispersion of firms around the mean employmentchange is substantial, so that even where employment inthe industry is constant, about 1 percent of the employ-ment industry’s labor force will be undergoing displace-ment.

The existence of a more-or-less constant backgroundlevel of displacement is important. Displacements aremost costly when they are entirely unexpected. If workersanticipate a nonzero probability of displacement, theycan prepare for that eventuality and reduce its conse-quences. This appears to be the case in some high-wageindustries such as aerospace, which have major boom andbust cycles as a result of military and civilian aviationprocurement policies.

Plant closings are a major source of displacements be-cause attrition can do little to offset employment declineswhen plants close. Closings are particularly likely whena firm experiences a sharp decline in demand: below acertain level of production, it is simply uneconomic tostay in business.

Earnings Loss Findings

Earnings losses due to displacement are highly cor-related with attrition rates—high losses go with lowattrition. This makes good economic sense; if fewworkers are leaving voluntarily, this is strong evidencethat there are few good alternative jobs, and a workerwho is forced to leave will have large losses. Displacedworkers experience a transitional period of earningslosses due to lengthy unemployment (a good part of

7 8

which may be simply waiting until all hope of recallvanishes*).

When workers begin to search for work, they oftenwill try out several jobs until they find suitableemployment. Workers’ earnings then begin to reboundbut usually, in cases where initial losses are high, theynever fully catch up with what they would have earnedhad they not been displaced. The “permanent” loss isgenerally between 7 and 15 percent of annual earn-ings and amounts to about 50 percent of the total earn-ings loss. (The total loss in industries such as steel andautos is generally equal to about 2 year’s earnings. )The transitional loss is about equal to the permanentloss, but it lasts only a year or two. The temporaryloss, however, is more likely to be offset by unemploy-ment insurance, SUB, and severance pay.

In analyzing earnings losses, we measured how theyvary with respect to age, tenure, labor market char-acteristics, and the size of the employment decline.Again, tenure was a key determinant of the size of theloss. New hires and workers close to retirement ex-perience small losses, while other workers tend to havelarge losses. The largest losses occur among workerswith about 5 years experience. This is because theyhave many years over which to accumulate the per-manent part of the loss, and because they may be ina position to assume more responsible positions butdo not yet have the experience needed to convinceanother employer that they are ready for such a job.

Labor market characteristics are also important.Loss in the transitional period can be doubled ifdisplacement occurs when area unemployment is high(one standard deviation over average–about 1.3 per-centage points). Losses are substantially larger if thedisplacement has occurred in a small labor market andmarginally greater if a large number of similar workersare searching for work. Workers who have been dis-placed because of a plant shutdown do not have muchlarger losses than similar workers displaced underother circumstances. A shutdown will, however, dis-place higher tenure workers who have larger thanaverage losses.

Methodology

This section discusses, in general terms, the meth-odology used to obtain the above results. The detailsare discussed in the reports referenced at the end ofthis paper. The basic estimating equations used to ex-amine attrition and earnings loss are quite simple. At-trition is assumed to be a function of worker charac-

● A large fraction of workers laid off eventually are recalled. In fact, tem-porary unemployment is the largest component in the cost of an employ-ment reduction, but the cost is low per capita, and industrial workers generallyanticipate several episodes of substantial temporary unemployment.

teristics-age, race, sex, tenure, earnings level, and therate of change of earnings; plant characteristics—em-ployment level (size), employment trends, whetherpart of a multiplant firm, average wage rate, and trendof wages; and labor market conditions—stage of thebusiness cycle (current unemployment rate divided bythe rate at the previous trough), size of the labor mar-ket, recent growth rate, average wage rate, and meas-ures of industrial diversity.

Earnings are estimated as a function of many of thesame variables used to measure attrition. Prior earn-ings, however, is the key determinant, since it capturesthe way in which a host of determinants, which re-main more or less constant over time (such as educa-tion, health, and marital status) affect a worker’s ear-nings potential. The earnings equations are designed tocompare the earnings of displaced workers to thoseof similar workers who are not displaced but are ini-tially employed in the same industry. Econometric testsare applied to ensure that the equation matches theearnings of the two groups exactly.

The procedure for estimating how employmentchanges are distributed across firms in an industry ismore complex. We assume that units use a strict LIFO(last-in-first-out) displacement rule. Given this assump-tion, the seniority distribution of displaced workersis also the cumulative distribution of employmentchanges across all the individual units. The actualseniority distribution is calculated by observing thetenure of displaced workers and relating tenure toseniority. This procedure is used to examine displace-ment under a range of circumstances and then econo-metric techniques are applied to generalize the results.

Data

All three estimating procedures are carried out withthe same data set—Social Security’s Longitudinal Em-ployer-Employee Data (LEED). These data contain theage, race, and sex of a continuing 1 percent sampleof the work force, each individual worker’s quarterlyearnings from each employer, and the employer’s in-dustry, location, and firm size. The data cover theyears 1957 through 1975. Job changes are noted by achange in the employer who files a quarterly earningsreport for the worker.

A key problem is that there is no explicit indicatorof the reason for separation in the data. Thus, we mustinfer whether a worker who changed jobs was dis-placed or left voluntarily (attrition). To measure at-trition, we isolated a sample of workers who changedjobs when employment in their plant (or sometimesarea) was increasing or stable. We assume that undersuch circumstances, no displacements were takingplace. The earnings loss due to displacement is esti-

7 9

mated using a simultaneous equation approach thatcompares the earnings loss of leavers in firms whereemployment is growing or stable to the loss of leaversin firms where employment is falling. It is thus possi-ble to net out the effect of attrition from the effect ofdisplacement. In practice, it turns out that holding con-stant the other factors mentioned earlier, earnings re-ductions are relatively insensitive to whether displace-ment or attrition was the reason for the job change;both the displaced and voluntary leavers have similarearnings reductions.

The LEED file is not the only data base suitable forestimating earnings losses. We have also used datafrom State employment security agencies (UI offices).These data include wage records that closely resem-ble those in the LEED file, but they also include infor-mation on the worker’s receipt of unemployment com-pensation, last day worked, occupation (in somecases), and the reason for separation (if the workerclaims UI). Thus, these data contain a richer set ofvariables. A major value of these data is examiningthe extent to which transitional losses are offset by thereceipt of UI. For example, we recently used data ofthe type to study TAA (Trade Adjustment Assistance)recipients in Pennsylvania and concluded that morethan 50 percent of the transitional loss in earnings isreplaced by UI benefits.

Summary and Conclusion

Our research on the effects of technical change onemployment has been aimed at measuring technicalchange and its effects on labor in the steel, auto, andrelated industries. State-of-the-art econometric modelswere used to determine the effect of technical changeon employment and to measure how displaced workersadjust.

Our basic conclusion about the effect of technicalchange on employment is that, while technical changewas labor saving, it occurred at a steady and relative-ly slow pace. It appears that new technologies wereadopted principally to avoid increases in labor costs.

Detailed description and measurement of the instal-lation of new technologies, such as transfer machinesin autos and the basic oxygen furnace in steel, seemsno better than the time trend in measuring how tech-nical change affects employment. This reinforces ourconclusion that technical change in established in-dustries (process innovation) is slow and steady.

We also linked technical change in one industry toits effect through factor prices on the other industries.This allows comprehensive measurement of the netemployment effect of innovation. Although technicalchange in steel reduced employment in the steel in-

dustry, the change in steel increased employment inautos by reducing the price of a key input.

Our models have been carefully checked and testedfor sensitivity to key assumptions. They represent agood way to measure the effect of technical change;the same methodology can successfully be used infuture research. In particular, we believe tracing theeffects of individual technologies is not worth the highcost. The effects of changes in consumer preferences,factor prices, and foreign competition on labor de-mand are likely to be far more important and canchange far more swiftly than production technology,as we learned in the 1973 oil embargo. One fruitfulextension of this approach would be to investigate thedeterminants of technical change itself, such as the ex-tent changes in factor prices induce factor-saving tech-nical change.

We have also analyzed how employment reductionsaffect workers employed in a given industry. Employ-ment reductions are costly to labor primarily whenworkers are displaced from established jobs, but reduc-tions that do not result in displacing workers arealmost costless. The work, therefore, addressed howemployment reductions are distributed across plantsand how much of the reductions can be accommodatedby attrition. The cost to displaced workers was alsoassessed.

Our basic conclusion is that employment reductionscan be handled largely through attrition. The mainthreat of dislocation is from plant closings. Althoughmajor plant closings occur rarely, we calculated thatthey are responsible for half the displacements in thesteel industry. Our work on earning losses showed thatwhen potential losses are large, the attrition rate is low.Thus, in industries where few workers leave volun-tarily, such as steel and autos, major employmentreductions are likely to be very costly and can involvesubstantial displacement. On the other hand, employ-ment reduction in high turnover industries, such as tex-tiles and apparel, have minimal adverse effects onworkers.

In terms of future research, further work on attri-tion is called for. We found that attrition is largely afunction of the tenure structure in an industry, butother research suggests that cyclical conditions are akey determinant of attrition. Although these two ex-planations are not necessarily contradictory (the tenurestructure changes as a result of cyclical hiring andlayoffs), it is important to determine whether badbusiness conditions reduce attrition holding tenurestructure constant, If, contrary to our evidence, attri-tion falls in a recession because general business con-ditions worsen, we will have overestimated an indus-try’s ability to adjust to employment reduction withoutdisplacing many workers.

80

A second objective of future research is to explainwhy workers, in industries such as steel and autos, suf-fer such large losses and what government can doabout it. There is a widespread feeling that these lossesare an indication of a failure in the functioning of thelabor market. In the extreme form, it is assumed thatdisplaced workers “fall off the end of the earth” andnever adjust. Competing explanations are that theearning reductions represent normal job search costsplus loss of high wages attributable to unionization orspecific human capital.

If market failure is responsible, an effort should bemade to determine its cause. If specific failures can beisolated, there is a reasonable chance that governmentaction can help eliminate them at low cost. If loss ofunion protection or human capital are involved, pol-icymakers should recognize the constraints thisimplies.

Our evidence is that where losses are large, displacedworkers adapt to new jobs but never regain the earn-ings level they would have attained had they not losttheir jobs. The adjustment is painful; often 5O percentof the loss occurs in the first 2 years. The “permanent”loss is about 7 to 15 percent, about that estimated forunion/nonunion wage differentials. This evidence isconsistent with the hypothesis that losses are due toloss of union protection or specific human capital. Ifthis view is correct, a policy that will fully eliminate

the loss must raise the income of an experienced work-er about $2,500 a year or about $40,000 in presentvalue terms. Thus, any government actions to elimi-nate losses such as training or enhancing unemploy-ment insurance will be expensive.

References

1.

2.

3.

4,

Jacobson, Louis S., “Earnings Losses and WorkerDisplacement When Employment Falls in the SteelIndustry” (unpublished Ph.D. dissertation, North-western University, August 1977).Jacobson, Louis S., “Earnings Losses of WorkersDisplaced From Manufacturing Industries,” in TheImpact of International Trade and Investment onEmployment, William Dewald (cd. ) (Washington,D. C.: U.S. Department of Labor, Bureau of Inter-national Affairs, 1978).Jacobson, Louis S., and Jondrow, James, and Levy,Robert, “Labor Adjustment to Technical Change:A Multi-Industry Study” (progress report to the Na-tional Science Foundation, 1980).Jondrow, James, Levy, Robert, and Jacobson,Louis, “Labor Adjustment to Technical Change inSteel and Related Industries” (paper presented atAmerican Economic Association meetings held inWashington, D. C., December 1981).

PROGRAMMABLE AUTOMATION: ITS EFFECT ON THESCIENTIFIC= ENGINEERING LABOR MARKETby William N. CookeKrannert Graduate School of Management, Purdue University

July 27, 1982

Adjusting to TechnologicalAdvancement

Introduction: An Analytical Framework

Technological change and innovation play a pre-dominant role in productivity and general economicgrowth. They also create shifts and movements in thesupply and demand for labor. In particular, they arepresumed to require the labor force to adjust or re-spond to demands for alternative knowledge and skills.Generally, the theoretical construct for determining thesupply and demand for labor (or the impact of tech-nological change on shifts in supply and demand oflabor) is the production function (6). This function

(whether Cobb-Douglas or C. E. S.) expresses how thefinal product (output) is a function of inputs (forsimplicity, capital, and labor) and how the inputs arerelated. For example, the marginal product producedby capital or labor is derived by differentiating the pro-duction function by either input. As either increasesits marginal product (ceteris paribus), the demand forits services increases over the other.

The relationship of inputs to outputs, and inputs toinputs are influenced heavily, however, by the existingtechnology. Technological change, in turn, alters theserelationships in several important ways—includingchanges in: 1) the efficiency of technology, 2) econ-omies of scale, 3) capital intensity, and 4) elasticityof substitution. Moreover, technology change either

8 1

is neutral or nonneutral. Neutral changes in technol-ogy affect technological efficiency and technologicallydetermined economies of scale. These alter the rela-tionship of inputs to outputs. Nonneutral changes, onthe other hand, affect the intensity of technology be-tween inputs. An increase in capital intensity is laborsaving when capital is a more rapidly growing factorof production than labor. “Many economists have thefeeling that technological change has been quite labor-saving, but they generally acknowledge that the evi-dence is indirect and too weak to permit a clear-cutjudgment” (12).

A further dimension to the production functionanalysis is found in what has been termed embodiedand disembodied technological change (17). Theformer generally refers to actual physical changes inthe capital equipment being operated and assumes thatit becomes more productive. The latter concept isusually characterized by managerial or organizationalchanges which cause the interaction of capital andlabor to be more productive. Generally speaking theanalysis of programmable automation can be treatedas embodied technological change. In summary, theproduction function framework suggests that em-bodied technological change raises the marginal pro-ductivity of capital relative to labor. It does thisthrough increasing the intensity of technology and theelasticity of substitution. Thus, everything else con-stant, the demand for labor shifts. For the most part,such shifts cause the substitution of capital for labor,and/or one type of labor for another.

Postschool Education, On-the-Job Training,and Turnover

Although the production function offers a usefulframework for conceptualizing the impacts of tech-nological change on the demand for labor, our pres-ent interests require us to go beyond this framework.We seek a framework that explains the type and levelof human capital adjustments made by labor as thedemand curve for specific skills is shifted by embodiedtechnological advances. The widely applied theory ofinvestment in human capital lends itself well to eco-nomic choice modeling and to an analysis of earnings.The theory takes a labor supply perspective. Workersmake decisions about their labor force behavior inresponse to the derived demand for alternative skillsor knowledge. Thus, demand conditions are given andthe supply of workers responds. Human capital theorymaintains that worker response is in the form of in-come maximization over the working life. Since de-rived demand is dynamic, labor supply is dynamic.What may have been a good decision previously maybecome a poor one later if the worker does not adjust

to changes in demand. According to human capitaltheory the incentive to adjust (or not adjust) primari-ly is pecuniary. In effect, the theory holds that a per-son makes an investment of resources in upgrading hisor her productivity. Couched in the marginal produc-tivity thesis, workers get paid the value of their pro-ductivity. Consequently, increases in productivity areassumed to lead to increases in income. Since the in-vestment of resources (primarily foregone earnings) iscostly, individuals weigh the present value of the in-creased flow of future income against the present valueof the cost of the investment. Income maximizing in-dividuals will make investments in those alternativeswith the highest rates of return.

The theory is an exact interpretation of very inexactreasoning on the part of individuals. The theory hasreceived much criticism, especially for its simplicityof the “economic man” concept (ignoring the non-pecuniary incentives of selecting occupations or fur-thering education—at least in empirical work). How-ever, the concept of maximizing utility readily fits intothe investment framework. It is the measurement ofnonpecuniary returns that eludes the researcher. Thetheoretical framework is a useful one for the presentanalysis because it explains behavior as a function ofprice (cost and earnings) and it encompasses formaleducational training (both early and later in thecareer), on-the-job training (OJT), * quit behavior, andoccupational changes.

Although not fully developed at this point (norwidely embraced by economists) “implicit contract”theory promises to improve our understanding of labormarket adjustments (1,2,3,8,9,11). The implicit con-

tract framework addresses better the employer’s partin the human capital investment decisionmaking. Inparticular, it addresses the decision of the firm to pro-vide OJT and to influence turnover. Succinctly, be-cause workers are more risk averse vis a vis employers,employers are willing to provide more stable employ-ment and less risky OJT investments in exchange forlower wages. Without developing the theory morefully here, let me emphasize that an examination oflabor market adjustments to technological changeshould consider employer as well as worker decision-making.

What kind of alternative choices do workers andemployers make when rapid advancements in technicalknowledge and skills lead to obsolescence? For scien-tists and engineers (S/Es), these choices include readingprofessional and trade literature, participating in work-shops and conferences, returning to school (part- orfull-time) (10), investing in formal and/or informal

● Editor’s note: OJT is more broadly defined in this paper than in ch. 3of the text where it refers only to informal skill acquisition.

82

OJT, and changing jobs. Changing jobs includes bothchanging employers and/or changing occupations(either within or out of S/E occupations). Employers,likewise, have alternatives for retooling their workforces experiencing obsolescence. Primarily, these op-tions include providing informal and formal OJT, re-assignment of job tasks, and laying off workers withobsolete human capital (and, in turn, recruiting newtalent).

Investment decisions made to replace obsoletehuman capital probably are influenced most stronglyby age, which affects investment behavior in severalways. In its simplest form the relationship between ageand investments is considered dependent on the timehorizon for receiving returns. Older persons simply in-vest less because they have fewer years to receive in-vestment returns. For the same reason, employers re-strict opportunities for older workers. Secondly, in-vestments will decrease at an increasing rate with ageif older workers become less efficient in producinghuman capital. Ben-Porath addresses this possibilityin testing his “neutrality hypothesis. ” The neutralityhypothesis assumes that the efficiency of human cap-ital production and market production remains con-stant throughout the career. Any shift in efficiencytowards the production of goods and services in theproduct market reduces the likelihood of producingfurther human capital.

A third dimension to the relationship between ageand investment probabilities is deterioration. Likephysical capital, human capital can deteriorate overtime as a result of obsolescence and/or depreciation.According to Rosen (16), obsolescence indicates thatmore vintaged knowledge has become outdated bymore recent knowledge. This can be caused by ad-vancements in the sciences, production innovationswhich render existing skills useless, and the “increas-ing abilities of successive generations. ” Depreciation,on the other hand, indicates a loss of ability by in-dividuals to apply their existing knowledge and skills.This presumably goes hand in hand with age as physi-cal and mental capacities diminish.

Although most modeling treats obsolescence and de-preciation as one (usually a depreciation factor), thedistinction between the two is especially important tothe study of S/Es where technological and scientificadvances make relatively specialized skills obsoletemore quickly. To keep abreast of these advancementsand to at least protect themselves in the labor market,scientists make further investments in their humancapital stock. In contrast, depreciation reduces thecapacity to apply one’s stock of human capital to pro-duction, which reduces the rates of return to furtherinvestments, and, thus reduces investments.

Other important determinants of investment behav-ior for S/Es include education level, labor market con-ditions, years of experience, tenure with presentemployer, sex, and government contracting. When thedeterminants of investment behavior cause disincen-tives to retooling obsolete skills, S/Es are more likelyto change employers, occupations, industries, and in-cur spells of involuntary unemployment.

Adjustments by Occupation and Industry

Below are several tables providing frequencies onthe incidence of various adjustments. The figurespresented in these tables are based on the NationalScience Foundation’s (NSF) Longitudinal Survey ofScientists and Engineers (NLSSE). There is no attempthere to tie these adjustments to technological change.(See ref. 7 for a more in-depth description and analysisof these adjustments. )

Table C-1 provides observations on the frequencyof investments in postschool education and OJT byS/Es. Occupational classifications are based on therespondent’s self-concept.

Roughly 6 percent of the scientists and 4 percent ofthe engineers made formal educational investments inthe same field, whereas about 10 percent of bothsamples invested in other fields. Considerably moreinvestments were made in OJT and in course work pro-vided at the employer’s training school: 19 percent ofthe scientists and 18 percent of the engineers investedin OJT and 16 percent of the scientists and 20 percentof the engineers took courses at the employer’s train-ing facility.

Table C-2 provides figures on the percent of S/Esundertaking OJT by selected manufacturing industry.The incidence of formal OJT ranges from a low of 15percent of the S/E work force in the fabricated metalsindustry to a high of 36 percent in the electronic com-puter industry. The incidence of reported informal OJTranges from a low of approximately 16 percent in theaircraft industry to a high of 36 percent in the elec-tronics computer industry.

Table C-3 provides observations on the frequenciesof voluntary and involuntary turnover by S/Es overthe 1969-72 period. Quit frequencies indicate that agri-cultural scientists were the least likely to quit duringthis period (7 percent) and that physicists had the high-est quit percentage (15 percent). Among engineers,aeronautical/astronautical engineers had the highestfrequency (5 percent) and agricultural scientists hadthe lowest frequency (0.4 percent). Among engineers,aeronautical/astronautical engineers had the highestpermanent layoff incidence (8 percent) and mining/

8 3

Table C-1 .–Frequencies of Postschool Education and On-the-Job Training byScientific and Engineering Occupation, 1972-74

Postschool education On-the-job training

Occupation Same field Other field Formal Informal

Scientists:Agricultural . . . . . . . . . . . . . . . . . . . . . . 0.035 0.055 0,242 0.315Biological . . . . . . . . . . . . . . . . . . . . . . . . 0.048 0.123 0.055 0.141Chemist . . . . . . . . . . . . . . . . . . . . . . . . . 0.064 0.070 0.130 0.158Earth/marine . . . . . . . . . . . . . . . . . . . . . 0.052 0.054 0.212 0.196Physicist . . . . . . . . . . . . . . . . . . . . . . . . 0.088 0.075 0.076 0.087

Engineers:Aeronautical/astronautical. . . . . . . . . . 0.052 0.133 0.217 0.156Agricultural . . . . . . . . . . . . . . . . . . . . . . 0.055 0.055 0.273 0.164Chemical . . . . . . . . . . . . . . . . . . . . . . . . 0.049 0.090 0.184 0.210Civil/architectural . . . . . . . . . . . . . . . . . 0.043 0.084 0.153 0.186Electrical/electronic . . . . . . . . . . . . . . . 0.063 0.106 0.240 0.176Industrial . . . . . . . . . . . . . . . . . . . . . . . . 0.012 0.129 0.226 0.219Mechanical . . . . . . . . . . . . . . . . . . . . . . 0.032 0.095 0.189 0.163Metallurgical/materials. . . . . . . . . . . . . 0.041 0.090 0.178 0.134Mining/petroleum . . . . . . . . . . . . . . . . . 0.023 0.062 0.229 0.209

Table C-2.—Formal and Informal On-the-Job TrainingFrequencies by Selected Major industries, 1972.73

(working full-time year round in same job)

Percent of Percent ofIndustry formal OJT informal OJT

Aircraft a. . . . . . . . . . . . . . . . . . . . 25.4% 15.50/0Chemicals b . . . . . . . . . . . . . . . . . 22.2 23.6Electrical machinery . . . . . . . . 30.1 19.7Electronic apparatus . . . . . . . . 28.0 17.6Electronic computerese . . . . . . . 46.3 35.5Fabricated metalsf. . . . . . . . . . . 15.3 17.3Machinery (except electrical)g . 18.6 17.4Motor vehiclesh . . . . . . . . . . . . . 31.2 21.8Ordinance i. . . . . . . . . . . . . . . . . . 31.2 19.3aAirCraft, aircraft engines, and Parts.bchemicals and allied products.cEle~trical machinew, qu~pfllent and supplies for the generation, storage, trans-

formation, transmission, and utilization of electrical energy.dEIKtronic apparatus, r~io, television, and communication equipment and partS.OElectronic computers, accounting, calculating and office machinery and equip-

ment.fabricated metal products (except ordnance, machinew ~d transportationequipment.

gMachinery (except electrical) including engines and turbines, farming and COrl-

struction machinery, mining, metalworking, and other manufacturing and serviceindustry machines.

hMotor vehicles and motor vehicle equipment including trucks, buses, automo-

biles, railroad engines and cars.iordnance, including manufacture of arms, ammunition, tanks, and comPleteguided missiles, space vehicles and equipment.

petroleum engineers had the lowest incidence (2percent).

Table C-4 reports the percentage of engineers chang-ing occupations during the 1969-72 period. Within theselected sample, 3.6 percent of engineers changed oc-cupations, and about 54 percent of them shifted intoother engineering occupations. The greatest exoduswas out of aeronautical/astronautical engineering (6.7percent) and the greatest entrance was into mechanical

Table C-3.—Frequencies by Voluntary and involuntaryMobility by Scientific and Engineering

Occupations, 1969-72

PercentOccupation Percent quit Iaidoff

Scientists:Agricultural . . . . . . . . . . . . . . . . . . 0.0680/0 0.0040/0Biological . . . . . . . . . . . . . . . . . . . . 0.124 0.018Chemist . . . . . . . . . . . . . . . . . . . . . 0.095 0.042Earth/marine . . . . . . . . . . . . . . . . . 0.122 0.034Physicist . . . . . . . . . . . . . . . . . . . . 0.147 0.053

Engineers:Aeronautical/astronautical. . . . . . 0.075 0.081Chemical . . . . . . . . . . . . . . . . . . . . 0.093 0.049Civil/architectural . . . . . . . . . . . . . 0.179 0.034Eiectricai/electronic . . . . . . . . . . . 0.143 0.074industrial . . . . . . . . . . . . . . . . . . . . 0.157 0.066Mechanical . . . . . . . . . . . . . . . . . . 0.130 0.068Metallurgical/materials. . . . . . . . . 0.083 0.057Mining/petroleum . . . . . . . . . . . . . 0.158 0.022

engineering (9.6 percent of those changing occupa-tions). Of all occupation leavers, 73 percent movedinto what appears to be equivalent engineering orscientific occupations. Another 6 percent shifteddownward to technician jobs, and the remaindershifted into a wide mix of occupations ranging fromsecondary teachers to laborers.

An examination of the incidence of various labormarket adjustments indicates that OJT (both formaland informal) is more widely experienced across S/Eoccupations than any other type of market adjustment.Although to some degree I am comparing apples tooranges, the above figures are consistent with the no-

84

Table C-4.—Frequencies of Occupational Changeby Engineering Occupation, 1969=72

PercentOccupation changed occupationsAeronautical/astronautical. . . . . . . .Chemical . . . . . . . . . . . . . . . . . . . . . .Civil/architectural . . . . . . . . . . . . . . .Electrical/electronic . . . . . . . . . . . . .Industrial . . . . . . . . . . . . . . . . . . . . . .Mechanical . . . . . . . . . . . . . . . . . . . .Metallurgical/materials. . . . . . . . . . .Mining/petroleum . . . . . . . . . . . . . . .

0.067%0.0270.0180.0390.0520.0320.0360.015

tion that OJT plays a predominate role in adjustingto technological change.

The Case of the Computer Industry

The following section is a brief summary of my re-cent analysis of human capital adjustments by S/Esin computer manufacturing (9). The summary beginswith a cursory overview of the structure and growthof the computer industry in the United States. Subse-quently, I present some empirical evidence on the rela-tionship between technological change and OJT.

TECHNOLOGICAL CHANGE IN THECOMPUTER INDUSTRY

Soma describes the computer industry structure asit has evolved from the early 1950’s. During the 1950’sthere were three distinct segments: the electrical com-ponent manufacturers, the computer manufacturers(mainframe assemblers, sales, and maintenance), andthe end users. By the 1970’s this structure became muchmore complex. For example, some mainframe assem-blers integrated vertically into electronic componentmanufacturing, and independent leasing and mainte-nance firms were established along with softwarefirms, computer utilities, a used computer market, andperipheral manufacturers. A study of the computer in-dustry, therefore, requires the examination of electricalcomponent manufacturers, the various types of hard-ware manufacturers, and software firms. Of course theusers of computer technologies include nearly everytype of enterprise.

Technological change (as defined for present pur-poses) primarily has been contingent on major im-provements in electrical components. The transitionfrom vacuum tube technology (used in the early 1950’s)to transistor-based product lines marked a substantialadvance in computer technology. This adoption inelectrical components appears to have underpinned thesecond generation of computers—from 1959 to 1964.The third generation of computers was founded on the

development of integrated circuits which became eco-nomically competitive in the mid-1960’s. If a fourthgeneration exists, it is not as well defined as theprevious generations. “Most companies have intro-duced new lines of equipment with improved technol-ogy since 1965, but there is no single technical advanceto be used in specifying a new generation. In terms ofcircuits, the primary change has been a movementfrom integrated circuits to large scale integration . . . “(5)0

Based on these technologies, technological changesin computer hardware have varied widely and grownsubstantially. Phister plots annually many of thesechanges over the 1955-75 period. For example, heshows that the average internal memory bytes pergeneral practice system remained roughly constantfrom 1955 to 1965 but steadily increased from 1965to 1974. Other types of measurable innovations in-clude increases in the: storage density of magnetic-corememories and moving-head files, off-line storagecapacity of magnetic tape and disk pack media, aver-age number of moving-head files, magnetic tapedrives, and terminals per system (among many others).

EMPIRICAL ANALYSIS

Exploratory interviews with personnel managers ofa convenient sample of firms in the computer and elec-tronics industry yielded two important general obser-vations. First, formal OJT plays the predominate rolein human capital adjustments for employed S/Es. For-mal OJT typically takes the form of training sessionsranging in duration from 2 to 3 days to 2 to 3 months,sometimes on an intensive full-time schedule but usual-ly on a less intensive part-time schedule. The secondmajor observation from the open-ended interviews isthat employers decide who, when, and how muchtraining is required to meet planned production andservice goals. According to the sample of personnelmanagers, employees who survive the industry takethe training as required—’’its just part of the job.”

It is hypothesized that rapid technological advance-ment in the computer industry requires considerableOJT to keep abreast of rapid obsolescence in humancapital. The probability of OJT in any given year,therefore, is a direct function of technological change.As a general index of technological change, the changein the rate that new computers (including minicom-puters) are introduced annually is employed in theprobability model below. By using an index of thechange in the rate of technological advancement, I amarguing that employers respond to noticeable shifts indeciding to provide OJT. The larger the shift upwardin the introduction of new technology, the larger theprobability that OJT is provided. Conversely, the

8 5

larger the shift downward, the larger the reduction inOJT requirements.

Annual observations on S/Es in the computer in-dustry are drawn from the NLSSE. Observations arepooled for the 1972-75 period for which the incidenceof formal OJT is reported in the survey.

Restricting the sample to S/Es working full-time yearround for the entire 1972-75 period, let P (OJT)ij bethe probability of receiving formal OJT in year i byindividual j. The cumulative logistic function:

P (OJT) ij = 1/1 + e- ( Bij X + Uij)is estimated by maximum likelihood. P (OJT) ij = 1,if OJT was made in year i by individual j, O otherwise;U ij is the error term, and

B ij X = BOij + Blij) (NMODELS) + BziJ (CTOTREV)+ B3ij (AGE) + B4ij (BS) + B5ij (MS) + B6ij (PHD) +B7ij (OCCUP) + B8ij (TENURE)

where:NMODELS =

CROTREV =

AGE =BS .

MS =

PHD =o c c u r =

TENURE =

rate of change in number of new computermodels introduced in year i-3percentage change in annual total industryrevenue in year i-1age in year i for individual j1 if had bachelors degree in year i for in-dividual j1 if had masters degree in year i for in-dividual j1 if had Ph.D. in year i for individual jzero-one dummy variables for engineers,computer specialists, and managers (scientistsand others in benchmark)years employed in job in year i for in-dividual j

Once a new innovation is introduced to the market,there is a delay before it becomes widely adopted.Thus, obsolescence and the need for retraining be-comes a function of that adoption delay. Knowing theappropriate lag of the change in the rate of introduc-tion of new computers is problematic. Unfortunatelythere is little empirical evidence about the rate of dif-fusion of other technological advancements on whichto base a judgment. Work by Mansfield, et al. (13),however shows that the introduction of numericallycontrolled machine tools has taken anywhere from 6to 15 years to become widely adopted by major userindustries. Since technological advancement in thecomputer industry has been particularly rapid, onewould infer that the lag is considerably shorter in thecomputer industry than in other industries. In orderto estimate the most “appropriate looking” lag, themodel is tested against 1- to 5-year lags on theNMODELS variable.

RESULTS

The results of the estimated probability model arereported in table C-5. The partial derivatives of P

Table C-S.—Estimation of the On-the-Job TrainingProbability Model, 1972-75

Variable Mean Coeff icient a t-value

NMODELS . . . . . . . . . . . . . . . –2.75 0.004 1.811CTOTREV . . . . . . . . . . . . . . . . 6.50 –0.007 –0.979AGE . . . . . . . . . . . . . . . . . . . . . 39.25 –0.015 –3.555BS . . . . . . . . . . . . . . . . . . . . . . 0.59 0.157 1.291MS. . . . . . . . . . . . . . . . . . . . . . 0.30 0.128 0.998PHD . . . . . . . . . . . . . . . . . . . . . 0.03 0.667 3.226Computer specialist . . . . . . . 0.35 0.266 3.985Manager . . . . . . . . . . . . . . . . . 0.12 0.179 1.951Other. . . . . . . . . . . . . . . . . . . . 0.04 0.442 2.810TENURE . . . . . . . . . . . . . . . . . 6.63 0.010 1.333Intercept. . . . . . . . . . . . . . . . . 0.173 0.885Chi square , . . . . . . . . . . . . . . 45.55 (10 d. f.)N . . . . . . . . . . . . . . . . . . . . . . . 324.00

aThe c~fficients reported above are B (~) (1 – $), where @ iS the lo9it eStimateand ~ is the mean probability of on-the-job training (= 0.386).

(PJT) ij with respect to each variable are given, eval-uated at the mean P (OJT) ( = 0.386). Using a 3-yearlag on the rate of change in the number of new modelsintroduced annually (NMODELS), a significantpositive relationship with the probability of OJT isfound. The increased probability in OJT in 1972(where the rate of change between 1968 and 1971 =–25), is estimated to be as much as 15 percentagepoints. Given that the mean probability of investingduring the 4-year period under consideration (1972 to1975) is 0.39, the impact of technological change is ap-parently considerable.

No significant relationship is found between the per-centage change in total revenue and the probabilityof OJT. This holds regardless of the lag employed.

Age shows the expected negative relationship, wherethe partial derivative is –0.015 per year of age; signifi-cant at the 0.01 level. As a point estimate, this indicatesthat a 40 year old S/E would be less likely to receiveOJT than an identical 30 year old by as much as 15percentage points. Further tests to examine the linearityof the relationship (i.e., using age in log and quadraticforms) support the inference that the training-age pro-file is downwardly linear.

Only for Ph.D.s does the probability of OJT differby level of education. But here the estimate is peculiar-ly large, albeit, highly significant. Given the size ofthe estimate (0.667) and the fact that only 0.03 of thesample are Ph.D.s, I find it difficult to place much faithin the estimate, On the other hand, the results suggeststhat Ph.D.s are especially prone to receiving formalOJT—perhaps because their functions are so closelytied to generating technological change.

The set of occupational dummy variables shows thatthere are substantial differences among occupations.Since any occupational classification must be com-pared to all others (i.e., all others, including engineers,are in the benchmark), it is impossible to estimate the

86

differences between any two occupations. Yet it canbe inferred that computer specialists and managers re-train more frequently than engineers.

Finally, TENURE obtains an unexpected positivesign but it is insignificant at conventional levels of con-fidence. The positive sign does suggest, however, thatOJT is more likely received later than earlier in one’stenure.

CONCLUSIONS

Using a pooled cross-section time series logit prob-ability model, it was found that technological changehas a substantial impact on the probability of receiv-ing OJT during any given year. The results are limited,however, and must be treated as first approximations.First, the time period under study is limited to 1972to 1975 because of data limitations—both with respectto individual data and available indices of technologi-cal change. Second, the pooling technique suffers froma potential violation of the assumption of independ-ence. That is, if an important variable correlated withthe dependent variable is omitted at the cross-section,pooling that omitted variable (in effect) over time re-sults in autocorrelated error. Although the estimateswould be unbiased, the standard errors would be un-derstated.

As a first approximation, (and bearing in mind theabove limitations) the results support the consensusof a sample of personnel managers in the computerindustry that formal OJT plays in a crucial role inadapting the S/E labor force to changing productionfunctions. Furthermore, the model estimates that thelag between the introduction of new computer modelsand subsequent training is approximately 3 years onaverage.

A similar empirical analysis of the linkage betweenpostschool educational investments and technologicalchange yielded no significant relationship. Becausetechnological change in the computer industry is sorapid and universities cannot afford to provide train-ing based on expensive technologies, continuing educa-tion programs do not provide a viable alternative ofadjustment to technological change—at least for S/Esin the computer industry.

Programmable Automation andLabor Saving Adjustments

What Can We Expect?

The above analysis of OJT indicates that the acquisi-tion of new skills and knowledge is widespread for theS/E labor force. My impression is that the professionalS/E labor force is probably the most highly adaptable

segment of our labor force. By the very nature of S/Eeducational training, workers are well equipped tomake adjustments to outdated skills and knowledge.Indeed one of the primary roles of the S/E labor forceis to advance our knowledge and pursue the goal oftechnological and innovative improvements.

Consequently, growth in programmable automationin manufacturing is unlikely to cause considerabledisruption for the experienced S/E labor force. Adap-tation to technological and innovative advance causedby programmable automation will come in the formof formal and informal OJT. The more rapid the de-velopments in programmable automation, the greaterwill be the extent of OJT and the less likely formaleducation will play an important role in providing ex-perienced S/Es with new skills and knowledge. Thislatter conclusion is based on what has been observedin the computer industry. Universities in general sim-ply do not have the resources for accumulating anddeveloping the necessary physical capital to train S/Esin the latest technologies. Because of limited resources,research and development (R&D) in programmableautomation unfortunately will greatly limit universitiesfrom providing students with advanced training,which in turn will slow the reindustrialization of U.S.manufacturing.

Therefore, the potential fly in the ointment is thelack of resources available to universities to play leadroles in R&D and the education of new S/Es. Theproblem is that newly trained S/E graduates will notprovide the manufacturing industry with cutting-edgetalent. Instead, the industry will need to providesubstantial training to new S/E entrants. Consequent-ly, instead of recruiting S/Es with skills capable ofsatisfying immediate technical needs, manufacturingfirms will experience a lag in recruiting S/Es to imple-ment and improve new programmable automation op-erating objectives.

To aggravate the problem, private industry appearsto be recruiting some of the best talent in S/E Ph.D.s;not only newly minted Ph.D.s but also experiencededucators. The heart of this problem lies in thesubstantial differences between current salaries ofPh.D.s in universities visa vis private enterprise. Thisbrain drain from the universities implies that oppor-tunities for more widespread and advanced educa-tional training will be retarded.

The above scenario suggests that although program-mable automation will be ready technically for diffu-sion throughout manufacturing, the S/E labor marketwill not have the requisite skills to implement the tech-nology-at least in the short run. This will hamper dif-fusion of programmable automation and any imme-diate expected improvements in productivity attribut-able to programmable automation.

8 7

Research Needs

If my best guess about what to expect is reasonablyaccurate, then appropriate social policy should focuson providing universities with the necessary resourcesto establish educational curricula in programmableautomation. Additional resources can come from gov-ernment and/or private industry. The primary re-search need is to examine the current and plannedR&D and educational activities of higher education.The

1.

2.

3.

4.

5.

basic questions to investigate are:What programs focusing on programmable auto-mation have been established? We know, for ex-ample, that several major universities have estab-lished research centers in programmable automa-tion; including Stanford, Purdue, Carnegie-Mel-lon, Massachusetts Institute of Technology, andthe University of New Hampshire (amongothers).Where have universities received resources forthese programs? The above-mentioned univer-sities have relied primarily on NSF grants, dona-tions from corporations, and university budgets.Are universities gearing up for R&D and educa-tional programs in programmable automation?What resource limitations are the universitiesfacing?To what extent is private industry attracting topPh.D. talent? Is this recruitment causing shortrunbottlenecks in developing programmable automa-tion curricula and R&D activities? How can uni-versities maintain their S/E faculty in light of lowrelative salaries?

Answers to these questions are important not onlyto government policymakers but also to higher educa-tion administrators and manufacturing executives.Thus, I would recommend that OTA investigate theanswers to the above set of questions. A survey ofmajor educational institutions could provide OTAwith reliable information to evaluate the above sce-nario. If such an assessment warrants considerable in-terest by the U.S. Government, then a task force com-posed of administrators from higher education, ex-ecutives from manufacturing, and officials from gov-ernment should be established. The purpose of the taskforce would be to design and coordinate educationalefforts to train future S/Es in programmable auto-mation.

In the next section I briefly overview several datacollection efforts by NSF (Division of Science Re-sources Studies). Ongoing efforts by NSF in collectingand analyzing data about S/E labor markets are ex-tensive and in depth. They are potentially very suitablefor collecting the type of information I have suggestedabove. Although I am not recommending at this point

any additional research, data collection efforts by NSFalso provide the most useful data source for monitor-ing how well the S/E labor force adjusts during an eraof programmable automation—both in terms of newentrants and experienced S/Es.

National Science Foundation Data Bases (see NSF)

SURVEY OF SCIENTIFIC AND ENGINEERINGEXPENDITURES AT UNIVERSITIES AND COLLEGES

This survey is conducted biennially to provide in-formation on three areas of academic spending forscientific activities: 1) R&D budgets, 2) expendituresfor departmental research and instruction, and 3) cap-ital expenditures.

SPECIAL SURVEYS

NSF also conducts special surveys. Prominentamong these is the Higher Education Panel Surveywhich is conducted by the American Council of Educa-tion. The survey is conducted several times a year withthe primary objective of providing quick responses tocurrent policy questions relevant to S/E labor markets.(This survey may be especially well suited for the typeof quick evaluation suggested in “Research Needs”above. )

A second example is the survey of Scientific Equip-ment in Academic Institutions. Its purpose is to meas-ure the adequacy and utility of available equipment.

SURVEY OF SCIENTIFIC AND ENGINEERINGPERSONNEL EMPLOYED AT UNIVERSITIESAND COLLEGES

This survey collects data about academic S/Es byfield of employment and primary function.

SURVEY OF GRADUATE SCIENCE STUDENTSAND POST-DOCTORALS

The objective of this survey is to obtain data on thecharacteristics of graduate science and engineeringenrollment at the departmental level. Data from thissurvey provide a base for assessing the relationship be-tween financial support and shifts in graduate enroll-ments.

THE NATIONAL LONGITUDINAL SURVEY OFSCIENTISTS AND ENGINEERS

This biennial survey is a continuing longitudinal ef-fort to maintain a comprehensive picture of the de-velopment and utilization of individuals who were partof the S/E labor force in 1969. The survey elicits in-formation about human capital investments, earnings,and employment/reemployment experiences.

88

THE NATIONAL SURVEY OF RECENT SCIENCE ANDENGINEERING GRADUATES

This survey is conducted biennially, furnishing in-formation on graduates in science and engineeringfields; including data on employment, earnings, andother labor market experiences.

THE NATIONAL SURVEY OFDOCTORATE RECIPIENTS

This is a biennial survey with the primary objectiveof estimating the national supply and utilization ofdoctoral S/Es.

References

1.

2.

3.

4.

5.

6.

7.

8.

Azariadis, Costas, “Implicit Contracts and Under-employment Equilibrium, ” ]ourna] of PoliticalEconomy, vol. 83, December 1975, pp. 1183-1202.Baily, Martin N., ‘Wages and Employment UnderUncertain Demand,” Review of Economic Studies,vol. 411, January 1974, pp. 37-50.Baily, Martin N., “On the Theory of Layoffs andUnemployment,” Econometrica, vol. 45, No. 5,July 1977, pp. 1043-1063.Ben-Porath, Y., “The Production of Human Cap-ital Over Time, ” in Education, Income and HumanCapita], Lee Hansen (cd. ) (New York: NBER,1970), pp. 129-147.Brock, G., The U.S. Computer Industry (Cam-bridge, Mass.: Ballinger Publishing Co., 1975).Brown, M., On the Theory and Measurement ofTechnological Change (Cambridge, Mass: MITUniversity Press, 1966).Cooke, William N., Labor Market Adjustments byScientists and Engineers: Probabilities and Out-comes (report to NSF, 1980).Cooke, William N., “Permanent Layoff Probabil-ities: What’s Implicit in the Contract, ” IndustrialRelations, vol. 20, spring 1981.

TECHNOLOGY AND LABORby Markley RobertsAFL-CIO

July 27, 1982

9.

10.

11.

12.

13.

14.

Cooke, William N., Human Capital Adjustmentsto Technological Change in the Computer Indus-try: The Case of Scientists and Engineers (reportto NSF, 1981).Cooke, William N., and Cobern, Morris, “Part-Time Post-School Investments in Education andTheir Impact on Earnings Growth for Engineers,”in Human Resources and Employment in the De-veloped Countries: International Economy andEmployment, Proceedings of the 6th World Con-gress of the International Economic AssociationMeetings, forthcoming.Gordon, Donald F., “A Neo-Classical Theory ofKeynesian Unemployment,” Economic Inquiry,vol. 12, December 1974, pp. 431-459.Mansfield, E., The Economics of TechnologicalChange (New York: W. W. Norton & Co., Inc.,1968).Mansfield, E., et al., The Production and Applica-tion of New Industrial Technology (New York: W.W. Norton & Co., Inc., 1977).National Science Foundation, Scientific and Tech-nical Personnel Data System: Strategy and Design,proceedings of a conference, October 1979,

15. Phister, M., Data Processing Technology and Eco-nomics (Santa Monica, Calif.: Santa Monica Pub-lishing Co., 1976).

16. Rosen, S., “Measuring the Obsolescence of Knowl-edge, ” in Education, Income and Human Behavior,T. Juster (cd.) (New York: NBER, 1975).

17. Solow, R., “Investment and Technical Progress,”in K. J. Arrow, et. al. (eds. ), Mathematical Meth-ods in the Social Sciences (Stanford, 1960).

18. Soma, J. T., The Computer Industry (Lexington,Mass.: D. C. Heath & Co., 1976).

19. Somers, G., E. Cushman, and N. Weinberg (eds.),Adjusting to Technological Change (New York:Harper & Row, 1963).

Technology changes the way goods and services are Therefore, workers and their unions have a direct andproduced and distributed—and for all its potential vital interest in how technology is introduced in thebenefits, including creation of new jobs, technology workplace—to make sure people get priority overalso has destructive effects on workers and their jobs. technology, to make human values prevail.

8 9

Technology often involves labor-saving opera-tions—increased production with the same number orfewer workers. This may wipe out many existing jobs.It may raise new dangers to workers’ safety and health.Of course, new jobs may also be created. New pro-tections may be achieved for workers’ safety andhealth. But the impact of new technology is often toeliminate some jobs, change the job content of others,change skill requirements, and change the flow ofwork.

Technology often causes changes in industry loca-tion—shutdowns of departments and entire plants andshifts to new locations in suburban or outlying areasand sometimes overseas. No industry is immune tosuch changes, which are constantly shifting the struc-ture of skills, occupations, jobs, and earnings ofAmerican workers.

Collective Bargaining

Collective bargaining holds a vitally important rolein meeting the challenges, opportunities, and dangersof new technology. There is much to be learned frompast experience in collective bargaining. The flexibili-ty of this institution, the American system of labor-management bargaining at the plant, company, andindustry level, helps workers negotiate and settle withemployers on reasonable and humane protections forworkers against the potentially adverse effects of job-destroying technological innovation. Mature collectivebargaining relationships between labor and manage-ment provide more opportunities and a sound basisfor special labor-management committees to deal withadjustment to technological change within the frame-work of collective bargaining.

Collective bargaining can help democratize labor-management relations and humanize the workplaceand work itself, including the impact of new technol-ogy on workers’ jobs and earnings. Collective bargain-ing can provide cushions to soften the adverse impacton workers by setting up adjustment procedures andprograms at the workplace. In a full employment econ-omy—linked with adequate employment services, em-ployment and training programs, and unemploymentcompensation—the disruption of workers’ lives andthe job displacement resulting from technologicalchange can be minimized.

Historically, unions have responded in a number ofways to the introduction of new technology. In 1960,Sumner Slichter, James J. Healy, and Robert Liver-nash, reported that major determinants of union pol-icies toward technological change are:l

‘Sumner Slichter, James J. Healy, and Robert Livernash, The Impact ofCofkctive Bargaining on Management (Washington, D. C.: The BrookingsInstitution, 1960), ch, 12, “Union Policies Toward Technological Change.”

1. the nature of the union, meaning specificallywhether it is a craft or industrial union;

2. the economic condition of the industry or the en-terprise, or occupation, whether it is expandingor contracting, whether the industry is highIycompetitive or not;

3. the nature of the technological change, the effecton jobs and on the bargaining unit, the effect onworkers’ skills and job responsibilities; and

4. the stage of development of the technologicalchange and the stage of development of unionpolicy toward the technological change.

Slichter, Healy, and Livernash distinguish five prin-cipal policies that unions adopt when faced with tech-nological change: 1) willing acceptance, 2) opposition,3) competition, 4) encouragement, and 5) adjustmentwith an effort to control use of the new technology.

They note:The most usual policy of unions toward technologi-

cal change is willing acceptance. This happens in thenumerous cases in which the technological changemakes little difference in the kind and degree of skillrequired and has little immediate effect on the numberof jobs. But the gain in productivity from the changemay make it attractive by giving labor improved op-portunity to bargain for wage increases. Unions maybe led by favorable bargaining opportunities to acceptwillingly technological changes that involve a mixtureof advantages and disadvantages. Thus, the bargain-ing advantages that accompany a change requiringgreater skill may lead to willing acceptance even thoughit greatly reduces the number of jobs.Slichter, Healy, and Livernash go on to point out

that no national union in recent years has destroyeditself by fighting technological change.

Nor is there record of any union in recent years beingable to prevent technological change by opposing it—though many unions have retarded recent changes tem-porarily and locally. Union wage policies appear to havebeen partly responsible for stimulating technologicalchange under some circumstances and may have af-fected the distribution of gains. Three principal effectshave been produced by union policies toward techno-logical changes:

(1)

(2)

(3)

They have tended to give to the holders of jobson the new machines or new processes some-what higher wages relative to other workers inthe same plant—in other words, they havetended to introduce distortions in the wagestructure of the plant.They have tended to a slight extent to cause thenew techniques to be operated with excessivecrews and under make-work rules.They have considerably eased the hardship ofdisplacement, partly by forcing managementsto do advance planning in the introduction oftechnological changes and partly by giving dis-placed workmen an opportunity to qualify forother jobs.

Using the approach developed by Slichter, Healy,and Livernash to distinguish the five principal attitudesthat unions take toward technological change, DorisMcLaughlin of the University of Michigan made a sur-vey of union officials, management, and mediators andarbitrators on the impact of labor unions on the rateand direction of technological innovation.

The McLaughlin report2 found that willing accept-ance was the most common response American laborunions make to the introduction of new technology.The next most common response was initial opposi-tion, but this was followed by adjustment, so that, inthe long run, willing acceptance or adjustment were,by far, most common.

A negative union response to the introduction oftechnological change was invariably the result of beliefthat acceptance would have an adverse effect on a largeor important segment of the union’s membership. Ifthe employer convinced the union’s leaders that theirmembers would not be adversely affected, or thatthose who were adversely affected would receive someoff-setting benefit, union opposition disappeared.

The three most important variables in determiningunion reactions, in order of importance, were:

1. the state of the economy,2. union leaders’ perception of the inevitability or

necessity for the change, and3. the nature of the industry.McLaughlin noted that, depending on union percep-

tion of these three variables, a fourth variable-wheredecisionmaking power lay—became crucial. If the in-ternational union held the decisionmaking power, adecision on how to react to the new technology wouldbe made only on consideration of the first threevariables. However, if decisionmaking power lay withlocal union leaders, three more variables became rel-evant:

4. how local union leaders perceive the impact ofthe new technology on the bargaining unit,

5. how local union leaders perceive the “quid proquo” offered by the employer to the affectedunion members, and

6. how local union leaders perceive the impact onthose union members left in the unit after the newtechnology is introduced.

Third-party action by mediators, arbitrators, orjudges did not seem to affect the outcome, accordingto the report, but did appear to affect the process bywhich unions and management reached accommoda-tion to the effects of the new technology. These third-party agents, as outsiders, serve a useful function in

‘Doris B. McLaughlin, The Impact of Unions on the Rate and Directionof Technological Innovation (Detroit, Mich.: Institute of Labor and IndustrialRelations, University of Michigan–Wayne State University, February 1979),report to the National Science Foundation, grant PRA 77-15268.

taking the heat off local union leaders “when other-wise politically delicate decisions need to be made withregard to the introduction of new technology, ” thereport states.

Surprisingly, labor unions are not the major stum-bling block to new technology and higher productivi-ty, McLaughlin concludes, but “employer representa-tives, particularly at the middle management level,were often cited as constituting the real barrier to theintroduction and effective use of technologicalinnovation. ”

In 1964, the Bureau of Labor Statistics reported3 thatsome of the major labor-management efforts to pro-tect against the effects of new technology haveincluded:

1.

2,3.

4.5.

6.

guarantees against job or income loss and, insome cases, against loss of supplementary benefitsfor varying periods,compensation for employees who lose their jobs,guaranteed income for workers required to takelower paying jobs,provisions for retraining,provisions for transfer to other plants and pay-ment of relocation expenses, andagreements to provide workers with notice ofplant closings or other major changes.

Some agreements have established joint labor-management committees to recommend methods ofproviding for workers affected by automation. Thereport concluded that:

These arrangements typically are combined with pro-visions for retention of workers with greatest seniori-ty, but in a limited number of cases, efforts are madeto spread work among larger numbers of employees orto encourage early retirement of workers with relative-ly high seniority.In 1966, the Automation Commission called atten-

tion to the need for private sector efforts to facilitateadjustment to technological change including relianceon attrition, an advance notice early warning system,job counseling and job-finding assistance, training andretraining. The Commission noted the rationality ofusing the seniority principle in the case of layoffs andthe seriousness of the need for pension and health ben-efits to continue during periods of unemployment.They also pointed out that technological improve-ments can bring more flexibility to work schedules andmore leisure to employees through reduced hours ofwork per day, per week and per year,

3Cokctive Bargain@ and Technological Change (Washington, D. C.: U.S.Department of Labor, Bureau of Labor Statistics, March 1964), BLS reportNo. 266.

‘Technology and the American fionomy, vol. I (Washington, D. C.: Na-tional Commission on Technology, Automation, and Economic Progress,February 1966).

9 1

The record of collective bargaining response to tech-nological change offers many examples of both suc-cess and failure, the Commission noted:

Collective bargaining has proved to be an excellentvehicle for the effective management to change; it per-mits those directly affected by the change to deal withit firsthand and with a familiarity that takes into ac-count peculiarities and problems peculiar to an enter-prise. Especially in recent years, some managements andunions, occasionally but not usually with the help ofoutsiders, have developed, with varying degrees of in-genuity and success, plans to facilitate change.”

But the Commission warned:Despite its many successes, collective bargaining has

often failed, and sometimes has failed spectacularly,to deal effectively or even responsibly with the manage-ment of change. It has been argued, not unreasonably,that the failures are the fault of the parties, not of thesystem.

Procedurally, the process of collective bargaining onbasic issues has tended to stagnate during the life ofthe agreement and to accelerate frantically in an atmos-phere of crisis immediately preceding contract renewal.Happily, employers and unions in a number of indus-tries are abandoning this pattern in favor of more orless continuous discussion. Basic issues such as adjust-ment to technological change cannot be resolved, how-ever, by a small team of negotiators working themselvesinto a state of physical and mental exhaustion for a fewmonths every 2 or 3 years. These issues must be dealtwith patiently, carefully, and above all, continuously,until satisfactory solutions emerge. This kind of bar-gaining calls for ability of the highest caliber on the partof leaders of both labor and management.In the 15 years since the Automation Commission’s

report, with generally slow economic growth and re-cessions in 1969-70, 1973-75, 1980, and 1981-82, eco-nomic conditions have not been conducive to easy ad-justments to technological change. The impact of newtechnology has become much more pervasive in the1980’s than it was in the 1960’s.

It must be emphasized that it is easier to deal withadverse effects of technological change in a generaleconomic climate of full employment. National eco-nomic policies must aim at full employment for a vari-ety of economic, social, and moral reasons. Amongthose reasons we must recognize the need to facilitatesuccessful and humane adjustments to job-destroyingtechnology in both the private and public sectors.

Much progress has already been achieved throughcollective bargaining. For example, a 1981 Bureau ofLabor Statistics study,5 updating a similar 1966-67

‘U.S. Department of Labor, Bureau of Labor Statistics, Major CollectiveBargaining Agreements: Plant Movement, lnte~lant Trans& and Reloca-tion Allowances, Bulletin 1425-20, July 1981. The 1966-67 study was reportedin U.S. Department of Labor, Bureau of Labor Statistics, Major CollectiveBargaining Agreements: Plant Movement, Transfer, and RelocationAllowances, Bulletin 1425, July 1969.

study, presents a wide range of contract language andstatistical summaries of contract language on plantmovement, plant transfer, and relocation allowances,many of which relate to the effects of technologicalchange. Agreements limiting plant movement rosefrom 22 percent in the 1966-67 survey to 36 percentin the 1980-81 survey of some 1,600 contracts, whileworker coverage rose from 38 to 49 percent. InterPlanttransfer provisions increased from 32 to 35 percent andworker coverage went from 46 to 49 percent. Agree-ments dealing with relocation allowances increasedfrom 34 to 41 percent while worker coverage went upfrom 60 to 65 percent.

On the issues of the major technological change,work transfer, or plant closings, some major contractshave a variety of provisions.’ For example, the UnitedAuto Workers (UAW) -General Motors contract pro-vides for advance notice to the union in cases of tech-nology-related permanent layoffs, and negotiation ofrights related to plant closing, department closing, andcompany transfer of work. Workers have the right totraining for a new job in cases of technology-relatedpermanent layoff. In the case of plant closings, depart-ment closings and transfer of work, workers have theright to bump to another job in the same plant, transferto a replacement facility, or transfer to a new plant.They will receive preferential hiring at another plant,keep seniority with respect to fringe benefits, get mov-ing expenses up to $1,355, take layoff with recallrights, and get severance pay.

The United Steel Workers’ contract with KennecottCopper includes a no-layoff clause and attrition pro-tection for workers affected by technology changeswhich will permanently eliminate their jobs. Under thiscontract, workers have the right to bump to anotherjob in the same plant or in another plant. The TransitWorkers’ contract with the New York City transit sys-tem and the Newspaper Guild’s contract with the NewYork Times also have no-layoff contract protection.

The Steel Workers’ contract with American Can Co.calls for a 12-month advance notice of permanent lay-offs related to technological change. The United Foodand Commercial Workers’ contract with Armour callsfor 6-month notice, and the Guild-New York Timesagreement calls for 4 months. There are contracts withadvance notice requirements as short as 7 days andcontracts with advance notice requirements, but nospecified time period.

A broad range of labor-management cooperation isalready included in many other labor-managementagreements with negotiated specific procedures for ad-justing to technological change.

bThe following contract provisions are listed in Industrial Union Depart-ment, AFL-CIO, Comparative Survey of Major Collective Bargaining Agreements, Manufacturing and Non-Manufacturing March 1979, December 1979.

9 2

One method to ease the human costs of new tech-nology is to assure advance information to workersand their unions about management plans for futureinnovation which will affect workers with job loss orother serious problems. Major technology changesresult from management decisions taken long beforethe new technology is actually introduced, often yearsearlier. Certainly there should be long advance noticebefore any technological change which results in lay-offs or plant shutdown. The failure of managementto institute worker safety-health and environmentalprotections should not be the way workers learn aboutintended plant shutdowns or major layoffs.

An “early warning system” of advance notice helpsmake it possible to ease the problems of affected work-ers. Such “early warning” provisions have long beenstandard in many union contracts. With advance no-tice and labor-management cooperation, workers canlook for or train for a new job, perhaps with the sameemployer in the same plant or at another location,Employer-paid retraining is an important part of anyadjustment-to-innovation program.

There are other methods and techniques for labor-management cooperation to cushion adverse effectsfrom changing technology. These include incomemaintenance with work and/or pay guarantees. Oneway is through “no-layoff” attrition to reduce the workforce by natural turnover, deaths, retirements, andvoluntary quits, thus protecting the jobs and earningsof those workers who remain with the company. Ofcourse, attrition alone is not an adequate solution.“Red circle” earnings protection for workers down-graded through no fault of their own attaches a wagerate to an individual instead of to the job itself andthus protects workers against loss of income whichmight result from innovation-induced downgrading.

Seniority is a key principle in protecting workersagainst layoffs and downgradings. This rewards longservice, but does much more—properly reflecting theworker’s investment in the job and the company’s in-vestment in the worker. Early retirement is an optionthat older workers should have available when majortechnological change wipes out their jobs. But the op-tion should be available as a free choice, not as a re-quirement. Many older workers cannot afford to retireearly and others prefer to continue working.

Transfer and relocation rights and mobility assist-ance to workers are other ways to provide job and in-come protection. Within-plant and interplant transfers,relocation assistance, severance pay, pension rightsand seniority protections and supplemental unemploy-ment benefits can all help cushion adverse effects onworkers and their families when industrial innovationoccurs.

Shorter workweeks and reduced time per year onthe job, including longer paid vacations and sabbaticalleaves, also can ease the negative employment effectsof technology.

Electrical machinery manufacturing is an industrywhere extensive use of robots is expected in the future.The June 1982 General Electric agreement with the In-ternational Union of Electrical Workers includes theseprotections for workers who lose jobs to robots andautomation: 7

A production employee whose job is directly elimi-nated by a transfer of work, the introduction of a robotor of an automated manufacturing machine and whois entitled to transfer or displace to another job shallbasically retain the rate of the eliminated job for a peri-od of up to 26 weeks.

The company shall give the union advance notice ofa minimum of six months of plant closing or transferof work and of a minimum of 60 days of the installa-tion of robots or automated manufacturing machinefor production.

An employee who is terminated because of a plantclosing will be assisted to find new jobs and learn newskills under an employment assistance program whichwill include job counseling as well as job informationservices.

An employee with two or more years of service whois terminated as a result of a plant closing will be en-titled to receive education and retraining assistance, in-cluding reimbursement of $1,800 for authorized educa-tion expenses.Obviously these provisions do not constitute total

protection but they offer some protection and somehelp to displaced workers.

Public Policy

More information is needed on the effects of chang-ing technology on workers. Federal action is neededto set up a clearinghouse to gather information on acontinuing basis on innovation and technologicalchange and its effects on the welfare of the Americanpeople, on jobs, skills, training needs, and industrylocation. Few economic studies of the impact of tech-nological change exist because there is no systematicdata-gathering relating to the changing technology ofAmerican production. With more and better informa-tion, public and private adjustment programs can bet-ter avoid needless human hardship and suffering whichtoo often result from the disruptive impact of chang-ing technology.

Through this clearinghouse, the Federal Govern-ment could provide unions and employers with com-prehensive information and service, upon request, to

‘Bureau of National Affairs, Daily Labor Report, June 29, 1982.

9 3

help develop labor-management solutions for the com-plex problems related to the impact of technologicalchange at the workplace.

Technology-caused economic dislocation and otherkinds of dislocation—including plant shutdownscaused by corporate merger mania and by recession,job loss from trade policies and production shifts awayfrom defense-related industry-require cooperativelabor-management efforts and also national programsto deal with these complex problems. Further explora-tion is needed of a variety of such programs, includingproposals dealing with plant shutdowns and plant relo-cation and with reconversion of defense-related indus-try.

Occupational training and retraining may perhapshelp displaced workers acquire new skills and newjobs—but such new jobs may be at lower skill levelsand at lower pay. Furthermore, the loss of an industryand the skills and know-how that go with that industrydiminish the essential diversity and pluralism requiredfor a healthy economy and healthy society.

In mid-1982 Congress was moving to approve newfederally supported job training legislation, but thescope of the program is too small to help the millionsof workers who have lost their jobs to technologicalchange and economy shifts. The recession has sharp-ened union pressures to get retraining commitmentsfrom employers.8

Workers who lose their jobs because of plant clos-ings may not be able to find new ones or maybe forcedto work at reduced pay. Family life is often disrupted.The mental and physical health of displaced workersoften declines at a rapid rate. Research over a 13-yearperiod indicates that the suicide rate among workersdisplaced by plant closings is almost 30 times the na-tional average.9 Such workers also suffer a far higherthan average incidence of heart disease, hypertensionand other ailments.

Bills to deal with this grave economic and socialproblem have been introduced in Congress .’” Althoughthese bills differ in some respects, they would do muchto counteract the devastating effects of shutdowns andrelocations. Unfortunately, they do not address theproblems caused by the relocation of governmentalfacilities. Among other things, these bills would:

‘For example, see “Retraining Displaced Workers: Too Little, Too Late?”Week, JU]Y 19, 1982, pp. 178-185.

‘Barry Bluestone and Bennett Harrison, Capital and Communities: TheCauses and Consequences of Private Disinvestment (Washington, D. C.: Pro-gressive Alliance, 1980), pp. 78-82. The health-unemployment link is one ofthe most clearly documented social research conclusions, e.g., Harvey Bren-ner, Estimating the Costs of National lkonomic Policy Implications for Mentaland Physical ffeahh, and Criminal Aggression, Joint Economic Committee,U.S. Congress, Oct. 26, 1976.

IoFor example, ~ gbth Cong. bills H.R. 5040, introduced by Congre=manFord of Michigan; S. 1608, introduced by Senator Riegle of Michigan; S. 1609,introduced by Senator Williams of New Jersey; and S. 2400, introduced bySenator Metzenbaum of Ohio.

1.

2.

3.

4.

require firms to provide advance notice of theirintentions to close or relocate a major facility;advocate programs to support troubled busi-nesses, including incentives to promote employeeownership;call for the issuance of economic impact state-ments and Federal investigation of the circum-stance; andrequire employers, whenever existing jobs can-not be saved, to provide minimal protections totheir workers in such matters as transfer rights,relocation expenses, severance pay, pension pro-tection, health care, and job training.

Three states—Wisconsin, Maine, and Michigan—have laws relating to plant shutdowns, and some 15other States have proposals pending with State labororganizations pressing for action on protective plantshutdown legislation at the State level. However, be-cause of “competitive laxity” among the States in theirefforts to attract new business and “runaway” business,Federal legislation with national plant closings stand-ards is essential,

Unfortunately, since reporting on plant closings isvoluntary, the U.S. Government does not have cen-tralized, comprehensive information on this importantsocial and economic issue. We don’t even knowwhether most plant closings are related to technologychanges or tax incentives, to short- or to long-termeconomic pressures.

For labor it is crucially important to require employ-ers to recognize their responsibilities to their employeesand their communities before they shut down a plantand to provide economic protections to workers andtheir families who must suffer the consequence of toohasty corporate action. There is nothing radical orunusual about national legislation requiring advancenotice and other worker-community protections. Inother nations, private business firms—including affili-ates and subsidiaries of many American firms—findthey can live with laws requiring advance notice andother protections for workers and communities againstthe adverse effects of economic dislocation and plantshutdowns.

In terms of international comparisons, Sweden re-quires 6 months notice where more than 100 workersare involved, 4 months notice where 26 to 100 workersare involved, and 2 months notice where 5 to 25 work-ers are involved. Under Swedish law, no dismissalsmay take place until the unions have been contactedand granted an opportunity to negotiate concerningthe issues and consequences of the dismissals. In theUnited Kingdom, 90-day notices must be given where100 or more workers are involved and 30 days inplants employing 10 to 99 workers. Failure to commu-nicate with the unions and to give the appropriate

94

notice can make the employer liable for continuing payof the workers during the required notice period. InFrance, Greece, and the Netherlands, prior to makinglarge-scale dismissals, the firm must have permissionof the government to lay off the workers and in actualpractice the advance notice period is as long as halfa year to a year depending on the specific circum-stances.

These examples indicate that advance notice is apractice with which firms can live. It must also beremembered that in most foreign countries the benefitspaid workers are generally two-thirds of lost earningsfor up to 1 year after the layoff.

Unfortunately, in the United States, there area num-ber of tax advantages provided for corporations whichclose down even viable, moneymaking plants. Con-gress should look into these plant closings very careful-ly to determine if there is indeed an array of tax incen-tives encouraging businesses to close down plants. Leg-islation must be created which will stop such incen-tives and will prevent tax-related plant shutdowns.Legislation must also be created which will establishbasic job and income protections for workers and pro-tection of workers’ pension and health care and otherbenefits, to deal in an effective and humane way withthe economic and social dislocation resulting fromplant closings.

Industrial Democracy

The potential for misusing technology is great, butthe possibility of human progress through the wise andhumane use of technology is equally great. The oppor-tunity for new technology to be introduced with mini-mal social disruption will be greatly enhanced if work-ers and employers have an equal opportunity for dis-cussion and joint decisionmaking on the subjects ofchanging technology and the quality of working life.

Collective bargaining, an established institution inour democratic society, has been a fair and workableprocess for joint labor-management decisions onwages, working conditions, and other major issues.It is therefore a logical mechanism for increasing theinvolvement of workers in such areas of decisionmak-ing as adjustment to new technology.

New technology and rising expectations are forcingtransformations in the workplace. Applications of newtechnology should be humane for workers as well asprofitable to business. Human and social values mustbe more highly valued in the production process, notonly when the process is producing goods and services,but also when it is producing cultural and social valuesand leisure and unemployment. The human desire forgreater autonomy and greater participation in deci-sionmaking on the shop floor, in the corporate board-

room, and in national economic policymaking mustreceive higher priority. Improvements in the “qualityof work life” (QWL) include a broad range of issues,such as better occupational safety and health, as wellas work-organization, work environment, and longruninvestment, employment and training decisions, andthe introduction of new technology. These QWL issuesare logical subjects for joint labor-management negoti-ation and decision. But employers must not use QWLas a disguise for union-busting.

Irving Bluestone, a former UAW vice president, hasbeen a strong proponent of increased worker participa-tion in corporate decisions. He warns:ll

The joint union-management programs that are inexistence have not yet proven themselves in any per-manent sense. They must be subject constantly to re-view and change as management, the union, and theworkers learn by doing. Although it is not possible toset forth a precise blueprint to ensure the successful par-ticipation of workers in the decision-making process,experience already indicates certain criteria that arebasic:

●

●

●

●

●

●

●

●

� � � ✎

The programs should be voluntary. Workers musthave the free opportunity to decide whether or notto participate in the program. To order compul-sion is to invite resistance and failure.Workers should be assured that their participationin decision-making will not erode their job securityor that of their fellow workers, that they will notbe subject to ‘speed up’ by reason of it, and thatthe program will not violate their rights under thecollective bargaining agreement.Workers should genuinely experience that they arenot simply adjuncts to the tool, but that they arebent toward being creative, innovative, and inven-tive plays a significant role in the production (orservice) process.Job functions should be engineered to fit the work-er; the current system is designed to make the work-er fit the job on the theory that this is a more effi-cient production system and that, in any event, eco-nomic gain is the worker’s only reason for working.This theory is wrong on both counts.The worker should be assured the widest possiblelatitude of self-management, responsibility, and op-portunity for use of “brainpower.” Gimmickry andmanipulation of the worker must not be employed.The changes in job content and the added respon-sibility and involvement in decisionmaking shouldbe accompanied by an effective reward system.Workers should be able to foresee opportunities forgrowth in their work, and for promotion.The role of workers in the business should enablethem to relate to the products being produced orthe services being rendered, and to their meaningin society; in a broader sense, it should also enablethem to relate constructively to their role in society.

lllWi% BIuestone, Ch. 12, “Emerging Trends in Collective Bargaining, ” inWork in America: The Decade Ahead Clark Kerr and Jerome M. Resow(cd.) (New York: 1979), Van Nostrand Reinhold, pp. 249-50.

95

Quality of Work Life

The conflict theory of labor relations is the soundestbasis for worker representation, worker participation,and worker gains. Conflict is institutionalized in ourpolitical system. Conflict is institutionalized in ourlegal system. Conflict is institutionalized in our eco-nomic system. And we have institutionalized conflictin labor-management relations through the Americansystem of collective bargaining.

But the adversary role, which is appropriate to theconflict of collective bargaining, should be limited tothe period of negotiation—and during the life of thecontract, the adversary relationship can very logical-ly and appropriately be replaced by cooperation aimedat maximizing the potential success of the enterprise,the company, or the establishment. The labor relationscycle should be one of periods of conflict during thenegotiating period followed by the longer contract pe-riod of cooperation.

Collective bargaining is basic and fundamental tohonest labor-management cooperation—and such co-operation can be mutual self-help supplement to col-lective bargaining. Committees that exist outside col-lective bargaining or try to take over the process ofcollective bargaining are not to be trusted.

Labor unions today jointly participate with manage-ment in thousands of safety committees, appren-ticeship committees, communitywide labor-manage-ment committees, quality of work life committees,quality circles and other joint labor-management ef-forts. As a result, labor-management committees arejoining together to deal with matters of mutual interestsuch as foreign trade; Federal, State, and local pro-grams; community philanthropic purposes; and revi-talization and strengthening of their industry and theircommunity as well as those committees which giveworkers a direct voice in the issues of the workplace,including investment and innovation with new tech-nology.

But, only the collective bargaining stature of theirunions establishes workers as real partners in thoselabor-management committees. Any action that weak-ens a union, distorts the balance in its relationship tomanagement, or its ability to represent its member-ship, will damage that union’s ability and desire to par-ticipate in committees of any kind with a particularmanagement.

Any program which strengthens the union’s abilityto grapple with the new issues union members wantaddressed, including technology change issues, anyprogram which holds out real promise for the expan-sion of workplace democracy, ought to be grasped,minutely examined for flaws, reshaped as necessary,

polished, and put in place—and then watched verycarefully.

QWL programs in the United States have takenmany different forms and appeared in many differentguises-participatory management, employee involve-ment, shop-floor democracy, consultation schemes,labor-management committees, quality circles,autonomous work groups, QWL teams, profit-sharingincentive structures, etc.

As a tool used toward labor’s basic goals, theseQWL programs can develop skill improvement pro-grams, more flexible working schedules, greater jobsecurity and promotional opportunities, along withmany other matters of great importance to the mem-bers we represent. So, other things being equal, unionshave every reason to encourage and cooperate in anyenterprise that will work to those constructive ends,for the benefit of workers and management alike.

But hard experience has taught us to look for pit-falls. Too many employers are more interested in pro-grams that offer cosmetic changes that try to fool theworkers into believing that management really caresabout them, in spite of low pay and bad working con-ditions.

Too many union-busting “consultants” are pro-moting these QWL programs as an alternative toworker participation through trade unionism. Thatway, without the protection of a union contract, anyconcessions to workers can be revoked as easily as theywere given. At best, the QWL group concept posesa problem to the labor movement because of the poten-tial that exists for management to penetrate and in-fluence small, informal work groups and bust unions.

For strong unions, able to insist on an equal and ac-tive voice in how the QWL program works and able,if necessary, to veto actions that aim at subverting itsbargaining position, these are not insuperable prob-lems. That accounts for the general acceptance ofQWL programs by such dominant and secure unionsas the Auto Workers, Steel Workers, and Communica-tions Workers. Even they have sometimes had to takestrong action to prevent their employers from usingthe programs for company propaganda in bargainingsituations.

Other unions are in a more difficult posture. Theyhave organized only a piece of the action—the otherfacilities of the same firm unorganized, under contractwith another union, or a mixture of both. Here man-agement more often controls the introduction and im-plementation of technology and QWL programs.Sometimes, in fact, they move ahead to new technol-ogy with major layoffs or plant closings with barelya nod to the union.

In doing so, they establish a carefully orchestratedorganizational and communication link with the em-

ployees that can bypass or attempt to supplant theunion. In part, this accounts for the more antagonisticresponse of unions which have only bits and parts ofdifferent firms and thus more limited bargaining lev-erage.

QWL programs, under whatever name, can be oftremendous help in facilitating the dealing with thelarger issues of collective bargaining, including wagesand working conditions and the job impact of newtechnology. At the same time, QWL programs can dealwith other less visible but basic issues that affect theindividual at the workplace.

Labor has no intention of allowing management toco-opt any of these basic issues. But dealing with QWLprograms will present our unions with immense prob-lems of educating our members—training and retrain-ing of shop stewards and business agents; giving at-tention to the overall coordination of QWL programsplant by plant, employer by employer, and individualby individual; and developing at national staff levelsthe technical expertise to assist in the negotiation ofQWL programs and in their development and main-tenance; and resolving problems relating to sharingtechnology’s benefits; and deciding what are necessaryagreements and conditions before entering into QWLprograms.

Every union must continue in every way possibleto assert its rights and the rights of its members to ac-ceptance as legitimate equals in a partnership withmanagement, with collective bargaining as the essen-tial foundation for labor-management cooperation.

We recognize the valuable contribution that proper-ly constituted and equally balanced labor-managementprograms can make in fulfilling the American tradeunion member’s desire for individual recognition, dig-nity, safety, quality of work life, and job security.

To that end, unions will cooperate with manage-ments that recognize and support the right of workersfreely to join unions of their choice and who demon-strate willingness to work with unions as equal part-ners in all areas that affect their members’ interests,including the impact of technology.

But unions will reject, as a dangerous fraud, all ef-forts to use specious programs and rigged committeesto undermine unions, divert attention from the realneeds of workers, and weaken enforcement of the Na-tion’s labor laws.

Conclusion

Workers and their unions have reasonable, under-standable, and legitimate concerns about loss of jobs,loss of income, and loss of life and health. If these con-cerns are met adequately and effectively, workers willbe much more willing to accept and adjust to chang-ing technology.

There are no simple solutions to the task of protect-ing workers against the adverse impacts of changingtechnology. In thousands of labor-management con-tracts covering millions of workers in both the publicand private sectors, unions and management haveadopted a wide variety of provisions to cushion work-ers against these adverse impacts. These provisions fallinto a few general categories—job protection, incomeprotection, safety and health protection, retraining,and relocation assistance. The specifics include attri-tion or no-layoff protection, early warning of techno-logical change, seniority protections, early retirementopportunities, “red circle” pay protection, shorterworkweeks or work-years, relocation rights to followtransferred operations, severance pay, negotiatedsafety-health protections supplementing safety-healthlaws and regulations, and many other specific labor-management collectively bargained responses to tech-nological change.

Without full collective bargaining—no matter howenlightened or benevolent management may be—working men and women simply don’t have a senseof participation in the basic decisions which governtheir jobs, their income and their lives. Collectivebargaining is essential to help workers share thebenefits of technological progress and help workers tomeet the challenge of technological change with aminimum of social and human dislocation.

9 7

LABOR= MANAGEMENT RELATIONS IN AN ERA OFPROGRAMMABLE AUTOMATIONby William N. CookeKrannert School of Management, Purdue University

July 27, 1982

Introduction

Currently we know very little about the impact ofprogrammable automation on labor-management rela-tions. One gets the general impression that the impactto date has been relatively small, except on a handfulof occupations and industries. The potential fordramatic growth in the utilization of programmableautomation, however, is generally acknowledged, al-beit, the timing of rapid diffusion remains iffy. Thepotential for growth raises a number of importantquestions about labor-management adjustments dur-ing an era of programmable automation. First, wemust ask: What shapes the decisions of employers toinvest in programmable automation? Will unions im-pede the diffusion of programmable automation andto what degree (if any)? What impact will unions haveon the lag between the introduction of commerciallyavailable programmable automation and its diffusionin manufacturing? Simultaneously, we must ask: Whatwill be the degree of displacement of workers? Whathappens to displaced workers? What proportion willbe retrained by employers? How many will be laid off?What kind of changes in work rules will unions seek?What kind of changes will unions gain? Will white-collar workers seek union representations?

Answers to these questions require in-depthresearch. Toward developing a research design, I beginwith an overview of the current utilization of roboticsin manufacturing and the extent of collective bargain-ing. I then discuss several collective bargaining issuesrelevant to an understanding of labor-management re-lations and the utilization of robotics. Subsequently,I lay out a research agenda, including a theoretical ex-planation of collective bargaining, implicit modelspecifications, and data collection.

Programmable Automation andCollective Bargaining in Manufacturing

Although programmable automation encompassesmore than robotics, it is this form of programmableautomation that most directly impacts on blue-collarwork forces and, thus, existing union-managementrelations. From a research design perspective, workerdisplacement caused by the introduction of robots canbe more precisely quantified than say, for example,displacement caused by the implementation of com-puter-aided design techniques. Furthermore, firms andplants utilizing robots can be identified.

Currently, robots are capable of performing tasksassociated with a handful of occupations. Accordingto the Robotics Institute of America (RIA), there werefewer than 5,000 robots in use in the United States in1980. Table C-6 reports the estimated number ofrobots by broad occupational application (17).

The type of broad occupational categories most vul-nerable to replacement by robot applications are heavi-ly concentrated in the metalworking industries (1).Table C-7 lists the basic metalworking industries by2-digit Standard Industrial Classification (SIC) code,the number of unions representing workers within anindustry, and the estimated union membership by per-centage category.

It can be seen readily that metalworking manufac-turing is heavily unionized and by a fairly large num-ber of different unions. Table C-8 identifies some ofthe national unions that have been active in organiz-ing drives in metalworking industries during 1980.

Time restraints in preparing this report have notallowed me to investigate the scope of contractualagreements pertaining to programmable automation.However, using the Bureau of Labor Statistics’ (BLS)

Table C-6.–Robot Usage and Identification of Applications, 1980

Machine loading, Painting/Welding unloading Foundry finishing Assembly Other

Number of robots . . . . . . . . . . 1,500 850 840 540 100 600

98

Table C-7.—ldentlfication of Metalworking Industriesand Extent of Unionization

Percent inNumber of industry

lndusty SIC unions unionized

Primary metals . . . . . . . . . . . 33 50-75Fabricated metals. . . . . . . . . 34 28 50-75Machinery (except

electrical) . . . . . . . . . . . . . . 35 16 25-50Electrical/electronic

equipment . . . . . . . . . . . . . 36 14 50-75Transportation equipment. . 37 16 75-1ooSOURCE: Bureau of Labor Statlstlcs, LXrectorY of fW/orra/ Urt/orK? and Hr@oyee

Assoclat\ons, 197fi Bulletin 1837.

Table C-8.-Unions Representing Workers inMetalworking Manufacturing Industriesa

1. Allied Industrial Workers of America (AIW)2.3.

4.

5.

6.7.

8.9.

10.

11.

12.13.14.

15.16.

17.

18.

19.

20.21.

22.23.24.

25.26.27.28.—

Aluminum Workers international Union (AWU)Automobile, Aerospace, and Agricultural implementWorkers of America (UAW)Boilermakers, Iron Shipbuilders, Blacksmiths, Forgers,and Helpers; International Union (BBF)Carpenters and Joiners of America; United Brotherhoodof (CJA)Chemical Workers Union; international (ICW)Clothing and Textile Workers of America; Amalgamated(ACTW)Communications Workers of America (CWA)Electrical Workers; International Brotherhood of (IBEW)Electrical, Radio, and Machine Workers of America United(UE)Electrical, Radio, and Machine Workers of America; In-ternational Union of (IUE)Engineers; international Union of Operating (IUOE)Furniture Workers of America; United (FWW)iron Workers; international Association of Bridge Struc-tural and Ornamental (BSOIW)Laborers; international Union of (LIU)Longshoremen’s and Warehousemen’s Union; interna-tional (ILWU)Machinists and Aerospace Workers; internationalAssociation of (IAM)Marine and Shipbuilding Workers of America; industrialUnion of (IUMSW)Metal Polishers, Buffers, Platers, and Helpers; interna-tional Union (MPBP)Molders and Allied Workers Union; International (i MAW)Oil, Chemical, and Atomic Workers; international Unionof (OCAW)Painters and Allied Trades; international Union of (IUPAT)Paperworkers; United international Union (UPI)Plumbing and Pipe Fitting industry; United Associationof Journeymen and Apprentices (PPF)Service Employees; international Union (SEIU)Sheet Metal Workers’ international Association (SMW)Steelworkers of America; United (USA)Teamsters, Chauffeurs, Warehousemen, and Helpers ofAmerica; International Brotherhood (TCWH)

~he above list of unions was taken from NLRB union election files. The unions(and others) were involved [n union representation elections in 1980 in themetalworking industries (SiC 33-38).

data base on union contracts (Characteristics of MajorCollective Bargain@ Agreements), an examination ofrecent contracts negotiated by unions in manufactur-ing would reveal how unions alter the scope of collec-tive bargaining as programmable automation becomesmore widely applied.

A glimpse at contracts and various publications ofthe UAW, the USW, the IUE, and the IAM indicatethat joint union-management committees have beenestablished to address the general issue of productivi-ty and the specific issue of technological change. A cen-tral purpose of these committees is to address in ad-vance impending technological change: how to preparefor changes in the production process and how to miti-gate its impact on the work force. Below I discuss theset of alternatives that unions and employers are like-ly to consider. One can only imagine that the scopeof contractual agreements will change as more work-places apply programmable automation to the produc-tion process. I propose below that we examine thesecontracts and the undertakings of joint union-manage-ment committees in order to understand more fullyhow labor-management relations cope with the ap-plication of programmable automation.

Current Collective Bargaining:Relevant Issues

In the following section I discuss several issues thatare especially relevant to understanding union-man-agement relations. My intention is to describe brieflysome important parameters that help set the stage formy subsequent discussion of collective bargaining inan era of programmable automation.

Concession Bargaining

The current mood and trend in collective bargain-ing in manufacturing differs substantially from thepast. Due to a deep recession and growing interna-tional competition, unions and employers have begunmaking concessions in wages, benefits, and work rules.Concessions by unions have covered wage and benefitpackages and, to some degree, restrictive work prac-tices. Concession bargaining is being witnessed inmany places of employment—well beyond the highlypublicized automobile, agricultural implement, rub-ber, electronic machinery, and ongoing steel negotia-tions. A recent mid-May poll by Louis Harris & As-sociates of 600 large corporations found that 26 per-cent of the unionized firms had obtained wage andbenefit concessions in recent negotiations (4).

Employers, likewise, have had to make conces-sions—albeit types of concessions that do not lend

99

themselves to easy pecuniary estimation. These con-cessions have included more union and worker inputinto management decisions and a variety of securityprovisions (e.g., added SUBS and reversals or post-ponement of plant closure decisions).

Several points can be made about the current con-cession bargaining. First, the recession itself would nothave precipitated these concessions. The recessionmerely brought the longrun ills of some major in-dustries to a head. The surge of foreign competitionis the cause of the ongoing decline in much of U.S.manufacturing. Consequently, collective bargaining isbeginning to address the long-term livelihood of manyindustries.

Second, concessions are being made by both unionsand employers. This suggests that unions in these man-ufacturing industries are not ipso facto in weakerbargaining positions than employers. Instead, the in-dustry is in a state of demise and, consequently, bothemployers and unions need to change standard oper-ating procedures.

Third, job security has become the major bargain-ing chip. Rank and file have insisted on additionalforms of job security (or monetary cushions to dis-placement) as the quid pro quo for concessions. Jobsecurity issues will continue to hold the limelight infurther bargaining—primarily due to the longrun im-pact on existing employment in manufacturing; bothin the potential reduction of existing jobs and in re-quiring major changes in the structure of work.

Programmable automation, therefore, will beviewed by labor much like foreign competition isviewed—as a threat to longrun employment. Jobsecurity, in turn, will be the bargaining chip forcooperation with management in restructuring hard-hit industries. In order to maximize the utilization ofprogrammable automation, unionized employers willhave to higgle and haggle over job security provisions.

Legal Requirements of Collective Bargaining andthe Issue of Programmable Automation

Union-management relations have been shaped sub-stantially by the National Labor Relations Act (NLRA)and any departure from the legal duty to bargain ingood faith is highly unlikely. The National Labor Rela-tions Board (NLRB) is the regulatory agency chargedwith the interpretation (along with the courts) and ap-plication of NLRA. It has laid down the ground rules(albeit foggy ones at times) to protect both parties fromunfair labor practices.

In particular, the NLRB has attempted to delineatemandatory bargaining subjects (i.e., the subjects andissues for which the parties must negotiate in goodfaith). Good faith bargaining requires discussion but

not necessarily concession to either party’s demands.However, the NLRB generally has interpreted the lackof compromise as evidence of bad faith bargaining.Mandatory subjects fall under the heading of wages,hours, and other terms and conditions of employment.Ever since the landmark Supreme Court case of Fibre-board Paper Products v. NLRB (1964), the NLRB hasinterpreted such issues as subcontracting and plantclosures as mandatory bargaining subjects. Until 1972(Summit Tool Co. Case), employers had the legal re-sponsibility to negotiate in good faith about decisionsto subcontract (where such subcontracted work waspreviously done by current employees) and closeplants. The Summit Tool Co. Case appears to allowemployers to make unilateral decisions to close plantsbut still requires employers to negotiate in good faithabout the displacement effects of plant closures.

The question to raise is, what will the NLRB requirein terms of good faith bargaining when workers facedisplacement by programmable automation? It ap-pears that if employers close plants and resort to sub-contracting previous in-house production, they will beobligated to negotiate with unions—at least about theimpact of these decisions on labor. It is difficult tosecond-guess the NLRB since: 1) it is not clear to whatextent employers will close plants and/or subcontract,2) the complexity of cases leads to unclear precedentand, thus, case-by-case resolution, and 3) the chang-ing make-up of the Board leads to inconsistency in theinterpretation of unfair labor practices (10).

Although it is unclear what impact the NLRB willhave in resolving disputes over programmable auto-mation, it is clear that NLRB governance of labor-man-agement relations will help shape labor-managementrelations affected by programmable automation. It isalso clear (given the enormous workload of the NLRB)that the typical long delay in NLRB conflict resolu-tion will add to the transitory problems we might an-ticipate as we move into an era of programmable auto-mation.

Union Organizing Activity

Table C-9 provides a description of union electionactivity since 1950. Column 1 shows that the percentof workers unionized in the nonagricultural labor forcehas declined steadily since 1955—from a high of 33percent to the present low of approximately 24 per-cent. Part of this decline is attributable to a long-termdrop in union success in representation elections; froma high of winning 75 percent of elections in 1950 tothe current low of winning only 46 percent of elections.Concurrent with increased election losses has been asubstantial drop in the average size of work unitsholding elections; from 519 workers in 1950 to only

100

Table C-9.—Annual Observations on Selected Parameters ofUnion Representation Elections

(1) (2) (3) (4) (5)Percent of thenonagricultural

labor force Percent of Average Percent PercentYear unionized elections wonb unit sizeb manufacturing consent b

1950. . . . . . . . . . 31.5 74.5 158.5 39.0 (c)1955 . . . . . . . . . . 33.0 65.3 122.4 38.6 42.61960 . . . . . . . . . . 31.5 58.6 75.9 36.6 42.21965 . . . . . . . . . . 28.5 60.2 70.0 35.6 46.91970 . . . . . . . . . . 27.3 55.2 75.4 33.2 26.51971 . . . . . . . . . . 27.0 53.2 70.1 31.8 23.11972 . . . . . . . . . . 26.4 53.6 66.3 31.6 20.21973. . . . . . . . . . 25.9 51.1 57.8 31.8 16.31974. . . . . . . . . . 25.8 50.0 61.5 31.2 14,71975. . . . . . . . . . 25.3 48.2 66.3 29.4 11.61976. . . . . . . . . . 24.5 48.1 55.0 29.4 10.41977. . . . . . . . . . 24.1 46.0 60.2 29.2 8.91978. . . . . . . . . . 24.0 46.0 57.3 29.2 7.9aDataare for calendar years.bDataareforflscal years.cNo comparable flgureavallable for Im.

SOURCES: Columns2,3,and 5aretaken fromvarlousannual reports oftheNLRB.Column4 Isadoptedfromthe 1979Hand-book of Labor Statlstlcs. Column 1 IS adoptad from the Directoty of National Unions and Employee Assoclatlons,1979.

57 workers in 1978. Obviously, this has had a substan-tial affect on total union membership. Column 4 pre-sents the percent of workers employed in manufactur-ing, The loss in manufacturing employment is widelycited as a leading cause in the decline of union member-ship, primarily because manufacturing has been a his-toric stronghold for the union movement. A recentstudy of union elections over the 1970-78 period showsthat roughly 28 percent of all private sector electionswere held in manufacturing industries and that, onaverage, workers were more likely to vote againstunion representation (all other things held constant)than their counterparts in nonmanufacturing industries(19).

Stepped-up employer resistance, however, has per-haps had the greatest impact on union representationelections. Employers, for example, rarely consent tounion elections today. Instead, employers have beencampaigning actively against union representation. Forinstance, using the percent of elections consented toby employers, column 5 shows that consent electionshave dropped dramatically from a high of 47 percentin 1965 to under 8 percent by 1978. Recent studies finda large negative and highly significant relationship be-tween not consenting to elections and those electionoutcomes (9,19).

Another area of interest is white-collar unionization.Chamberlain, et al. (7), report that as of 1976, roughly

18 percent of the membership in national unions waswhite-collar (professional-technical, clerical, and salesworkers). That figure represents an increase in white-collar unionization in recent years. For instance, in1960 only 12 percent of national union membershipwas white-collar. This reflects an increase from2,200,000 union members in 1960 to 3,850,000 in 1976.If membership in employee associations is added tounion membership, the proportion of members fromwhite-collar occupations becomes 27 percent. Cooke(9) also finds in an analysis of private sector electionsin 1979 that both professional-technical and clerical-sales work groups were much more likely to vote forunion representation than blue-collar workers votingin representation elections. However, it should also benoted that only 13 percent of 1979 elections involvedprimarily white-collar work groups.

Several important points can be drawn from theabove discussion. First, the union movement in theprivate sector has been experiencing a long-termdecline in its relative power–at least as proxied bymembership figures. Second, the relative bargainingpower in manufacturing has experienced the greatestslippage. Third, employer resistance to further unionorganizing has increased dramatically over the last 15years. Finally, white-collar unionization has been in-creasing and remains a potential growth area for theunion movement.

101

Collective Bargaining in an Era ofProgrammable Automation:A Research Agenda

Absent one of those funny little crystal balls, thetask of anticipating the impact of programmable auto-mation on labor-management relations will requiremore in-depth and sophisticated research than is cur-rently available. In the following section I sketch outa general theory of collective bargaining. I then raisea set of questions about the impact of union-manage-ment relations. Subsequently, I develop a set of im-plicit models to answer these questions. Finally, Idiscuss data collection issues and propose a researchplan.

A General Theory of Collective Bargaining

Dunlop (11) advocates the use of a systems frame-work in the general analysis of union-managementrelationships. His framework encompasses threebroadly defined actors: 1) workers and their represent-atives, 2) management, and 3) interested governmentagencies (the NLRB for present purpose). These actorsinteract within a set of environmental constraints (tech-nological, economic, and sociopolitical) to establishthe rules of the work relationship; both pecuniary andnonpecuniary. Although Dunlop’s framework is basedon a historically observed set of relationships, he failsto breathe any life into the framework. A suitabletheory, however, requires some underlying moti-vator(s) that helps explain and predict behavior. Mygeneral intention here is to utilize Dunlop’s set of ac-tors, but make them dynamic and explain their interac-tion within the environmental context.

Theoretical analyses of negotiations are based on theconceptual idea that bargaining “power” determinesthe outcome(s) of negotiations or, more generally,union-management relations. Concerned primarilywith the ability of unions to raise wages, Pigou andHicks (16,13) define bargaining power as the abilityof unions to increase wages above competitive wagelevels. This concept of bargaining power has becomepopularly known among economists as the union’s“monopoly power” to raise wages (20). Chamberlainand Kuhn (8) suggest a broader definition of powerthat encompasses more than wage gains: power is theability to secure some agreement that otherwise wouldnot be granted. This latter definition is more appro-priate for our present purpose. In the simplest context,unions (employers) use power to force employees(unions) to agree to sets of work rules.

Since power is primarily the means of attaining someunderlying goal(s), one needs to examine more close-

ly the central goals of the parties. It would seemreasonable to believe that unions, employers, andgovernment agencies act under some premise of utili-ty maximization. Although utility maximization of-fers an underlying motivator (i.e., maximizing behav-ior), it does not lead us to what actually motivates par-ties; except “whatever motivates the parties” (i.e., utili-ty). Unless the researcher assumes what utility rep-resents (e. g., profits, wages, etc. ), the researcher can-not model very well any cause-effect relationships.Since the analysis of industrial relations issues encom-passes economic, sociological, political, and psycho-logical relationships, we need to establish a motivatingprinciple which is amenable to incorporating such acomplex array of factors.

I argue here that unions, employers, and the NLRBattempt to “optimize control” over employment rela-tions. For unions (acting as the voice of workers), op-timization of control over employment relations in-cludes, for example, increasing job security andminimizing employer discretion with respect to wagesand benefits, work assignments, displacement, andworker discipline. Optimization of control, however,does not mean that unions want complete responsibili-ty for managing employment relations. Instead, unionswant to maximize control up to the point where theirmembers’ employment prospects are not jeopardizedthrough massive layoffs or business closures. In con-trast to union optimization of control, employers wantnear unilateral control of all decisions affectingemployment relations. Obviously, unionized employ-ers have given up substantial control of the work rules.The optimization thesis holds that unions and employ-ers are motivated to wrestle back as much control aspossible; which implies that any state of equilibriumis short-lived. Such an optimization principle, there-fore, places unions and employers in a natural stateof conflict (although not necessarily in a destructiveone).

During negotiations both unions and employersdraw on their respective bargaining power to wrestlecontrol from one another. Under the thesis that bar-gaining power determines negotiation outcomes, thesources of bargaining power must be examined. Butfirst, bargaining power should be defined in relativeterms since union and employer willingness to usepotential power is a function of the perceived net gain(i.e., total benefit minus total cost) associated withusing that potential power. For instance, a union’s will-ingness to endure a strike to force an employer to agreeto a change in a contract is dependent on the perceivednet gain to the union of such a strike. Likewise, an em-ployer’s willingness to take a strike to avert signingthe change in contract is dependent on the perceivednet gain associated with taking a strike. In defining

1 0 2

relative power, Chamberlain, Cullen, and Lewis (7)argue that:

Only if the cost to management of not agreeing to theunion’s terms exceeds the cost of agreeing with them,and if the cost to the union of not agreeing to manage-ment’s terms is less than the cost of agreeing to them,does the union’s bargaining power surpass that of themanagement.

Such a definition of relative power is dependent onthe size of the demand, however. For example, if aunion’s demand is for a 30-percent increase in “costs”to the employer, the union’s relative power is less thanif the demand is only 10 percent. This implies thatunions and employers will be more successful in ne-gotiating contract changes as the cost of changebecomes smaller.

The sources of relative power are determined by anarray of economic, technical, legal, organizational,and sociopolitical factors. As any factor changes (ordiffers across firms), the relative bargaining powerchanges (or differs across firms). The set of implicitmodels described below are derived by focusing onparameters of the various sources of relative power.This set of structural equations is designed to provideanswers to some basic questions about union-manage-ment relationships in an era of programmable auto-mation.

Model Specification: Implicit Models

At this stage of the analysis I formulate several im-plicit models; models that begin to establish the con-ceptual understanding of cause-effect relationships be-tween the diffusion of programmable automation andunion-management relations. Explicit empirical speci-fications detailing functional form and variable meas-urement will require considerable additional effort.

Based on a theory of control optimization, anychange in the work rules attributable to programmableautomation will be a function of the employer’s bar-gaining power visa vis the union’s bargaining power.Stated algebraically,

Work Rules = f (POWm/POWu)where POWm = power of management and POWU

= power of union. As discussed above, the relativepower of the parties depends on the sources of powerand the perceived cost of the change in work rules.Thus,

POWm/POWU = f (sources of power + cost of change)

The utilization or diffusion of programmable automa-tion (which can be viewed as a change in work rules),therefore, becomes a function of relative power:

Diffusion = f (Powm/Powu)

Similarly, a union’s response to the utilization of pro-grammable automation is a function of relative power:

Union Resp. = f (Powm/ P o wu)Several of the questions raised in the introduction

basically ask: how will unions influence the diffusionof programmable automation? In effect, we want totest the hypothesis that a union’s response influencesdiffusion. The dependent variable, therefore, is the dif-fusion of programmable automation. Diffusion can beevaluated as: 1) the decision to invest, or 2) somemeasure of the amount of utilization (e.g., number ofrobots), and/or 3) the lag in diffusion. The depend-ent variable can be evaluated at firm or industrywidelevels.

Our theory implies that the diffusion of program-mable automation increases as the relative power ofmanagement rises above the relative power of theunion, and conversely. Since relative power is a func-tion of the perceived cost of the change of work rulesplus sources of power, relative power varies as the costof the pending change varies and as the parametersof the sources of power vary. Consequently, our anal-ysis focuses on the cost of proposed changes in workrules as well as the parameters of sources of power.

We can begin by postulating that the diffusion ofprogrammable automation is a function of: 1) theunion response to proposed or anticipated utilizationof programmable automation, plus 2) a vector (X) ofother variables.

Diffusion = f (union response, X)Our theory says that the union response will be a func-tion of the cost of proposed changes and its source ofpower.

Union Resp. = f (cost of changes + sources of power)Thus, everything else constant, as the cost to a unionof the application of programmable automation in -creases, the more negative the response of the union.As discussed in the section about concession bargain-ing, job security becomes the central issue and theprimary bargaining chip in negotiations. At the ex-treme, for example, if displaced workers are laid off(and say without severance or relocation pay), the costto the union in membership and status is quite high.Of course, the greater such layoffs, the greater the costto the union. Under these circumstances the union re-sponse will be strongly negative to the diffusion ofprogrammable automation. At the other extreme,where, for example, all displaced workers are retrainedor placed in alternative and equivalent jobs, the costto the union is negligible: causing no negative responseby the union (except perhaps to seek the establishmentof a joint union-management committee to prepare forfuture diffusion).

1 0 3

Under the negative response scenario, (and again,everything else constant) the union will attempt tonegotiate work rules protecting its membership. Thus,the union will attempt to increase the cost to theemployer of making changes in working rules. Onecan imagine a wide range of proposals at the bargain-ing table, including: 1) guarantees of employment (per-haps tied to years of seniority); 2) restrictions on tim-ing of the implementation of programmable automa-tion (e.g., implementation would be tied to normalwork force attrition or requirements of advancednotice); 3) retraining requirements; 4) postponementof plant closures and subcontracting; 5) cushions tolayoffs (e.g., severance, relocation, or SUB payments,and early retirement schemes); 6) reduction in workhours; and 7) union shop requirements covering new-ly created jobs.

Holding constant the cost of management’s pro-posed changes in work rules, we can now consider thesources of power that determine the union’s ability tonegotiate protective work rules. It is hypothesized thatthe greater the union’s power, the more successful itwill be in negotiating protective work rules, and themore costly it becomes for management to changework rules (i. e., utilize programmable automation).

Sources of power are gained or lost according tovariation in the economic and technological contextof the firm, union organizational strength, legal con-straints, and the sociopolitical environment. Withrespect to the economic context, the power of theunion varies with: 1) labor costs (e.g., wages, laborcost per unit produced, wage bill/total cost, etc.);2) the product market (profits, sales, industry concen-tration, insulation from foreign competition, etc.); and3) the labor market (employment growth in firm orindustry, layoffs, unemployment, etc. ) As an illustra-tion, in a firm or industry experiencing a loss in sales,declining profits and employment, and a growingthreat from international competition, the union’ssources of power are diminished as well as its abilityto shift the cost of work rules to the employer.

With respect to the technological context, the morethe production process is amenable to the utilizationof programmable automation and/or the less strategicthe work force in the production process, the less thepower available to the union. This relationship isanalogous, in part, to the study of union strike activi-ty, whereby union strike activity is reduced as the pro-duction process can be manned temporarily by non-union supervisors and workers.

Everything else constant, organizational strengthplays a role in determining the union’s strength.Worker unity in supporting leadership initiatives, thewillingness of workers to endure strikes, the financial

resources of the union, membership size and extent oforganization, and negotiation skills are all positivelyrelated to union power.

The NLRB also plays a potential role in shaping theunion’s power. The impact of plant closures and sub-contracting on the work force, for example, are man-datory bargaining issues. If the NLRB interprets its cur-rent legal precedent about displacement to also includedisplacement caused by the introduction of program-mable automation, then the union’s power to negotiateprotective work rules is enhanced.

Finally, both employers and unions are sensitive tothe sociopolitical climate. If public opinion supportsnotions of “job rights” and “employer responsibility”for displacement, then unions will draw on this sup-port in negotiating with firms. Sociopolitical supportfor union initiatives will depend in large part on theimpact of displacement on a given community. Com-munities have played, for instance, an active role inkeeping some plants open.

In summary, the impact of unions on the diffusionof programmable automation depends on the cost tothe union of its implementation and union sources ofpower. Likewise, the diffusion of programmable auto-mation depends on the power of the employer (whichis a function of the cost to the employer of changesin the work rules and the employer’s sources of power).The cost to the employer of utilizing programmableautomation can be treated as a standard investmentdecision. The rate of return is obviously reduced asunions are able to increase the cost of utilization (i. e.,more expensive work rules). The ability of employersto keep the investment cost down is a function of theirpower sources. The sources of power are determinedfrom the same set of economic, technological, orga-nizational, legal, and sociopolitical parameters thatdetermines the power of the union. Parameters thatyield power to the union generally take power awayfrom the employer, and conversely, Through negotia-tions over work rules, the parties will eventually signa contract that generally reflects compromises by both,and which theoretically reflect the relative power ofthe parties.

The types of renegotiated work rules will also reflectthe preferences of the parties. For example, unions andemployers may be willing to trade regular wages andfringes for greater security provisions like retrainingand relocation pay, since job security becomes a pri-mary concern. The types of outcomes, for labor, there-fore, are tied to relative bargaining power and pref-erences of the parties. One can imagine a wide varia-tion in outcomes among unionized firms.

It seems reasonably clear that as the capabilities ofprogrammable automation increase and as the price

104

per unit drops, there will be a substantial impact onunionized workers in selected occupations, industries,and localities. It would not be feasible for employersto retrain substantial portions of their work forces.Consequently, unions will attempt to bargain workrules that cushion the layoff experience of largenumbers of workers. It also seems reasonably evidentthat as the diffusion of programmable automation pro-gresses rapidly in nonunionized establishments, thelongrun threat to unionized groups of losses in employ-ment due to noncompetitive work rules will diminishthe sources of power for unions. As such, it becomesless likely that unionized employers will agree to pro-tective work rules.

Finally, let me briefly address the issue of unionorganizing activity of white-collar workers. First, thediffusion of programmable automation (especially inthe form of computer-aided design and manufactur-ing) also threatens existing white-collar jobs. The desirefor job security and due process are important factorsin work group decisions to unionize. As noted in thesection “Union Organizing Activity, ” white-collarunionization has been increasing and remains a poten-tial growth area for the union movement. Second,unions which can organize both blue- and white-collarworkers, obviously increase their sources of bargain-ing power to negotiate protective work rules. Thus,in firms and industries with the greatest potential forwhite-collar displacement by programmable automa-tion, the likelihood that white-collar work groups willunionize increases. One can note, for example, thatwhite-collar workers at General Motors very recentlyhave shown considerable interest in attaining UAWrepresentation primarily, it appears, because of declin-ing job security.

Data Analysis and Research Plan

Our knowledge of the relationship between the dif-fusion of programmable automation and labor-man-agement relations is very limited. We simply have noadequate empirical information to make reliablejudgments about the impact of unions on the diffu-sion of programmable automation, nor the impact ofprogrammable automation on union-management re-lationships and worker displacement. In part this canbe attributed to the modest and welcomed encroach-ment of programmable automation to date. However,if the type of projections of programmable automa-tion utilization made by RIA are reasonably accurate(a threefold increase for 1985 over 1980, and twenty-fold increase by 199o), the need to increase ourknowledge on this subject is highly warranted.

I suggest that we examine existing evidence in lightof the implicit models described above. The first step

would be to formulate a set of tentative explicit em-pirical models derived from the above implicit model-ing. Once the critical (and measurable) variables areidentified and the models are written in appropriatefunctional form, we can begin efforts to obtain thenecessary observations to test the models.

The information we seek covers: 1) the extent ofutilization and how employers weigh the reaction ofunions and/or workers in making investment deci-sions; 2) the negotiation experience (i.e., the changesin work rules); 3) the parameters of the sources ofpower for employers and unions; and 4) the impacton the work force (e.g., numbers of workers retrained,relocated, or laid off, etc. ) Toward this end we needthe cooperation of employers and unions. Unfor-tunately it is not clear at this point how much coopera-tion we will receive. The second step, therefore, callsfor a limited exploratory effort to test the waters. Intesting the waters, we can make a judgment of the like-lihood of successfully gathering the necessary data.Furthermore, we will learn: 1) what data on the vari-ables in the tentative explicit models can be collected,and 2) what additional parameters should be consid-ered in our hypothesis testing.

If the second step indicates that further data collec-tion is feasible, then as a third step we need to iden-tify the users of programmable automation; and ideal-ly, potential users. Given that in 1980 there were about5,OOO robots installed, we have a fairly large popula-tion of programmable automation applications. RIAis probably the best source of information on users andpotential users in manufacturing. After identifying thepopulation of users and potential users, we can thenbegin the process of selecting a representative sample.A small population of users, however, would allowus to examine the entire population.

Finally, once the data have been collected, the ex-plicit equations can be estimated. Only at that pointshould inferences be drawn, projections made, andpolicy scenarios evaluated.

References and Selected Bibliography

1.

2.

3.

4.

5.

Ayres, Robert, and Miller, Steve, “IndustrialRobots on the Line,” Technology Review, vol. 85,No. 4, May/June 1982, pp. 34-46.Ayres, Robert, “Robotics, CAM, and IndustrialProductivity,” NationalProductivity Review, vol.1, No. 1, winter 1981-82, pp. 42-60.Business Week, “Robots Join the Labor Force,”June 9, 1980, pp. 62-76.Business Week, “Concessionary Bargaining,” June14, 1982, pp. 66-81.Business Week, “Retraining Displaced Workers:Too Little, Too Late?” July 19, 1981, p. 178.

105

6.

7.

8.

9.

10.

11.

12.

13.

14.

15.

Carnegie-Mellon University, “The Impacts ofRobotics on the Workforce and Workplace,” man-uscript, June 14, 1981.Chamberlain, Neil W., Cullen, Donald E., andLewis, David, The Labor Sector (New York:McGraw-Hill Co., 1980).Chamberlain, Neil W., and Kuhn, James W., Col-lective Bargaining (New York: McGraw-Hill Co.,1965).Cooke, William N., “The Collective Decision toUnionize: Theory and Evidence, ” Industrial andLabor Relations Review, forthcoming, April 1983.Cooke, WiIliam N., and Gautschi, Frederick, III,“Political Bias in NLRB Unfair Labor Practice Deci-sions, ” Industrial and Labor Relations Review, vol.35, No. 4, July 1982, pp. 539-549.Dunlop, John T., industrial Relations Systems(New York: Holt, Reinhart, & Winston, 1958).Extejt, Marian, “The Impact of Robotic Tech-nology on Human Resource Management: A Pre-liminary Investigation, ” working paper, Univer-sity of Pittsburgh, 1982.Hicks, John R., The Theory of Wages (New York:MacMillan, 1932).McLaughlin, Doris B., “The Impact of LaborUnions on the Rate and Direction of TechnologicalInnovation, ” manuscript, University of Michigan-Wayne State University, 1979.Office of Technology Assessment, U.S. Congress,Exploratory Workshop on the Social Impacts ofRobotics—A Background Paper, OTA-BP-CIT-11, February 1982.

16.

17.

18.

19.

20.

21.

22.

23.

24.

25.

26.

Pigou, A. C., Economics of Welfare (London:MacMillan), 1938.Robotics Institute of America, WorldwideRobotics Survey and Directory (Dearborn, Mich.:1982).Robotics Today, “Surveys Reveal Robot Popula-tion and Trends,” February 1982, pp. 79-80.Seeber, Ronald L., and Cooke, William N., “TheDecline in Union Success in NLRB RepresentationElections,” Industrial Relations, forthcoming,winter 1983.Simons, Henry C., Economic Policy for a FreeSociety (Chicago: University of Chicago Press,1948).Somers, Gerald G., Cushman, Edward L., andWeinberg, Nat (eds.), Adjusting to TechnologicalChange (New York: Harper & Row Publishers,1963)Time, “The Robot Revolution, ” Dec. 9, 1980, pp.72-83.Wall Street Journal “Steel-Collar Jobs: As RobotAge Arrives, Labor Seeks Protection,” Oct. 26,1981.Wall Street Journal, “The Big Revolution on theFactory Floor,” July 12, 1982.Zager, Robert, “A Working Model,” New YorkTimes, Sept. 7, 1981.Zager, Robert, “T’he Problem of Job Obsolescence:Working It Out at River Works, ” Monthly LaborReview, July 1978, pp. 29-32.

0