ED 459 346

AUTHORTITLE

INSTITUTIONSPONS AGENCYPUB DATENOTE

CONTRACTAVAILABLE FROM

PUB TYPEEDRS PRICEDESCRIPTORS

ABSTRACT _

DOCUMENT RESUME

CE 082 708

Freeman, Peter; Aspray, WilliamThe Supply of Information Technology Workers in the UnitedStates.Computing Research Association.National Science Foundation, Arlington, VA.1999-00-00148p.; Also supported by the Intersociety Working Group onInformation Technology Workers.EIA-9812240Computing Research Association, 1100 17th Street, NW, Suite507, Washington, DC 20036-4632 (Single copy free) . Tel:202-234-2111; e-mail: info@cra.org; Web site:http://www.cra.org. For full text:http://www.cra.org/reports/wits/cra.wits.html.Information Analyses (070)MF01/PC06 Plus Postage.Classification; Computer Science Education; Data Analysis;*Demand Occupations; Education Work Relationship; Females;Government Role; Higher Education; *Information Technology;Labor Force Development; Labor Market; *Labor Needs; *LaborSupply; Minority Groups; Older Adults; School BusinessRelationship; Secondary Education; Supply and Demand;*Technical Occupations; Technology Education

This study is designed to improve understanding of thesupply of and demand for information technology (IT) workers in America andcontextual issues surrounding that topic. Chapter 1 examines aspects of thepolitical context concerning IT workforce issues. Chapter 2 outlines a way todistinguish IT workers from a much larger class of workers whose jobs areenabled by IT and classifies them into four categories--conceptualizers,developers, modifiers/extenders, and supporters/tenders--based on skills andknowledge required to do the job. Chapter 3 addresses dynamics of themarketplace and dangers of government intervention in the IT labor market,limitations on action to improve a supply-demand mismatch, costs of an ITworker shortage, and international considerations. Chapter 4 evaluateswhether there is a shortage of IT workers and determines that data areinadequate to ascertain a national supply-demand mismatch. Chapter 5describes an extensive supply system, including majors in 20 IT-relateddisciplines at the associate, bachelor's, master's, and doctoral degreelevels and majors in science, engineering, business, and non-technicaldisciplines. Chapter 6 looks at non-degree programs as a supply system.Chapter 7 focuses on underrepresentation of women, minorities, and olderworkers in the IT workforce. Chapter 8 focuses on seed-corn issues, whetherthe strong industrial demand for IT workers is harming the educationalsystem. Chapter 9 discusses data sources and their limitations. Chapter 10offers recommendations for federal and state governments, higher education,industry, professional societies, and individuals. (YLB)

Reproductions supplied by EDRS are the best that can be madefrom the original document.

Peter Freeman and William Away

MS-

. .11k. k

The.... e. Supply of.

O. 8. foiL AttgaTechnoja

LefEinancialOfficer grtkers Chi6f

Marl_Lj_gAmg InformationVP in the wiser

*134 led States0 Entreyeneur

Business.aaect

>t,ApplicationManager

!*. 13eloper

Ac_cpuntantTechnology

Chief

Ofitice r/0 SW

ProjectManager

cn.41BankTeller

(0)NetworkInstaller

Callre Consultant Developer

OS

SystemAdmin

Information TechnOlOgy Knowledge

Peter Freeman and William Aspray

TheSupply ofInformationTechnologyWorkersin theUnited States

3

This study was supported by Grant No. EIA-9812240 of the National ScienceFoundation. Any opinions, findings, conclusions, or recommendations expressed inthis publication are those of the author(s) and do not necessarily reflect the view ofthe organizations or agencies that provided support for this project.

Coordinated by the Computing Research Association (CRA), 1100 SeventeenthStreet NW, Suite 507, Washington, DC 20036, Tel. 202-234-2111.

Additional copies of this report are available from CRA. Single copies are availableat no cost. To request pricing information for multiple copies or to place an ordercontact:

Computing Research Association1100 Seventeenth Street NWSuite 507Washington, DC 20036Tel. 202-234-2111E-mail: info@cra.org

Copyright 1999 by the Computing Research Association (CRA).

4

Table of Contents

Executive Summary 9

Chapter 1. Political Context 15

What Is This Report's Evaluation of Earlier Studies? 15

How Does Recent Legislation on II-1B Visas Affect Any Shortage? 19

Chapter 2. Information Technology Workers 25

What Is Information Technology? 25

Who Is an IT Worker? 29

Now Many IT Jobs Are There, and Where Are They Located? 34

What Skills Does an IT Worker Need In Order to Be Effective? 37

Why Is Information Technology Becoming So Prevalent In Om Society? 39

What Are the Characteristics of Information Technology

That Affect IT Labor? 41

Chapter 3. Demand, Constraints, and Consequences 45

What Are the Dynamics of the Marketplace and the Dangers

of Government Intervention In the IT Labor Market? 45

What Factors Umit the Ability of the Government, industry,

University System, and Professional Community to

improve the Match between Supply and Demand? 46

What Are the Costs of an IT Worker Shortage? 48

What Are the International Considerations in Dealing

with a National Worker Shortage? 49

Chapter 4. Worker Shortage 53

How Does One Determine Whether There Is a Labor Shortage? 53

is There a Shortage of IT Workers? 54

Where Are IT Worker Shortages Occurring? 68

Chapter 5. SupplyThe Degree Programs 71

What Are the Sources of IT Workers? 71

How Have Career Paths for IT Workers Changed Over Time? 73

What is the Role of High Schools in the Supply System? 75

What Is the Role of TWo-Year College Programs in the Supply System? 78

What is the Role of Four-Year College Programs in the Supply System? 81

What is the Role of Graduate Programs in the Supply System? 88

Chapter 6. SupplyThe Non-Degree Programs 99

What Non-Degree Programs Do Traditional Colleges and

Universities Offer? 101

What Other Groups Supply Non-Degree Programs? 102

What Is the Role of Corporate Universities in Training and

Educating IT Workers? 103

What is the Role of Distance Learning in Educating the IT

Workforce? 106

is Retraining Occurring, and if so, How Long Does it Take to

Retrain for an IT Job? 109

6

Chapter 7. Women, Minorities, and Older Workers 111

How Do Women Relate to the Worker Shortage? 111

How Do Minorities Relate to the Worker Shortage? 114

How Do Older Workers Relate to the Worker Shortage? 115

Chapter 8. Seed-Corn Issues 117

is the Strong industrial Demand for IT Workers Henning the

Educational System? 11 7

Chapter 9. Data Issues 121

What Are the Sources of Data on IT Workers? 121

What Are the Limitation of Existing Data on the IT Workforce? 124

Chapter 10. Recommendations 127

1. Federal and State Governments 127

2. Higher Education 132

3. Industry 138

4. Professional Societies 138

5. Individuals 140

Appendices:

A. Methodology 14 7

B. Related Studies 149

C. About the Authors and the Sponsoring Organization 151

D. Study Group Members 153

E. Computing Research Association Board of Directors 155

E External Reviewers 157

0. Acknowledgments 159

Executive Summar y

The purpose of this study isto improve understanding of thesupply of and demand for infor-mation technology (IT) workers inthe United States, and the surround-ing contextual issues. In conductingthis study, the authors receivedsupport from the National ScienceFoundation, collaboration from fiveother professional societies, andguidance from the ComputingResearch Association Board ofDirectors.

There are four majorcontributions in this study:

1. Evaluation of data. Thereport identifies and evaluates all themajor sources of statistical infotma-don relevant to this subject. Thestudy group found that federal dataare by far the most important andreliable, but that they have someserious shortcomings related tountimely reporting, occupationaldescriptions that are out of dateand based on ambiguous job titles,and incompatibilities betweensupply and demand data collectedby different agencies.

There are other data sources.However, it is questionable whether

results drawn from geographicallyrestricted and occupationallyrestricted data studies of IT work-ers can be generalized to thenational IT labor force; and manyof the national studies of ITworkers done by private organiza-tions have methodological weak-nesses (see chapter 9).

2. Definition of 'IT Worker'This report outlines a way ofdistinguishing IT workers from amuch larger class of workers whosejobs are enabled by informationtechnology. It then classifies ITworkers into four categories(conceptualizers, developers,modifiers/extenders, and support-ers/tenders) based not on their jobtitles, but on skills and knowledgerequired to do the job. An initialtest of this categorization scheme inthe State of Georgia has illustratedits value (see chapter 2).

3. Description of the SupplySystem. Some of the participants inthe national debate identifiedbachelor's degree training in com-puter science as the principal supplysource of IT workers. This reportpresents a detailed description of amuch more extensive supply system,

8 9

10

including not only majors in twentydifferent IT-related disciplines at theassociate, bachelor's, master's, anddoctoral levels, but also manypeople majoring in science, engi-neering, business, and even non-technical disciplines who often takesome course work in IT subjects(see chapter 5).

The supply system alsoincludes an increasingly importantand rapidly growing continuing-education element. This continuingeducation is supplied not only bythe traditional higher educationsystem through short courses andcertificate programs, but also byfor-profit educators and bycompanies that are employing theIT workers. New modes ofdelivering instruction, such asasynchronous learning over theInternet, are rapidly being deployed(see chapter 6).

4. Analysis of shortage claims.The report evaluates the question ofwhether there is a shortage of ITworkers in the United States. Thestudy group determined that thedata are inadequate to ascertainwhat mismatch there is, if any,between national supply anddemand. Therefore the reportmakes use of a variety of otherquantitative and qualitative kinds ofevidence. These include: secondaryindicators, such as wage growthand labor certificates awarded,based on federal data; quantitativestudies restricted to specific geo-graphical regions or specific IToccupations; private studies of thenational IT labor market whosemethodologies have come underquestion; anecdotal evidence abouthow employers have acted in theirsearch to recruit or retain workers,or take alternative solutions such asrefusing work or replacing workers

9

by machines; and other kinds ofqualitative evidence.

The preponderance ofevidence suggests that there is ashortage of IT workers, or at least atight labor market. None of thisevidence has the certainty of a directcount of supply and demand, andwithout this kind of direct count itis impossible to distinguish an actualshortage from a mere tightness inthe labor market. Moreover, thereare credible reasons for questioningthe evidentiary value of virtually anypiece of evidence that is available(see chapter 4).

One of the problems with thenational debate is that IT workershave been treated as a single,undifferentiated group. However,the phrase 'information technologyworker' encompasses many differ-ent occupations that require a widearray of skills and knowledge. Itwould be helpful in future discus-sions to segment the class of ITworkers into classes of occupationsthat have similar levels of knowl-edge and skill. It would be surpris-ing, given the dynamism anddemand in the IT labor market, ifthere were not some spot shortages(and some spot surpluses). Unfor-tunately, existing data do not allowthis kind of segmented analysis.

Of the many contextual issuesthat need to be considered to gain afull understanding of the supply anddemand of IT workers, this reportexamines four:

5. Political context. The studygroup evaluated the reports by theInformation Technology Associa-tion of America (ITAA) and theDepartment of Commerce, as wellas the criticism of these reports bythe U.S. General Accounting Office

(GAO). We agree with GAO thatthe low response rates in the surveysof industrial demand are a seriousweakness in the ITAA and Com-merce reports; but this speaksagainst the quality of the evidence,not necessarily against the conclusionthat there is a shortage. The ITAAand Commerce reports can also befaulted for their narrow focus onrecipients of computer sciencebachelor's degrees when discussingthe supply of IT workers (seechapter 1).

The legislation providing atemporary increase in the numberof temporary visas permittedannually under the H-1B visaprogram was also reviewed. CRAand the other professional societiesparticipating in this study did nottake a position on the H-1B increasewhen it was being debated in 1998,and it is not our intention tosecond-guess the program now. Webelieve that the legislation probablywill increase the total supply ofworkers during a period of epi-sodic higher demand created by theY2K problem, while limiting therisk to the indigenous supply systemthrough a sunset clause on theincrease in visas awarded (seechapter 1).

6. Types of demand. This reportdifferentiates two kinds of demand.There is episodic demand, such as thiscountry is experiencing currently asit struggles with the Y2K problemand the sudden spurt in Internetactivities. There is also a long-termdemand, created by fundamentalchanges in technology and society.The long-term demand for ITworkers is driven by the relentlessdecrease in the size of informationtechnology, as well as its relentlessincrease in performance, reliability,

flexibility, and price-for-perfor-mance. Because of these changes,information technology is rapidlybeing adopted by every sector ofAmerican society and made afundamental part of organizationaloperations and personal activity.Perhaps the greatest weakness in theH-1B legislation is that it focuses ona short-term problem created by anepisodic demand. It does notaddress the long-term problem ofproviding an adequate supply tomeet the demand for IT workersthat is likely to continue to growunabated well into the new millen-nium (see chapter 3).

7. Limitations on action. Evenwhen organizations recognize amismatch between supply anddemand that they would like toovercome, there are almost alwayslimitations on their ability to act. Agovernment organization cannotregulate supply and demand; it canonly provide incentives, such asfellowships, to encourage studentsto choose an area of expertise thatappears to be in short supply. But itis difficult for a government tostimulate labor supply by just theright amount since the market isconstantly changing, knowledgeabout supply and demand isimperfect and difficult to obtain in atimely fashion, and there are oftenunforeseen consequences of anygovernment action.

Industry has its own con-straints. For example, companiesare forced by short product life andshort product development cyclesto hire new employees or reassignexisting workers in ways that do notrequire a lot of break-in trainingbefore they can be productive.

11

12

The traditional higher educa-tion system is constrained by itsinability to change directions quickly.This results from its limited re-sources to allocate to new orgrowing disciplines, the long-termcommitment colleges and universi-ties make in buildings and capitalequipment or in tenured facultyappointments, and its deliberativestyle of decision-making (seechapter 3).

8. International consideration&There is a rising internationaldemand for information technol-ogy. Companies throughout theworld are taking advantage of thechanges in information technologythat allow them to use it to improvetheir operations. There is increasingglobal competition to supply ITproducts and services. Internationaland foreign firms are thus compet-ing with U.S. firms for the skilledworkforce to develop IT productsand services and place them intooperation. Only a few countries(e.g., India and Ireland) have asurplus of IT workers; and there isstrong competition from manycountries, including their homecountries, for these workers. TheUnited States will have to assure anadequate supply of IT workers if itwants either to retain its world leadin the IT sector, or remain competi-tive in other industry sectors thatrely on information technology.Foreign workers can play animportant role in the United States,but they are unlikely to help meetour growing demand for ITworkers in this manner (seechapter 3).

Other topics. This study couldonly touch on a number of othertopics that are important to ad-equately understand IT workforceissues. A number of groups are

11,

underrepresented in the ITworkforce and in the educationalprograms that prepare people forcareers as IT workers. Theseinclude women, Hispanics, AfricanAmericans, and Native Americans.If these groups were represented inthe IT workforce in proportion totheir representation in the U.S.population, this country would havemore than an adequate supply ofworkers to fill even the most direestimates of a shortage. Becauseother studies have investigated theunderrepresentation of women orminorities in professional work,scientific and engineering fields,including the computing profession,this study chose to focus its effortson other issues that have been lessthoroughly investigated. However,some basic information andstatistics about the issues concerningthe participation of women andminorities in the IT field have beencollected here. Some of the issuesconcerning women mentioned hereare the lack of adequate exposureto computers during the K-12 years,the scarcity of role models, andperceptions about what it is like tobe an IT worker (e.g., competitivework environment, focus onmachines rather than people, and soon). This study found many of theissues for minorities to be similar tothose for women; but there is theadded problem that minorities arenot as likely as white males orwomen to attend college or gradu-ate school. It is clear that theseissues of underrepresentation needmore attention than they could begiven in this study (see chapter 7).

It may also be true that olderworkers are underrepresented in theIT workforce. There is certainly awidespread perception that pro-gramming is an activity for theyoung, and that IT workers tend to

get "burned out" and leave the fieldby the age of 40. The absence ofalmost any data precluded this frombeing a major topic of study in thisreport. The study group looksforward to the examination of thisimportant issue in a forthcomingstudy by the National ResearchCouncil (see chapter 7).

Some people are concernedabout a seed-corn problem: thatthe high industrial demand for ITworkers is siphoning off too manygraduate students and faculty fromthe universities, leaving an insuffi-cient number to educate the nextgeneration of IT workers. Thisstudy detected preliminary signs ofa seed-corn problem. The coordi-nated efforts by government,industry, and academia to solve aseed-corn problem in computerscience that occurred around 1980are recounted. Some of theproblems today are the same as in1980, such as the rapid increase inundergraduate enrollments placingheavy loads on faculty. Howeverthere are also some differences:universities today have betterresearch equipment, comparedwith that in industry; than they didthen; however, research supportnow emphasizes short-termresearch too much, and it is harderfor industry to coordinate volun-tary restraint on faculty and studentraiding today because the industrialemployers of research computerscientists are today much morewidely spread across many differ-ent industries (see chapter 8).

The authors struggled withthe decision whether to includerecommendations in the report. Themandate for this study was toprovide an understanding of theissues surrounding the supply of

and demand for IT workers, notto provide a call for action. Inmost policy reports, the recom-mendations have primacy and theanalysis is included merely in asupporting capacity. The studygroup did not want the presenceof recommendations to under-mine the attention paid to theanalysis. Also, as a study group, wedo not have any particular standingwithin the government, industrial,or academic sectors from which torecommend actions.

On the other hand, a numberof important issues were raisedand actions suggested by the studygroup during the course of thestudy. Given the wide range ofknowledge and experience repre-sented by the study group, wedecided it would be useful to putthese suggestions forward in thehope that they will stimulatefurther discussion and action.Mostly, the recommendationsidentify a problem and a generalcourse of action without trying tobe specific about implementationmechanisms.

The thirty-seven recommenda-tions are grouped around a smallnumber of issues: data-collectionpractices, industry-academic coop-eration, industry hiring and trainingpractices, certification of educa-tional and training programs,broadening the supply pipeline,improving the research and teachingenvironment to retain and recruitfaculty, and curriculum develop-ment. The recommendations areorganized according to intendedaudience: government, highereducation, industry, professionalsocieties, and individuals (a summa-ry of the recommendations can befound at the end of chapter 10 asbox 10-1).

2 13

Chapter 1 Political Context

What Is ThisReport's Evalu-ation of EarlierStudies?

Human resource issues relatedto information technology havefrequently been on the nationalagenda since the early 1960s, whenthe National Academy of Sciences(NAS) and the National ScienceFoundation (NSF) began preparingstudies and collecting data on thissubject. These efforts supple-mented data that have long beencollected by the Bureau of LaborStatistics (BLS).

The most recent nationaldebate began in 1997 with a reportprepared by the InformationTechnology Association of America(ITAA), a trade association repre-senting 11,000 companies. ITAAreported a large shortage of ITworkers in large and mid-size U.S.companies. In 1998, ITAA pub-lished a second report based on alarger sample of companies, whichindicated an even more seriousshortage of workers. The Depart-ment of Commerce's Office of

Technology Policy then issued areport that largely mirrored theconclusions in the original ITAAreport. The General AccountingOffice (GAO) criticized the meth-odology used to gather the data putforward by both ITAA andCommerce, and questioned theirconclusions about a shortage.Individuals and private organiza-tions weighed in on the debate,which quickly crystallized aroundproposed legislation to increase thenumber of temporary H-1B visasthat could be awarded annually.Compromise legislation was passedby Congress in October 1998 andsubsequently signed into law.

However, it is clear that thislegislation does not end the nationaldiscussion of IT workers. Thelegislation is temporary, lasting onlythree years, and it addresses onlyone aspect of this complex laborissue. Questions about the treat-ment of older IT workers, forexample, were largely ignored in theH-1B debate. The importance ofIT workers to national competitive-ness and national wealth, and thedynamic nature of informationtechnology, which is rapidly beingincorporated into every sector of

1315

16

the U.S. economy, strongly suggestthat issues surrounding the ITworkforce will be part of nationalpolicy discussions for years tocome.

Although this report primarilyaddresses current and future issuesabout IT workers, this chapter isdevoted to an examination of someaspects of the political contextconcerning IT workforce issues inthe United States. This analysisbegins with a review of some ofthe arguments presented in theITAA and Commerce reports.

The discussion will focusprimarily on vacancy rates, whichare used as a principal kind ofevidence in the two ITAA studies.'The first study, undertaken by theCato Institute, a libertarian think-tank, involved telephone interviewswith randomly chosen companies(in various sectors, not just ITcompanies) that employed 500 ormore workers (counting all theirworkers, not just IT workers). Only271 of 2,000 companies responded,giving a response rate of 14percent. The second study, under-taken by the Continuing EducationDivision of Virginia Tech, was alsoa telephone survey, but this time thetargets were companies with 100 ormore employees. From a sample

size of 1,500 companies, 532interviews were completed, giving asomewhat better but still lowresponse rate of 36 percent. Thefirst study argued that 190,000 ITjobs were unfilled in the UnitedStates in 1996. The second studyclaimed 346,000 IT vacanciesrepresenting a 10-percent vacancyrate.

GAO argued that the lowresponse rates and small samplesizes made the results of thesesurveys highly questionable.2 It istrue that the results might beskewed because companies thatwere experiencing more seriousshortages may have been morelikely to respond. However, there isno hard evidence from any statisti-cally valid study that refutes eitherthe ITAA results or the presence ofan IT worker shortage. While theITAA results may not have theweight of scientific evidence, theydo suggest that many U.S. compa-nies are having difficulties fillingtheir IT positions. This evidence isconsonant with many other anec-dotal reports heard while research-ing this study.

This finding is bolstered by aGAO observation that the unem-ployment rate for IT workers wasonly 1.3 percent in 1997, which is

I "Help Wanted: The IT Workforce Gap at the Dawn of a New Century," InformationTechnology Association of America, Arlington, VA, 1997; "Help Wanted 1998: A Call forCollaborative Action for the New Millennium," Information Technology Association ofAmerica and Virginia Polytechnic Institute and State University, March 1998. For a moredetailed critique of the first ITAA report, see Burt S. Barnow, John Trutko, and RobertLerman, "Skill Mismatches and Worker Shortages: The Problem and AppropriateResponses," Draft Final Report, The Urban Institute, February 25, 1998. It persuasivelyargues that there are methodological problems with one after another of the kinds ofevidence of shortage given in the ITAA report.

2 See, for example, U.S. General Accounting Office, "Information Technology: Assessmentof the Department of Commerce's Report on Workforce Demand and Supply," GAO/HENS-98-106, March 1998; GAO, "Information Technology Workers: Employment andStarting Salaries," GAO/HEHS-98-159R, May 1998.

.4

much lower than the 4 percentoften considered by economists torepresent "full employment." Infact, even when the IT unemploy-ment rate hit its historical high forrecent times-3 percent in 1991itwas below the 4 percent thresholdindicating full employment. This isperhaps to be expected, given thatmost professional occupations thatare filled primarily by highly edu-cated workers have low unemploy-ment rates.' Since 1987 the ITunemployment rates have trackedup and down in parallel with thenational general unemployment rate,although the IT rates are consistentlylower by a factor of 2 to 3. Giventhat the general unemployment rateis currently at a 25-year low, it is notat all surprising that the IT labormarket is experiencing sometightness.

The Department of Com-merce report,' like the first ITAAreport, compares the projectednumber of IT jobs to the supply asan indicator of a shortage. Fordemand, Commerce used a Bureauof Labor Statistics (BLS) projectionthat between 1994 and 2005 thenumber of IT jobs in the UnitedStates would increase annually by95,000. Commerce contrasted the

projected annual increase in demandwith data from the National Centerfor Education Statistics (NCES).The NCES data indicated that only25,000 bachelor's degrees in com-puter science are produced annuallyin the United Statesand that thesenumbers have been steadily decreas-ing from a high of 42,000 in 1986(a 40-percent decline).

GAO rightly criticizedCommerce's argumentby notingthat there are additional kinds offormal computing training besidesbachelor's degrees, such as associatedegrees and certificate programs,that prepare people for IT careers;and also that most IT workersreceive their formal education infields other than computer science.In consonance with GAO's line ofreasoning, this study group believesthat to understand whether there isan IT worker shortage, one mustconsider both the multiple sourcesof supply and the various occupa-tions for which each of thesesources prepares an IT worker.There are many IT jobs, such ashelp desk attendant or Web de-signer, for which an undergraduatecomputer science degree is neithercommon nor appropriate training.'Another problem with the ITAA

3 According to a data brief from R. Keith Wilkinson, Science and Resource Studies, NSF,the unemployment rate for science and engineering degree-holders was less than halfthe overall national average: 2.2 percent for those working in science and engineering,and 2.8 percent for science and engineering degree-holders in other than science andengineering occupations, versus 5.6 percent for the U.S. labor force as a whole in 1995.See http://www.nsf.gov/sbe/srs/databrf/sdb/98325.htm for details.

4 U.S. Department of Commerce, Office of Technology Policy, "America's New Deficit: TheShortage of Information Technology Workers," Fall 1997.

5 Brian Hawkins, the president of EDUCAUSE, has noted that there are reasons why ITjobs that do not require a computer science degree are advertised with this degreerequirement: "Often HR personnel do not fully understand such issues all that well,when job specifications and job descriptions are written. Sometimes this is done in anattempt to artificially elevate the position deliberately, in an effort to get the positiongraded at a level that the salary could attract reasonable candidates, who may or maynot have such qualifications." Personal communication, March 8, 1999.

17

18

line of reasoning is that the declinein the production of bachelor'sdegrees in IT-related fields hasended. There was a very significantnational upswing in the number ofstudents entering bachelor's degreeprograms in computer science in1997 and 1998. The CRA TaulbeeSurvey of Ph.D.-Granting Depart-ments of Computer Science andEngineering indicates that newdeclarations of majors in computerscience at the bachelor's level havedoubled over these two years, andmuch higher graduation numberscan be expected as these studentscomplete their undergraduatedegrees. Enrollments still are not ashigh as they were at their peak adecade earlier, however.

As another indicator of ashortage, the Commerce reportcited the fact that some U.S. compa-nies were outsourcing work toother countries. GAO correctlyobserved that the statistical data tosupport this claim are not compel-ling. The Commerce report dis-cussed the size of the IT workforcein other countries, such as India, andmentioned percentages of thatcountry's IT work being done forexport. However, it was difficult toapply these statistics meaningfully asan indicator of a worker shortage inthe United States. It would havebeen useful to have statistics aboutthe percentage of IT work for U.S.companies being outsourcedoverseas and how that percentagehas changed over time. Thesignificance of even these statistics,however, would not be unambigu-ously clear. Outsourcing is a tried-

and-true method used by compa-nies for financial and other reasons,and the presence of outsourcingdoes not, in itself, represent evi-dence of a worker shortage.Outsourcing outside of the UnitedStates may be an indicator oflimited U.S. capacity to do thiswork, but it is hard to tell withoutfurther examination. For example,the lower cost of having program-ming done overseas or the desire ofa company to establish a presence ina particular country might bereasons for foreign outsourcing thathave nothing to do with a shortageof American workers.

The final argument given byCommerce as evidence of a workershortage is the increase in salaries ofITworkers. The report cites threeprivate studies of IT workercompensation that show annualsalary increases in 1996 and 1997averaging between 7.4 percent and20 percentmuch higher than theoverall average increase in compen-sation for U.S. workers (4.1 per-cent). GAO gives two counter-arguments. The first is that this maybe evidence of a tightening labormarket rather than an actual short-age of workers. On this point,GAO's argument may be focusedtoo much on the near term. Re-gardless of whether there is a'shortage' or merely a 'tightness'today, if the demand for ITworkers continues to grow rapidly inthe next decadeas the BLS itselfpredicts and our study groupexpectsany tightness would soonbecome a shortage, unless supply canfind a way to keep up with demand.'

6 BLS has projected that between 1996 and 2006 the IT-related occupations (i.e.,computer systems analysts, engineers, and scientists) will grow by 107.6 percent,compared with an overall job growth of only 14 percent in the United States. BLS, MonthlyLabor Review, November 1997.

16

GAO's second criticism wasthe use of BLS data to argue thatIT occupations have, at best, merelykept pace with average salaryincreases in the professional andspecialty occupations overall from1983 to 1997 (although IT workersstarted from a considerably higherbase salary). One problem with thisargument is that BLS data arestrongly at odds with private data,such as employer compensationsurveys. One reason may be thatthe private studies take bonuses andstock options into consideration,whereas BLS considers only basesalary. There is anecdotal evidencethat spot shortages have driven upsalaries. Two or three years ago, forexample, new graduates with abachelor's degree in computerscience were being offered startingsalaries up to double the average oftheir peers if they were experiencedin building local area networksanew skill that was then in shortsupply. Some schools are reportinga similar increase today in salaryoffers for students graduating withexperience in information security.

Most people interested inthis labor issue have focused on thedifferences between the positionstaken by Commerce and GAO, butthere are, in fact, many areas ofagreement. GAO has publiclystated, for example, that it agreeswith the Commerce report on pastand expected growth in demandfor IT workers, current lowunemployment rates for IT work-ers, and the need for a better

understanding of supply-demanddynamics and better categories anddata in order to develop goodpolicy.

How DoesRecentLegislation onH-1B VisasAffect AnyShortage?

This study group did not takea position for or against the legisla-tion that temporarily increased thenumber of H-1B visas that may beawarded annually.' Nevertheless, anumber of points can be madeabout the likely effects of thislegislation on the IT workforce.'

Given that there is a sunsetclause in the legislation that increasesthe cap, this legislation will clearlynot be a long-term factor unless it isrenewed.

The number of IT workersentering the United States under theH-1B program is and will continueto be a small percentage of the totalIT workforce in this country.According to BLS, there are ap-proximately two million IT workersin the United States. The limit on thenumber of workers of all typesentering this country on H-1B visaswas 65,000 per year. This limit willincrease to 115,000 for 1999 and2000, before dropping back to the

7 For a discussion of the legislation that increased the annual cap on H-1B visas, plus alabor perspective on the legislation, see Greg Gillespie, "Congress Expands H-1B VisaProgram," IEEE, The Institute, December 1998, p. 1.

8 For another view on H-1B, Y2K, and the larger, long-term issues of IT workers, see HowardRubin, "The United States IT Workforce Shortage," Dr. Dobb's Journal, Fall 1998, on the Dr.Dobb's Web site at http://www.ddj.com/articles/1998/9814/9814b/9814b.htm

19

20

65,000 level. H-1B workers enter anumber of occupations, such asbaker and fashion model; not all ofthem are IT workers. The bestestimate from unreliable data is thatfrom 20,000 to 25,000 of these65,000 visas have been going to ITworkers, and it is difficult to predicthow many will go to IT workers inthe future. Since the visas aretemporary, the workers can stay inthe country on this visa for only sixyears. In fact, a number leavebefore their visas expire. For thesereasons, H-1B workers are likely torepresent a small portion (perhapsless than five percent) of thenational IT workforce.

Of H-1B workers who doIT work, there is no good evidenceabout where they are likely to beemployed or what effect they willhave on the domestic IT labor pool.The specific kinds of IT work theywill do is unknown. The top tenusers of these visas in the past havebeen companies providing contractlabor and services. By far thelargest user has been the MastechSystems Corp. in Pittsburgh, whichhas received visas for 1,733 of itsemployees-80 percent of itsworkforce. These workers arevirtually all programmers who holda bachelor's degree as their highestlevel of education. However, thereis also some demand for H 1-Bworkers in more highly skilled,higher paying positions. Theyinclude research scientists in indus-trial research laboratories or facultymembers in research universities,where they serve the nation by

creating new technologies thatgenerate national wealth and bytraining the next generation of ITworkers. For example, Intel Corp.says that, of its 67,000 employees,about 3 percent have been hiredusing H-1B visas, and nearly 80percent of these hold a master's ordoctoral degree. There is no way todetermine whether the largernumber of temporary workersentering the United States under theexpanded program will continue tobe employed by companies likeMastech or by companies like Intel.

The cost of the visa, whichapproaches $15,000 when legal feesare included, limits the extent towhich H-1B visas will be used to filllower-skill positions.

The attention paid to theH-1B visa has obscured the exist-ence of another major source ofskilled foreign labor, including ITworkers. The TN visa (commonlyknown as the 'NAFTA visa')permits unlimited numbers ofCanadian IT professionals to enterthe United States for work, and itwill offer similar opportunities toMexicans beginning in the year2003. There is some evidence that abrain drain of IT workers fromCanada is hurting the Canadian ITindustry.'

Making the increase in thenumber of visas awarded under theH-1B program a temporarymeasure is well justified. It is verydifficult to accurately forecastnational shortages. Witness the NSF

9 On the IT worker situation in Canada, see the Software Human Resource Council's report,"Taking Action on Canada's IT Skills Shortage" (http://www.shrc.ca/Taking%20Action/front_page.htm); SHRC's Occupation Skills Profiles Model (the Canadian AdvancedTechnology Association's call to "double the pipeline" (http://www.cata.ca/cata/advocacy/skillshortage.htm); and the British Columbia Technology Industries Association report,"Technology Industries in BC: A 1997 Report Card" (http://www.bctia.org/industry/report/).

8

forecast in 1990 of a "shortfall" ofscientists and engineers in general,which did not materialize; in fact,many people now believe there is anoversupply. Generally speaking,legislation has never been a veryeffective tool for balancing supplyand demand in any occupation. It ispossible to harm the indigenoussupply system by bringing in toomany foreign workers. It is alsopossible that foreign workers arefilling positions that might otherwisehave been filled by Americancitizens, such as older electricalengineers who have lost their jobs inthe downsizing of the defenseindustry, although the study groupuncovered no hard evidence of this.A limited-term increase in thetemporary visa program willprovide an opportunity to studysome of these issues before anyadditional immigration legislation isconsidered.

The current legislation mayserve well to ameliorate the current,but episodic, demand for ITworkers caused by the Y2Kproblem by covering some of thetotal demand for IT workers (butnot necessarily by doing Y2K workitself). For reasons that are ex-plained elsewhere in this report, thedemand for IT workers willprobably continue to grow steadilyand rapidly over the comingdecades. In fact, because informa-tion technology affects so manysectors of American industry, thegrowth in demand for IT workersis likely to continue unabated, evenin times of general economicmalaise or when IT companies

themselves have a downturn. Thus,in the long term, the country willhave a large capacity to absorb newworkers from both domestic andforeign sources. Y2K is a specialand important problem, but it istemporary. It represents a sharp,short-term peak on the longupward curve of continuing, long-term demand for IT workers.There will be other short-termpeaks as well, such as the end-timedate for Unix systems early in thenext century. While eventually therewill be a sufficient growth indemand for IT workers that theU.S. economy could possiblyabsorb all of the foreign workershired to fix Y2K problems, it isunclear whether there will be sometemporary displacement of ITworkers after this short-term crisisis over. To lessen the risk ofdisplacement of American workers,it appears to be sound policy to fillsome of this short-term spike indemand for the Y2K problem withtemporary foreign workers.

To complete this discussion ofthe H-1B issue, it is useful tocompare this visa program withlegislation in 1989 dealing with laborshortages of nurses reported bysome hospitals and other employergroups. The history of this legisla-tion and its effects on the shortageof nurses is an illustrative case studyfor understanding the IT workershortage. In fact, this H-1A visalegislation for nurses provided thestructure for the legislation in 1990that created the H-1B visa program.The case study is recounted in box1-1.

21

Box 1-1. ii-1A Visas and the Nursing ShortageIn response to labor shortages of nurses reported by hospitals and

some other employer groups, the Congress passed the ImmigrationNursing Relief Act of 1989 (INRA, Public Law 101-238). This lawadded a new provision allowing admission of nonimmigrant registerednurses (RNs) during a five-year pilot period to expire in September 1995.

The history of this program was carefully reviewed in 1995 by theImmigration Nursing Relief Advisory Committee established under theabove law, and its report provides an instructive model for discussion ofthe issues surrounding the debate about foreign IT workers. The rationalesfor the H-1A nursing visas were:

a) reports of a nationwide shortage of nurses;b) increasing dependence on foreign temporary nurses admitted

under other visas;c) pending expiration of work authorizations for many existing

temporary foreign nurses admitted under other programs;d) concern that foreign nurses were detrimentally affecting the

pay and working conditions of the domestic nursingworkforce; and

e) declining numbers and quality of applicants to basic nursingeducation programs.

The new act, first, allowed foreign nurses previously admitted ontemporary visas to convert their status to legal permanent resident, andwaived numerical limits in existing law in order to allow this to happen.Second, it created a new temporary nursing visa (the H-1A visa) thatincluded provisions intended to: 1) encourage employers to reduce theirdependency on foreign nurses, 2) provide protection for the wages andworking conditions of nurses who are citizens and legal permanentresidents of the United States, and 3) foster the development of a stablepool of domestic RNs so that future shortages could be minimized.

According to the Advisory Committee, the debate on this particularlegislation embodied many of the issues that repeatedly arise in discussionsregarding the admission of foreign workers to meet skill shortfalls andlabor shortages in the United States. The Advisory Committee summa-rized these issues with a long quotation from a 1991 staff repore°

The debate about relying on immigration more significantly to meet"labor shortages," and thereby contribute to America's competitiveness inthe global marketplace, inevitably included the need to provide realisticprotection for U.S. workers. For immigration policy, this issue involvedtwo interrelated points. First, how to evaluate independently an employer'sclaim that a foreign worker is needed. And second, how to strike an

10 Gary B. Read and Demetrios G. Papademetriou, "U.S. Legal Immigration Reform: RecentDevelopments," Immigration Nursing Relief Advisory Committee report, 1995, pp. 12-13.

22

intellectually and politically satisfactory balance between being responsiveto employer needs while also being sensitive to concerns that greateraccess to foreign workers by U.S. employers might affect adversely thewages and job opportunities of U.S. workers. Such adverse effects couldoccur through direct displacement of U.S. workers or through significantinterference with the market's natural propensity to adjust to a tighterlabor supply, thereby leading to an increasing dependence on foreignworkers. This debate clearly raised the issue that over-reliance on immi-gration to meet labor shortages, as opposed to educating and training thedomestic workforce, could turn temporary labor market shortages intostructural deficiencies.

The conclusions of the Advisory Committee were mixed. Theynoted that because only a tiny percentage of the U.S. nursing workforceever came to be accounted for by H-1A nurses (about 13,800 in 1994,less than 10/0 of employed RNs), at the national level essentially all theeffects of this program were negligible. However, because the H-1Anurses were heavily concentrated in only a few metropolitan areas (overone-third in the New York City area alone, and two-thirds in New York,Chicago, Houston, Los Angeles, and Dallas together), H-lAs in thesecities mitigated a tight nursing labor market with "no adverse impacts onpatient care," but also "may have lessened the pressure to find long-termsolutions to nurse staffing problems." "

The Committee found the "attestation" procedures required ofemployers to be ineffectual, and reported that the "use of employer-specific vacancy rates as a justification for the need for H-1A nurses wasproblematic, as these rates could be calculated in several ways, makingthem difficult to verify." It noted further that the prevailing wage deter-minations were often of doubtful validity and reliability, and that the act'srequirement that "employers take timely and significant steps to recruitand retain U.S. nurses was ineffective because it did not require any newsteps" beyond those that most employers had long practiced.12

Lastly, the Advisory Committee reported that the rate of increasesin RN employment had slowed since passage of the 1989 Act; that pressreports had begun to appear about nurse layoffs; and that "the futurelabor market for registered nurses is uncertain."" The H-1A nursing visaprogram was allowed to expire in September 1995. In its final year,FY1994, approximately 6,300 nonimmigrant nurses had been admittedunder this visa program.

Source: Computing Research Association, Intersociety Study Group on InformationTechnology workers, based on the "U.S. Legal Immigration Reform: Recent Develop-ments," Immigration Nursing Relief Advisory Committee report, 1995.

11 Ibid., p. 5.

12 Ibid., p. 6.

13 Ibid., pp. 7, 31.

2123

Chapter 2. Information Technology Workers

What IsInformationTechnology?

This report uses the terms'information technology' and'information technology worker'because they are used in the nationaldiscussions about these labor issues.Unfortunately, these terms aresomewhat imprecise and are used indifferent ways at different times.Figure 2-1, taken from a dated butstill useful study by JohnMcLaughlin and Anne Birinyi, givesa broad definition of the 'informa-tion business.' From this figure, onecould infer a broad definition ofinformation technologies thatwould include, among many others,computers, telephones, radios,televisions, books, and filingcabinets. This report does not usesuch a broad definition.

In this discussion, informationtechnology (IT) refers only tocomputer-based systems. Itincludes computer hardware andsoftware, as well as the peripheraldevices most closely associated withcomputer-based systems. We define

2 2

'computer-based systems' broadlyto include the full gamut of techno-logical considerations, ranging fromthe design and production of chips(for example, Intel is widely re-garded as an IT company); throughthe design and creation of complex,computer-based systems for aparticular application (the modern-ization of the U.S. Internal RevenueService tax-processing system wascertainly considered to be an ITproblem); to the end-use of suchsystems (most of the electroniccommerce startup companies areconsidered to be part of the 'ITrevolution,' at least for the purposeof tracking and reporting).

There is a certain amount ofambiguity to this definition. Toclarify, it may be helpful to compareit with some other commonly usedterminology and concepts. Theterm 'information system' issometimes used to refer to com-puter-based systems that provideinformation for decisionmaking inorganizations, which results in theuse of 'information technology' and'information systems' in closelyrelated ways. This usage (e.g., "Heheads up the corporation's IToperations") focuses on the purpose

25

26

Figure 2-1

THE "INFOF

U.S. MAILPARCEL SVCSCOURIER SVCS

OTHER

DELIVERY SVCS

PRINTING CO'SLIBRARIES

RETAILERS

NEWSSTANDS

PRINTING ANDGRAPHICS EQUIP

COPIERS

CASH REGISTERS

INSTRUMENTS

TYPEWRITERSDICTATION EQUIPFILE CABINETSPAPER

TELEPHONETELEGRAPH

MAILGRAMIRC'S

SCC'SVAN'S

CABLE OPERATORS

MULTIPOINT DIST. DVCS

SATELLITE SVCSFM SUBCARRIERS

PAGING SVCS

RADIOSTV SETSTELEPHONESTERMINALSPRINTERSFACSIMILEATM'SPOS EQUIPANTENNASFIBEROPTICS

CALCULATORSWORD PROCESSORSPHONO'S, VTR'S, VIDEO DISC

MICROFILM MICROFICHEBUSINESS FORMS

INDUSTRY NETWORI.q

DEFENSE TELECOM SYSTEMSSECURITY SVCS

PABX'S

TELEPHONE SWIMODEMS

CONCENTRA

MULTIPLEXERS

TEXT EDITIN

COMMUNICA

MASS STOR

CONDUITATM - Automated Teller MachinesIRC - International Record CarrierPABX- Private Automatic Branch ExchangPOS Point-of-Sale

Source: John McLaughlin and Anne Birinyi, "Information Business," Harvard College, 1980.

23

TION BUSINESS"

DCAST NETWORKSE NETWORKSDCAST STATIONS

NEWS SERVICE

DATA BASESTELETEXT

TIMESHARING SERVICE BUREAUS

SOFTWARE SVCS

PROFESSIONAL SERVICES

FINANCIAL SVCS

ADVERTISING SVCS

ON-LINE DIRECTORIES

UTERS LOOSE-LEAF SVCS

G EQUIPMENT

SOFTWARE PACKAGES

NEWSPAPERS

IP SHOPPERS

WP'S

CONTENTSCC Specializied Common CarrierVAN Value Added NetworkVTR Video Tape RecorderWP Word Processor

27

28

Table 2-1

IT-related Academic Disciplines Offered in the United States

1. Computer Science2. Information Science3. Information Systems4. Management Information

Systems5. Software Architecture6. Software Engineering7. Network Engineering8. Knowledge Engineering9. Database Engineering10. System Security and Privacy

11. Performance Analysis (CapacityPlanning)

12. Scientific Computing13. Computational Science14. Artificial Intelligence15. Graphics16. HCI (Human Computer Interface)17. Web Service Design18. Multimedia Design19. System Administration20. Digital Library Science

Source: Peter Denning, "Information Technology: Developing the Profession,"Discussion Document, December 4, 1998.

of the system rather than its under-lying technology. In this report, theunderlying technology of aninformation system (as used in theexample above) is considered to bean example of information technol-ogy. The definition of informationtechnology, however, is not re-stricted to any particular applicationarea. Indeed, one of the attributesof information technology thatmakes it worthy of study is itspervasiveness in society.

There may be as many astwenty academic specialties thatstudy various aspects of informa-tion technology and its use andapplications (see table 2-1). Box 2-1and the accompanying text inchapter 5 lists and reviews defini-tions of nine of these disciplinescompiled by a National ResearchCouncil (NRC) study panel in theearly 1990s. Only the three mostpopular IT-related disciplinescomputer science, computerengineering, and informationsystemswill be considered here.

25

In a strict sense, computer science isfocused on the study of algorithms,the software that implements them,the properties of computers, andthe processes for creating thesetechnologies. Computer engineeringtraditionally has focused on theengineering of the components andhardware systems that make up acomputer. In this strict sense, itfocuses on the underlying technol-ogy that implements computerhardware. Information systems,although less well defined as adiscipline of study, has focusedinstead on the use of computertechnology for end-purposesrelated to decisionmaking of somekind. All three of these disciplinescapture some aspects of what weregard to be information technol-ogy, but none of them covers allaspects.

In the past decade, and evenmore rapidly in the past five yearswith the spread of the Internet, therapid merging of traditionalcommunications and computer-

Box 2-1

Undergraduate Degree Programs in Information Technology

Computer engineering - Graduateswork primarily in computer hardware.

Computer science and engineering -Graduates work primarily in hard-ware, firmware, and software,depending on program and choicesmade by the student.

Computer science - Graduates workprimarily in software design andimplementation.

Software engineering - Graduateswork with the engineering ofsoftware, with special attentiondevoted to large and critical systems.

Computer information science -Graduates work on the developmentof information systems, probably

with more emphasis on information asan enterprise resource than is given inprograms in computer science orsoftware engineering.

Information systems Graduatesdesign, develop, implement, andmaintain business informationsystems.

Management information systems -Graduates design, develop, implement,maintain, and manage informationsystems with a greater emphasis onthe management of the systems thanon the other aspects.

Information science - Graduatesusually work in libraries or developother facilities to provide informationto users.

Source: Adapted from "U.S. Degree Programals Changing Needs for the 1990s, National

s in Computing" in Computing Profession-Academy Press, 1993.

based systems has added to theconfusion. Although most telecom-munications technology has beencomputer-based for some years, therapid miscegenation of the func-tionality of computers and oftraditional communications systemshas come to the forefront. Theability to make telephone calls overthe Internet or make computationsvia a Web page devoted to aparticular topic, or the provision ofgreatly increased content (such asbank account information) using atraditional telephone hookup, areexamples. There is no preciseboundary between informationtechnology and telecommunicationstechnology. Some cases, such as the

provision of enhanced, computer-based information services as partof standard telephone service,probably should be consideredinformation technology; others,such as installing telephone lines inhomes or fiber cables under theocean, may not be.

Who Is an ITWorker?

Defining an IT worker iscomplicated, not only becauseinformation technology itself is notclearly defined. A wide range ofoccupations might be considered IT

26 29

work. They vary enormously in thetechnical and other skills required todo the job. These jobs are notlocated solely in the IT industry (theindustry whose primary business isto make and sell IT devices, soft-ware, services, and systems), andthey do not always involve thedesign and creation of informationtechnology artifacts. Instead, theyare distributed throughout virtuallyevery sector of society, includinggovernment, all sectors of industry,and most nonprofit organizations;and they may involve many peoplewho propose, implement, enhance,and maintain systems that rely uponinformation technology. Not everyjob in an IT company is necessarilyIT work (Are the janitors at IBM ITworkers? We think not). Many jobsinvolve some contact with informa-tion technology, but not all wouldbe considered IT jobs; otherwise,this category would soon becomeso large as to be useless.

It is not surprising that thedifferent studies of the IT workershortage have employed differentdefinitions. As the Department ofCommerce report noted:"

What is an IT worker? Itdepends on whom you ask. In abroad sense, the term 'informationworker' can be applied to data entrypersonnel, auto mechanics who usecomputer diagnostic equipment,medical technicians who operate CATscan equipment, and loan officerswho use computers to assess

creditworthiness, as well [as]computer programmers, systemsanalysts, and computer scientistsand engineers.

Commerce used the narrowdefinition of the Bureau of LaborStatistics classifications: computerscientists and engineers, systemsanalysts, and computer program-mers. The Information TechnologyAssociation of America (ITAA)used a broader definition: anyskilled worker who performs anyfunction related to informationtechnology, which itself is definedas the "study, design, development,implementation, support or man-agement of computer-basedinformation systems, particularlysoftware applications and computerhardware.""

The General AccountingOffice (GAO) has noted how thelack of a good definition has causedproblems in making good policy: "

The GAO and CommerceDepartment research into the ITindustry labor issue reveals that it isnecessary to make a distinctionbetween the IT industry as a wholeand the various occupations withinthe industry. This distinction is oftenoverlooked or is not clear in thedata; there is often difficulty inidentifying peop/e who are workingin IT occupations if they are notworking for an IT business. If oneasks what government or companiesshould do about the IT labor issue,

14 U.S. Department of Commerce, Office of Technology Policy, "America's New Deficit: TheShortage of Information Technology Workers, Fall 1997, p. 3.

15 "Help Wanted: The IT Workforce Gap at the Dawn of a New Century," InformationTechnology Association of America, Arlington, VA, 1997, p. 9.

16 Carlotta Cooke Joyner, "Is There a Shortage of Information Technology Workers?"Symposium Proceedings, The Jerome Levy Economics Institute of Bard College, June 12,1998, p. 4.

30 27

the answers will be more apparent ifthe question is phrased more clearly.There is a substantial difference insalaries, employment opportunities,and labor supply by IT occupation. Itis difficult to compare statistics thatexamine these issues because thestudies use different definitions ofoccupations and therefore come upwith widely different estimates ofstarting salaries, job vacancies, andlabor supply.

An NRC report on computerprofessionals written in the early1990s called for a simple classifica-tion scheme, which has yet to besupplied.'7 Jane Siegel from theSoftware Engineering Instituteindicated:'8

I would be thrilled if in the nextmajor national surveys...they didnothing more than simply have alogical, simple structure that broke outpeople doing computer-relatedwork...If I could get even very roughestimates of the degree field andsome simple demographics aboutwho these people are and a little bitabout their turnover rate and whatthey do in life, I would have a wholebody of knowledge that I think wouldhelp a large set of our users.

Alan Fechter from the NationalAcademy staff, following up Siegel'ssuggestion, "cautioned that although amoderate level of detail may bevaluable for corporate planning, agreater level of aggregation may beappropriate for purposes of nationalplanning and estimation."9

This report will not attempt toprovide the ultimate definition of anIT worker. However, two categori-zations are presented that the studygroup believes can help with nationalplanning and estimation. The firstdistinguishes IT workers from otherkinds of workers who may some-times use information technology intheir jobs (see figure 2-2). Each IT-related occupation is located at asingle point on the graph. As onemoves from left to right, the occupa-tions require increasing amounts of ITknowledge. As one moves frombottom to top, the occupationsrequire increasing amounts of domainknowledge (knowledge of businesspractice, industry practice, technicalpractice, or other kinds of knowledgeparticular to an application domain).The diagonal line separates the IT-related occupations into two classes,depending on whether IT knowledgeor domain knowledge is moreimportant. If more than half thevalue provided by a worker involveshis or her IT knowledge, then thisperson is considered to be an ITworker. If the person's occupationinvolves the use of informationtechnology but it adds less than halfthe added value to the work, then weregard the person as an Irenabledworker. A few occupations areplotted on the exhibit, as examples.

The second categorizationfocuses only on the IT workers.Table 2-2 differentiates four catego-ries of IT workers, depending onthe principal functionality in theiroccupation. The table includes

17 The workers addressed in the NRC studythe "computing professionals"probablyconstitute a slightly narrower class of occupations than are addressed in this study.National Research Council, Computing Professionalschanging Needs for the 1990's,National Academy Press, 1993.

18 Ibid., p. 18.

18 Ibid.

2 8, 31

Figure 2-2

Distinguishing IT Workers from 1T-Enabled Workers

Source:

A

Marketing CFOVP

BankTeller

IT-Enabled Workers

ProductDeveloper

BusProjectMgr

CallConsultant

SW.ProjectMgr

ApplicationDeveloper

SystemA dmin

CIO

IT Workers

OSDeveloper

Information Technology Knowledge

cro

Computing Research Association, Intersociety Study Group on InformationTechnology Workers, April 1999.

32

examples of particular IT occupa-tions that would fall under each ofthe four categories (conceptualizers,developers, modifiers/extenders,and supporters/tenders).2°

This approach is somewhatdifferent from the Standard Occupa-tional Classification (SOC) schemeused by the Bureau of LaborStatistics (BIS),21 in which thecategories are essentially a distillation

of job titles. This study found itdifficult to classify workers on thebasis of what they are called, at leastin a way that is helpful to makingpolicy. It decided instead to return tofirst principles and figure out whatthe workers do.

The categorization is built froma developmental perspective of theworld. It is based on an experienceand familiarity with the IT industry,

20 Professor Daniel Papp of the Sam Nunn School of International Affairs at the GeorgiaInstitute of Technology tested these categories in a survey of information technologyeducation programs he conducted. He found that the respondents were able to use thiscategorization easily and that it seemed to have value for grouping the IT workforceissues occurring in Georgia. See Papp, "ICAPP Information Technology StrategicResponse Educational Capabilities Inventory," draft report, December 16, 1998.

21 The SOC categorization scheme was in the process of being updated in 1998, but itwas not clear that the revisions would meet the criticisms lodged in this report. SeeOffice of Management and Budget, 1998 Standard Occupational Classification Revision;Notice, Federa/ Register, August 5, 1998.

9

Table 2-2

Categorization of IT JobsConceptualizers - those who conceiveof and sketch out the basic nature of acomputer system artifact:

EntrepreneurProduct designerResearch engineerSystems analystComputer science researcherRequirements analystSystem architect

Developers - those who work onspecifying, designing, constructing, andtesting an information technologyartifact

System designerProgrammerSoftware engineerTesterComputer engineerMicroprocessor designerChip designer

Source: Computing Research Association,Technology Workers, April 1999.

Modifiers/Extenders - those whomodify or add on to an informationtechnology artifact:

Maintenance programmerProgrammerSoftware engineerComputer engineerDatabase administrator

Supporters/Tenders - those whodeliver, install, operate, maintain, orrepair an information technology artifact

System consultantCustomer support specialistHelp desk specialistHardware maintenance SpecialistNetwork installerNetwork administrator

Intersociety Study Group on Information

where the workers are responsiblefor creating IT artifacts. However,this categorization should also applyreasonably well to all kinds of ITworkers in all sectors of theeconomy (i.e., to those who develop,use, and maintain systems driven byinformation technology), and itshould provide insight into currentpolicy issues regarding supply anddemand.

This belief is bolstered by thefact that there is a reasonably goodmatch between level of formal

education and category of worker.Table 2-3 maps formal educationonto the four categories. There isnot an exact one-to-one correspon-dence between educational degreeand category of work. However,the exhibit clearly shows a correla-tion.22 Occupations that fall underthe conceptualizer category arecommonly populated with recipi-ents of master's or doctoraldegrees. Occupations that fallunder the developer or modifiercategories are usually filled bypeople with bachelor's or master's

22 This correlation breaks down, however, in the case of the earliest stage ofconceptualization of an IT system, where the initial functional idea often comes frompeople with little IT education, but great applications knowledge.

30- 33

34

Table 2-3

Typical Educational Preparation

GUI@taCCD

for IT Jobs

1:03321:1500 Bachelor's fkihea3Vb Doctorate

Conceptualizers / / V V VDevelopers V V /Modifiers / V V if

Supporters / V V

Unlikely= (blank) Occasionally= .4 Common= V Frequent= si,

Source: Computing Research Association, Intersociety Study Group on InformationTechnology Workers, April 1999.

degreesand in the case of themodifier category, sometimes bypeople with associate's degrees.Supporter occupations tend to befilled most commonly with peopleholding an associate's degree, orperhaps only a high school diploma.Chapter 6 discusses at length thefact that an increasing percentage ofIT worker training is providedoutside of formal degree programs.This kind of training providesvaluable knowledge of specifictechnologies, company culture, andthe practices within that industry. Italso hones skills such as communi-cations, teamwork, and self-learning.However, there is some questionwhether it can adequately replace thefoundational knowledge acquired inthe formal degree programs, whichis critical preparation for at leastsome IT occupations.

In this report, the term 'ITworker' is used throughout andalways in the general sense described

above. However, many of thesources cited either use alternativeterminology ('computer profession-als,' `computer scientists,' 'computerand information scientists,' etc.) orthey have a different meaning of 'ITworker' in mind. In these cases, thesource's terminology is generallyused, and an effort is made toclarify the intended meaning in thecontextual discussion.

How Many ITJobs Are There,and Where AreThey Located?

Table 2-4 shows the numberof IT workers in the United Statesand the annual percentage change inemployment, using data fromBLS.23 Over the period 1988 to1997, employment in the IToccupations (as they define them)grew from 1,259,000 to 2,063,000

23 These statistics are taken and adapted from Exhibits 7 and 8 in Burt S. Barnow, JohnTrutko, and Robert Lerman, "Skill Mismatches and Worker Shortages: The Problem andAppropriate Response," Draft Final Report, The Urban Institute, February 25, 1998.

31

Table 24

YearNumber of

Workers Annual %(thousands) Change

1988 1,259

1989 1,366 8.5

1990 1,411 3.3

1991 1,422 0.7

1992 1,435 0.9

1993 1,583 10.3

1994 1,687 6.6

1995 1,703 0.9

1996 1,863 9.4

1997 2,063 10.7

Source: Adapted from Burt S. Barnow,John Trutko, and Robert Lerman, "SkillMismatches and Worker Shortages: TheProblem and Appropriate Responses,"Draft Final Report, The Urban Institute,February 25, 1998, Exhibits 7 and 8. Basedon the Bureau of Labor Statistics data.

jobsa 64-percent increase. Thiscan be compared with an increase of29 percent in all professional jobsand an increase of only 13 percent inthe total workforce during this time.Over this period, IT jobs increasedfrom eight to eleven percent of allprofessional jobs in the United States,and from 1.1 percent to 1.6 percentof all jobs in the United States.

As figure 2-3 shows, the vastmajority of IT jobs as reported byBLS are in one occupationalcategory (Computer SystemsAnalysts and Scientists). Over theperiod 1988 to 1996, this categoryhas grown much faster (158 percent)

than the category of ComputerProgrammers (9.8 percent), while thecategory of Operations and SystemsResearchers has dropped by 4.3percent. From 1988 to 1996, thenumber of Computer Programmersdropped from 570,000 to 561,000,but in 1997 the number jumped to626,000 (an 11.6 percent increase inone year). This may be an artifact ofthe temporary demand created bythe Y2K problem.

The IT industry (that is, thecollection of companies that pro-duce IT products, services, orsystems as their principal business) isone of the largest and most dynamicindustries in this country. Thenumber of workers in the computerand software industries has almosttripled in the past decade. However,this sector is by no means the onlyindustrial sector in which informationtechnology is being produced orused, or the only place where ITworkers are employed. Indeed,there are IT workers in virtually everysector of American society. Infor-mation technology is rapidly beinginfused into the financial, retail,manufacturing, service, entertain-ment, transportation, and otherindustries; and numerous IT workersare going to work for companies inthose sectors (see table 2-5 for someexamples of the use of informationtechnology in American industry). ITwork also occurs in every geographicregion of this country, not just inhigh-tech centers such as SiliconValley or Route 128 in Massachu-setts!' Thus a shortage of ITworkers affects not only the ITindustry (hardware, software,

24 While there are IT jobs in every American community, there are also concentrations of ITcompanies in a number of locales. Silicon Valley is well known, but there are up-and-coming concentrations of IT companies and IT workers in a number of other places, suchas Austin, Texas; Champaign-Urbana, Illinois; Salt Lake City, Utah; and the Washington, DCarea. See Steven Levy, "The Hot New Tech Cities," Newsweek, November 9, 1998, pp. 44-56.

3 2: 35

36

Figure 2-3

Comparison of IT Employment 1988 and 1997

1988 Total IT Employment=1,259,000

ComputerProgrammers

570,00045%

1997 Total IT Employment=2,063,000

ComputerProgrammers

626,00030% f

Operations ResearchersSystems Analysts

201,00010%

Computer SystemsAnalysts and Scientists479,00038%

Operations ResearchersSystems Analysts210,00017%

Computer SystemsAnalysts and Scientists1,236,00060%

Source: Adapted from Burt S. Barnow, John Trutko, and Robert Lerman, "Skill Mis-matches and Worker Shortages: The Problem and Appropriate Responses," Draft FinaReport, The Urban Institute, February 25, 1998, Exhibits 7 and 8. Based on theBureau of Labor Statistics data.

computer systems, computerservices firms), but virtually everysector of the American economy."The Y2K problem drives this pointhome. Computers are so firmly

woven into the fabric of organiza-tions that Y2K is a problem foralmost every corporation and everygovernment organization, and everymember of society is affected.

25 See, for example, One Digital Day, Intel, 1998, for a snapshot of the many applicationsof computing.

r4%.1 ".)

Table 2-5

Use of Computer Systems in the Operation of American Industry

Inventory management by largeretailersShipping scheduling and qualityassurance by express courierservicesFinancial controls in virtually everylarge businessFrequent flyer programs by theairlinesCredit card validation by merchantsProduction of movies and videosDistance education

Source: Computing Research Association,Technology Workers, April 1999.

Control of manufacturing lines in thechemical and automobile industriesProcessing data for oil explorationcompaniesGlobal positioning systems used inthe trucking industry and inscientific agricultureLiterature searching in biomedicalresearchComputer-aided design by engineersAutomated switching in thecommunications industry

Intersociety Study Group on Information

What SkillsDoes an ITWorker Need inOrder To BeEffective?

An effective IT worker needsa variety of skills, including technicalknowledge about informationtechnology, business knowledge andexperience, and organizational andcommunications skills. The mix ofskills needed varies greatly from oneIT occupation to another. Forexample, a person doing IT workfor a producer of householdappliances will probably need toknow more about production andaccounting than an IT worker whois building general-purpose softwareutilities for a company in the IT orcommunications industry. It isimpractical to present a completeset of skills needed for all IToccupations, or even for a single IToccupation. But it is possible tomake a few observations, using

examples from software-relatedoccupations.

In the technical area, thereare skills as well as knowledge to beacquired. A programmer needs toknow how to design, program, test,debug, and modify programs.Someone specializing in operatingsystems would need to know howto analyze basic hardware opera-tions and how to deal with complexcommunications situations. In theperformance-testing area, a goodknowledge of statistics is useful.A worker would need to knowhow to measure and analyzeperformance information andhow to modify programs toimprove performance. In theproject management area, theworker would need to understandthe project management modelfor code development and testing,be familiar with industry standardssuch as ISO 9000, and be able toestablish requirements and functionalspecifications. In the project estima-tion area, a worker would need the

37

38

ability to determine how long aproject will take, what resources willbe needed, and what dependencieson others need to be satisfied.

IT workers also need to havebusiness skills and experience, whichin many ways are similar to thoseneeded by people in other serviceprofessions. Many workers needthe ability to formulate projectbudgets, set tasks within thosebudgets, and complete work withintime and budget. A worker mayneed to be familiar with specificapplication programs, such ascorporate and industry databases,operation support programs, ormanufacturing support programs.There is a need for knowledgeabout the specific applicationindustry and its vocabulary, such asknowing who the leaders are andkeeping up with industry standards.The worker needs to have knowl-edge of the customer's concernsand how to meet them; for ex-ample, how the customers use theIT product, their need for futureproducts, and how current projectsmeet these needs. Finally, theworker needs to know the businesspractices of the employer: howprojects are started, led, andterminated; the steps in getting aproduct to customers; and howcustomer needs are collected,distilled, and spread through thecompany.

IT workers also need commu-nications and organizational skills,similar to those required of anyworker involved in technical projectdevelopment. There are teamworkskills, such as the ability to workwith others who have diverseeducations, skills, backgrounds, andcultures; to understand the functionof each team member; and torespect the strengths and limitations

3 5

of others. The worker needs to beable to organize and present technicalmaterial to technical peers, manage-ment, and customers. Non-technicalskills relating to technical specifica-tions and documentationsuch asthe ability to work in a team todesign a project implementation, theability to write a clear description ofthe job of each person on the team,and the management skills todelegate tasks within the teamarealso required. The worker must beable to work to and revise specifica-tions, and work to deadlines. Finally,the worker should be able accuratelyto estimate rates of progresstowards goals and be able to reportproblems early. Table 2-6 shows thatdifferent kinds of IT jobs requiredifferent mixes of these technical,business, and communications skills.

Where does one acquire theseskills? University programs incomputer science traditionally havetaught some of the technical skills,but not the business and communi-cations skills (although studentssometimes acquired some of theselatter skills through internships). Theaccreditation bodies CSAB andABET currently make communica-tion skills a required component ofan accredited computer science andcomputer engineering program, butthe accrediting organizations do notspecify the department that is toteach the skills or how they are tobe applied. Indeed, almost noaccredited program's departmentactually teaches communicationsskills as a separate course, althoughmost of them require the applica-tion of these skills in some com-puting courses. These skills maynot yet be emphasized as much asindustry would like, but the trend isin the direction sought by industry.Business skills are not well ad-dressed in computer science or

Table 2-6

Typical Knowledge, Skill Mix for IT Jobs (scale: 1-4)

Bommunicati-informationTechnology

Business EfidIndustry 01 MO

Organization

Conceptualiziers 4 2 3

Developers 3 2 3

Modifiers 2 3 3

Supporters 1 2 3

Scale: 1- least important; 2- moderately important; 3- important; 4- critically important.

Source: Computing Research Association, Intersociety Study Group on InformationTechnology Workers, April 1999.

computer engineering programs,but they are addressed in informa-tion systems programs.26 Graduatesin other majors generally gain lesstechnical training, but often they geta better introduction to communi-cations skills and even industryknowledge. Technical schools andself-help courses tend to focus onthe technical skills. Corporate trainingprograms often focus on all three.

Why Is Informa-tion Technol-ogy BecomingSo Prevalent inOur Society?

There are many reasons whyinformation technology has becomeso prevalent in the modern world,and there is no indication that thesereasons will fade any time soon. Theprocessing power and storagecapacities of semiconductor devices,which are the building blocks ofinformation technology, have beendoubling every eighteen months forthe past thirty years. At the sametime, prices have continued to

26 For an example of an attempt to introduce communications, teamwork, conflictresolution, and ethics into the information systems curriculum, see John Lamp, ChrisKeen, and Cathy Urquhart, "Integrating Professional Skills into the Curriculum," http://www.man.deakin.edu.au/jw_lamp/acse-96.pdf ; also see "IS '97 Model Curriculum andGuidelines for Undergraduate Degree Programs in Information Systems," The DATA BASEfor Advances in Information Systems, Vol. 28, No. 1, Winter 1997, ACM Special InterestGroup on Management Information Systems (SIGMIS).

3 6 39

40

decrease. This means that informa-tion technology, which can beprogrammed to do practicallyanything, has become embedded inmany kinds of organizational andphysical systems. IT products havebecome commodity items. General-purpose semiconductor devices, suchas the microprocessor, can now beused for millions of differentpurposes, leading to economies ofscale. These devices are much morereliable than the mechanical, vacuumtube, and transistor devices that theyreplace. They add value to theproducts in which they are used,and they reduce the need for humanusers to do dangerous or boringtasks. Vast improvements over thepast decade in the connectedness ofcomputers and in human-machineinterfaces have driven new uses.

What does this growingprevalence of information technol-ogy in society mean for IT laborissues? It appears as thoughinformation technology will be anincreasingly important part of theU.S. national economy for many yearsto come. Although there are likely tobe increases in productivity throughnew technologies and other means,the production of informationtechnology will continue to rely on alarge and growing force of workerswho require high levels of skill andknowledge to do their jobs effec-tively. An inadequate supply of suchworkers will have harmful effects onthe economy and the wealth of thenation. Any tightness in the labormarket is likely to become ashortage within a few years, as thedemand for information technol-ogy-based products and servicesgrows. From a policy perspective,the focus needs to be not only onachieving a proper match betweensupply and demand today, but alsoon how the nation will supply the

37,

growing number of appropriatelytrained IT workers in the future.

What Are theCharacteristicsof InformationTechnologyThat Affect ITLabor?

Information technology has ashort life cycle. Figure 2-4 illustratesthis short life cycle by charting therevenue earned by the Hewlett-Packard Company during fouryears in the mid-1990s. It showsthat new products introduced inone year earn their greatest amountof revenue in the following year,and that by the second year aftertheir introduction their contributionto the company's revenue streamhas already diminished significantly.In fact, nearly two-thirds ofHewlett Packard's revenues arederived from products introduced inthe previous two years. This is trueof many other companies as well.The demands of competition, andthe opportunities presented bytechnological advances, have driventhe introduction of new productsevery few months, and an almostcomplete turnover of the productline in four years. This rapid turn-over in technology makes it impera-tive that IT workers adapt to newtechnologies and new products.This means that they must continu-ously work at keeping their skillsand knowledge up to date or riskbecoming obsolete and unemploy-able.

The fluidity of the ITworkforce gives labor a powerover management. While some ITworkers have gained knowledge

Figure 24

The Importance of New Products to Company RevenueThe Case of Hewlett Packard

0 '97

'96

0 95'94

0 93 & Prior

NMI

Nearly two-thirds of HP'sorders are derived fromproducts introduced inthe last two years.

1994 1995

Source: Hewlett Packard

1996 1997

about particular application areasthat represents a valuable asset intheir work, there is nothing inher-ently application-specific about theinformation technology itself. ThusIT workers are not generally boundto specific industries. There is,however, a great variation in theproductivity of IT workers.Especially in the software area, thebest workers can be as much as tentimes as productive as the leastproductive workers in the samecompany. Figure 2-5 presents onemeasure of this wide variation inthe productivity of softwareworkers. These kinds of variationsoccur among IT workers evenwhen the labor market is notparticularly tight. In a tight market,companies have to settle for lessqualified programmers, and theeffects of this variability in workerproductivity may hit them harder.Companies may be fully staffed butnevertheless suffer greatly inproductivity.

3

IT workers often have strongpreferences about the kind ofemployer they wish to work for.Given the wide availability of jobs,many IT workers are more willingto change employers than areworkers in many other occupations.They have little sense of being"locked in." Employers who aredeemed less attractive because ofthe nature of their work, the salariesthey pay, or the culture of theirorganization are more likely thanother employers to experience ITworker shortages or to employunder-skilled IT workers. Figure2-6 illustrates the hierarchy ofemployment. This is grim news fororganizations at the bottom of thepyramid, which often includesgovernment organizations.

IT work is stratified, and thereis much greater demand formanagers and other workers withsystem-level skills than for 'assemblyline' programmers. The "averageannual level of change in employ-

41

42

Figure 2-5

Variability in Programmer Productivity

200

180160

co1.) 140

120LE 100

cnu) 80m 60

40

Source:

207

1 5 10 15 20 25 30 35 40 45 50

Top 50 Programmers by Rank Order

"Not all Programmers Are Created Equal," G. Edward Bryan, IEEE, 1994.

ment for computer systems ana-lysts, engineers, and scientists was inexcess of 10 percent, well abovethe 1.2 percent" for computerprogrammers.27 Managers andmore advanced IT workers require

a longer time to train, both throughformal education and on-the-jobexperience. Consequently, there willbe longer lags in responding tochanges in demand for these morehighly skilled workers.

27 Burt S. Barnow, John Trutko, and Robert Lerman, "Skill Mismatches and WorkerShortages: The Problem and Appropriate Responses," Draft Final Report, The UrbanInstitute, February 25, 1998.

39

Figure 2-6

Who's Getting the Top Talent?Tier 1 Hot Software Companies

Software start-ups and boutique service firmsSoftware publishersWall StreetR&D (corporate and university)

Tier 2 - Software-aware CompaniesVARs, consulting firms, systems, integratorsSoftware intensive industries (computerhardware, communications, financial services)Aerospace systems firms

Tier 3 Everyone ElseOther industries with incidental softwareMost IS applications development andma intenanceDoD, federal, state, and local governments

Source: Stanford Computer Industry Project, 1999.

43

Chapter 3. Demand, Constraints, and Consequences

What Are theDynamics ofthe Market-place and theDangers ofGovernmentIntervention inthe IT LaborMarket?

In most industries, there areboom and bust cycles that affect therespective labor market. Even theIT sector, which overall has had arapid upward growth for the pasthalf-century, has experiencedeconomic downturns. It is com-mon in any active profession tohave occasional mismatches be-tween labor supply and demandbecause business cycles tend tomove up and down more rapidlythan changes in supply. Occasionally,supply changes more rapidly thandemand, as shown by the growthof 40 percent per year in newlydeclared computer science majors atresearch universities the past twoyears. The invisible hand of themarketplace will often correct for

shortages through wage pricing andother adaptations. Especially whenthere are many employers vying forthe labor pool, wages will typicallyrise enough to attract workers fromother fields. Government interven-tion in a market is generally re-garded as advisable either when thefree market is unable to operate onits own, or where the costs areregarded as too high. Examples ofthese costs are the long time it takesfor the market to self-correct, thepain caused to individuals orinstitutions, or risk placed on thenational economy or nationalsecurity. In fact, it is very difficultfor government organizations toeffectively control labor supplynot to mention that there is littlepolitical will for doing so in this eraof free markets. It is hard topredict future demand and tocollect timely data about the effectintervention is having; as a result, itis easy to over-stimulate a laborpool.

Even when wages rise, themarket cannot adjust more quicklythan the amount of time it takes totrain an adequate supply of work-ers. In the IT sector, such delays areoften quite short. Anecdotal evi-

4 1 45

46

dence suggests that it takes about sixmonths to retrain a worker for alow- or mid-level IT occupation,assuming the worker alreadypossesses some basic skills on whichto build. Some high-level positions,such as management positions orsenior research positions in labora-tories, may require a longer trainingperiod. Sometimes, instead of ashortage occurring immediately, thelevel of talent filling open positionsis gradually lowered. This may beespecially true for particular seg-ments of the labor marketforexample, a geographic region,employers who have a rapidincrease in demand, or companiesthat are regarded by workers assomehow less attractive as employ-ers (e.g., because of the nature ofthe work, the wages paid, or thecorporate culture).

The recent history of interven-tion by U.S. government agencies tomeet perceived shortages ofscientific and technical workers doesnot provide an encouraging picture.Perhaps the most notorious recentcase of failed policy pronounce-ments is the warning during the late1980s from then-senior manage-ment of the National ScienceFoundation (NSF) about looming'shortfalls' of scientists and engi-neers. These warnings were basedon methodologically weak projec-tion models of supply and demandthat were originally misinterpretedas credible forecasts, rather thansimulations dependent upon certainkey assumptions. The projectionsyielded numerical estimates of theshortfalls anticipated, eventuallyreported to be 675,000 scientistsand engineers by the year 2006.

Based in part on these worry-ing pronouncements, Congressagreed to increase funding for NSF

4 2 ,

science and engineering educationprograms. Several years later, in1990, again influenced by theshortfall claims, Congress agreed togreatly expand the number of visasavailable for foreign scientists andengineers, for both permanent andnon-permanent residents. (This billwas the origin of the H-1B visas,among other measures.) Manyeducational institutions moved toincrease the numbers of graduatestudents in these fields. By the timethese larger cohorts of graduatestudents emerged with their newlyearned doctorates, the labor marketin many fields had deterioratedbadly, and many found their careerambitions extremely frustrated.This experience proved embarrass-ing, leading to congressionalhearings in 1992 and harsh criticismof NSF management from severalprominent congressional supportersof science and engineering. Arepetition of this experience shouldbe avoided in handling the IT labormarket.

What FactorsLimit the Abil-ity of the Gov-ernment, Indus-try, UniversitySystem, andProfessionalCommunity ToImprove theMatch BetweenSupply andDemand?

It may be difficult for compa-nies to recognize that there is aworker shortage, especially if theyemploy only a few workers in a

given IT occupation, or if they areused to taking a long time to fillvacancies (which is common forpositions that require a high level ofeducation or training or many yearsof experience, even in times ofmarket equilibrium). But even if thegovernment, industrial, and aca-demic sectors do recognize thatthere is an IT worker shortage anddecide they want to deal with it,there are factors that limit theirability to do so. The high level ofcompetition and the short productlife and product development timeoften make it difficult for compa-nies to hire new employees whorequire a lengthy period of break-intraining before they can becomeproductive. It also makes it difficultto retrain an existing employee for asignificantly different job. Thuscompanies are sometimes forced,by competitive pressures, to lay offworkers of one type and hireworkers of another type. Or theymay refuse to hire anyone who doesnot already possess all the neededskills. These employer practicesreceive harsh criticism at times fromlabor unions and some governmentofficials, but to some degree this is arational and perhaps necessaryreaction to the realities of themarketplace.

Universities are sometimescriticized for their slow response tomarket conditions and their reluc-tance to allocate or reallocateresources to programs with highand growing demand. Perhapsmost importantly, it should benoted that the colleges and universi-ties do not control student demand.In this free market, student demandcan change quicklycertainly morequickly than most universities canreact. It should also be remem-bered that most universities havelimited resources, most of which

are tied up in long-term commit-ments such as buildings and tenuredfaculty. It is difficult for universitiesto shift tenured faculty from onesubject area to anotherespeciallyfrom one department to another,but often even within a givendepartment. Even when such shiftsdo occur (such as library sciencefaculty becoming managementinformation science faculty) thesefaculty retreads are often notaccepted as full citizens, they maynever conduct the kind of researchthat fits into the new department,and they may not provide goodleadership in the new area. How-ever, they do take up faculty slotsthat might have been assigned to ayoung researcher trained in the area.Part-time and adjunct appointmentsare one tool that departments canand do use to respond to rapidlychanging market conditions. But acommitment to build up a com-puter science or informationsystems faculty (other than withnon-tenure-track faculty) meansmaking a long-term commitment,which is almost certainly made atthe expense of some other worthyinitiative.

The slow response is alsopartly due to the decision andreview process. This process oftenseeks out views on major initiativesfrom many parts of the univer-sityfaculty, administration, andsometimes even students and staff.This deliberative process, whichlargely precludes a response timethat can keep up with industrytrends, is part of what universitiesbelieve gives them strength. Indus-try, however, often sees this operat-ing style as a weakness.

There are various factors thatlimit the government's ability to act.Supply and demand are not regu-

4 3 47

48

lated by the government; thegovernment can only offer incen-tives to encourage a student to studya particular field or a company tobroaden its hiring practices. It ishard for a government to stimulatelabor supply in any discipline by justthe right amount: the market isconstantly changing, the informationabout supply and demand is imper-fect and is difficult to obtain in atimely fashion, and it is hard topredict the effects that a governmentinitiative might have. Another factorinvolves state and local versus federalrights. Educational issues, especiallyat the K-12 level, are consideredprimarily a local prerogative; butnational labor issues require nationalaction, or at least national coordina-tion, of K-12 and higher education.Government organizations also havea limited ability and desire to inter-fere with the actions of a privateorganization, such as a company or auniversity.

For all of these reasons, evenwhen there is a desire to act, thereare often impediments to doing so.

What Are theCosts of an ITWorkerShortage?

A substantial shortage of ITworkers can incur many differentcosts. The effects are felt on manylevelsthe nation, various indus-tries, individual firms, and consum-ers:

A shortage of skilled ITworkers in the IT industry and other

high-tech industries would slowinnovation and product develop-ment, which in turn would harmexports and wealth creation in theUnited States.

Industries dependent on ITworkers would grow more slowlyif there were an IT worker short-age, and this would have a negativeimpact on employment overall.

American companies mightbecome less competitive globallybecause the up-push of salaries in atight labor market would have to bereflected in the prices of goods andservices produced, or because thesecompanies could not get theirproducts on the market as rapidly asforeign companies that are fullystaffed.

If American talent were tooexpensive or in short supply,companies would be more likely tomove jobs offshore.

Companies with intensive orcompany-critical IT functions thatwere unable to find an adequatesupply of qualified IT workersmight have to merge with compa-nies that have these talents. While amerger is not a bad thing in and ofitself, industry concentration for thepurpose of acquiring expertisemight harm economic forces andindustry efficiency.

Information technology isleading to enhanced productivity ina number of industries. A lack ofIT workers would lead to a slow-down in productivity improve-ment." When salaries are driven up,certain kinds of organizations might

28 In the early 1990s some economists questioned the productivity added by informationtechnology, but this attitude seems to be changing, coming into line with long-standinganecdotal evidence that this technology can provide significant productivity gains.

4 4

be less able to meet industry marketsalaries and would be affecteddisproportionately by the shortageof workers. This has been true, forexample, of government organiza-tions trying to fix their Y2K prob-lems. This factor has also affectedthe ability of universities and othernonprofits to attract qualified ITtalent.

IT graduate students andfaculty might be attracted awayfrom universities in sufficientnumbers to jobs in industry thatthere would not be an adequatenumber of teacher-scholars left totrain the next generation of ITworkers (the so-called "seed-corn"problem).

A shortage of skilled peoplewould mean that in some casesemployers would fill positions withpeople who are less than adequatelyqualified. This practice could leadto poorer performance, high ratesof project failure, a slowdown indelivery, and decreased innovation.

Software projects, especiallylarge ones, already have a highfailure rate (perhaps an artifact of along history of inadequate numbersof well-trained IT workers). Thiswould likely worsen with a shortageof skilled IT personnel.

Tight labor markets areoften characterized by "churning,"the increased movement of theemployees from one employer toanother. Churning significantlyincreases recruitment and retrainingcosts for employers; it also meansthat the time of technical staff, aswell as management and humanresources staff, is spent on recruit-

29 For a reflective article on the issue of jobIEEE Spectrum, November 1998, p. 17.

ing, retaining, and retraining em-ployees, rather than on moreproductive elements of the busi-ness." New employees are not asproductive as employees who havebeen with a company for sometime because new employees needto learn about company projects,processes, skills, and capabilities.

Workers who are respon-sible for conceptualization anddevelopment lose productivitywhen there is a shortage of ad-equate support staff and adequatelymaintained support systems.

When there is a shortage ofstaff, employees who might havebeen assigned to create new prod-ucts sometimes are reassigned tomaintain existing products orcompany support systems. Thiscould limit innovation and stiflecompetitiveness.

There could be a reductionin the range of products available toboth individual Americans and toAmerican companies who use themto improve their own products,services, and well-being.

What Are theInternationalConsiderationsin Dealing witha NationalWorkerShortage?

The IT marketplace and thecompanies in it are increasinglyinternational in scope. Virtually allof the major companies in the IT

churning, see Robert W. Lucky, "Job Churn,"

49

50

industry headquartered in the UnitedStates offer their products andservices around the world. Mosthave sales offices in many countries,and some have factories and evenresearch laboratories outside theUnited States. Beginning in the1970s, American companies beganto increase their use of overseasworkers to manufacture compo-nents and assemble products, aslabor costs escalated in the UnitedStates. Similarly, IT companiesheadquartered in other countrieshave opened foreign offices as theytry to expand their markets. A fewforeign IT companies have estab-lished a substantial presence in thiscountry, and a number of bothEuropean and Japanese companieshave established IT-related researchlaboratories here (e.g., Hitachi,Mitsubishi, NEC, Panasonic, Philips,Ricoh, Sharp, Siemens, and Sony).

This means that there isincreasing worldwide competitionfor IT contracts; if U.S. companiescannot provide the service andproducts, increasingly there areother options. This surely couldlessen the demand for IT workersin the United States; but it couldalso prove to be an opportunity forU.S. companies to build up theirworld business. It was longthought that American industrywould maintain control of both thenational and worldwide IT indus-tries because of its large civilian andmilitary domestic markets, stronghigher educational system, goodresearch laboratories, and domina-tion of software development. Butthere have been pressures againstthe American companies. Forexample, in the 1980s, a globaliza-tion of the software industry began,

driven by a search for talent. Sofar, the United States still dominatesthe software industry, perhapsbecause the capacity of foreignlabor sources is strictly limited bythe numbers of highly educatedindividuals and by the educationalinfrastructures in other countries.However, it is hard to know howlong this domination will last.

The flow of IT work andworkers is not limited by nationalboundaries. It is not uncommonfor IT workers from one countryto work for a few years or eventheir entire careers in anothercountry, although no statistics areavailable on the numbers. Anec-dotal evidence indicates that theUnited States is by far the destina-tion of choice for IT study andwork. Roughly half the graduatestudents and one-tenth of theundergraduates in IT-relateddepartments are foreign nationals;and foreign-born students who earnscience and engineering (S&E)doctoral degrees from U. S. aca-demic institutions are staying in thiscountry after graduation in increas-ing numbers. A recent NSF studyindicates that 63 percent of foreign-born students who earned S&Edoctorates from U.S. institutionsbetween 1988 and 1996 said theyplanned to locate here, compared to50 percent or less of those previ-ously studied. Two-thirds of thosewho planned to stay had firm plansfor further study or employment."

One might think that theinflux of workers and studentswould help meet the strong de-mand by American companies forIT workers. Indeed, it does help.However, it is fairly clear that

3° See "Statistical Profiles of Foreign Doctoral Recipients in Science and Engineering:Plans to Stay in the United States" (http://www.nsf.gov/sbe/srs/nsf99304/start.htm)

4 6

foreign sources of labor is not thelong-term solution to IT laborneeds in the United States. Thesupply of foreign workers islimited by numbers, resources,government policies, and demandsfor them in their home and othercountries. A number of countriesare reporting their own shortages

of IT workers and are placingpressures on or providing incentivesto their indigenous IT workforce tostay at home or return home.These countries are also in competi-tion with the United States forworkers from those few countries,such as India, that have a surplus ofIT workers.3'

" Some examples of reports on IT worker shortages and worker levels in countriesother than the United States:

Canada: "Software and National Competitiveness, 1995 Update," Software HumanResource Council. (hard copy). Also see http://www.shrc.ca/index.html; Update to theHuman Resource Study of the Software Industry, Human Resource Development Canada,1995 (hard copy), but also see http://www.hrdc-drhc.gc.ca/hrdc/corp/stratpol/arbsite/research/rsctoc_e.html and http://www.hrdc-drhc.gc.ca/hrdc/corp/stratpol/arbsite/publish/bulletin/contents.html)

The Netherlands: Advisory Council for Science and Technology Policy, Report 31. Thestructural need for computer scientists, February 1998. (PDF, in Dutch) Also see http://www.awt.nl/en/advlist.html

India: "The Software Industry in India 1997-98: A Strategic Review," NASSCOM, Alsosee http://www.nasscom.org/ and A. Barr and S. G. Tessler, "The Globalization ofSoftware R&D: The Search for Talent," http://www-scip.stanford.edu/scip/avsgt/cfr1296.pdf and "Software R&D Strategies of Developing Countries," http://www-scip.stanford.edu/scip/avsgt/cfr197.pdf

United Kingdom: "Skills Alliance Lists Projects That Need Help. Computerweekly (UK)article about study sponsored by Department of Trade and Industry. http://www.computerweekly.co.uk/news/20_11_97/08622625264/C6.html

Israel: "Information Technology in Israel: Human Capital & IT." Gives a count of ITworkers only: http://gurukul.ucc.american.edu/mogit/dz9586a/humapage.html.Forecast of a shortage (article refers to study done by Ministry of Industry and Trade):"Israeli Software Firms Meet Needs by Training, Hiring Yeshiva Pupils," Jerusalem PostService, http://www.jewish.com:80/bk970919/ifirms.htm

Belgium: Insea (a Belgian professional association) report (1997) counts the number ofBelgian workers and forecasts worker shortage and spiraling wages. "InformaticiensRois."

Worldwide headcount: Several years ago The Economist offered a chart of the number ofsoftware workers in various countries, based on some estimates of the META group.These numbers were not necessarily based on government reports by country, but are astarting point. "The Importance of Being American" (from Survey of the SoftwareIndustry) The Economist, May 25, 1996. Counting new graduates entering the workforce(outside of the US) is even tougher. But productivity and headcount are two differentthings. There are estimates of two million for the number of technical professionals inRussia, but their productivity is inhibited for a number of reasons. Some of thosereasons are mentioned in A. Barr and S.G. Tessler, "Software R&D strategies ofDeveloping Countries," http://www-scip.stanford.edu/scip/avsgt/cfr197.pdf

4 751

Chapter 4. Worker Shortage

How Does OneDetermineWhether ThereIs a LaborShortage?

Part of the controversysurrounding information technol-ogy (IT) workers stems from thefact that there is no definition of theterm 'labor shortage' that is univer-sally accepted. What one personmight consider a labor shortageanother might regard as only a"tightness" in the labor market, anda third person might dismiss thequestion entirely because of the wayin which the supply and demandcategories were defined. This

report uses the Department ofLabor's definition of shortage: a"market disequilibrium betweensupply and demand." A shortagerequires two things: an occupationalmarket in disequilibrium and a slowmarket response. Alternativedefinitions of labor shortage areprovided in sources cited in thefootnotes."

When a worker shortageoccurs, employees and workers takevarious actions, some of which canbe tracked statistically as indicatorsof the presence and severity of ashortageeven when one cannotmeasure supply and demanddirectly. These indicators include:vacancy rates, amount of employ-ment change (so-called 'churning')

32 For a description and analysis of various definitions of 'labor shortage' given byeconomists over the past forty years, see Burt S. Barnow, John Trutko, and RobertLerman, "Skill Mismatches and Worker Shortages: The Problem and AppropriateResponses," Draft Final Report, The Urban Institute, February 25, 1998, pp. 4-14. Onedefinition in particular that we do not adopt is the so-called Social Demand Model oflabor shortage. Under this definition, a shortage exists if there is not a sufficientnumber of workers of the type in question to meet a particular social goal. Thus somepeople might argue that there is a shortage of public school teachers because childrenseem to get a better education in smaller classes; or others might argue that there is asurplus of lawyers because our society would be better off if it were less litigious. TheSocial Demand Concept of a shortage depends on personal judgments about socialwelfare. The definition we adopt focuses on market equilibriumthe match betweensupply and demand as measured by economic indicators.

4 8 53

54

from firm to firm within the labormarket, unemployment rates,changes in wage rates, predictedemployment growth, and demandfor foreign workers as indicated bylabor certifications." In general, thepresence of multiple indicatorsprovides better assurance that ashortage exists. Perhaps the mostcommonly used indicator of a laborshortage is vacancy rates. Unfortu-nately, this indicator can be mislead-ing. Vacancies occur in labor marketsthat have supply-demand equilib-rium, and even in markets wherethere is a surplus of workers.Vacancies occur because of normalturnover and lags in filling openpositions; indeed, they are a naturalpart of any business operation. Onelarge company in the IT sectorreported to the study group that itregards vacancy rates of five percentas normala management decisiondriven by the flexibility needed to fillpositions quickly when the needarises in this rapidly changing field.

There are many kinds ofshortages. They may be specific toone geographic region, or to aparticular set of occupations orsubspecialties within a labor market.They may be episodic (e.g., createdby a one-time phenomenon oflimited duration, such as the Y2Kproblem or the Euro conversion)or more enduring.

There may be non-statisticalindicators of a labor shortage aswell. If a large number of peoplesay there is a labor shortage prob-lem and act as though there is, thenthere may well be a problem. Thisis especially true if the action comes

from companies ranging acrossmultiple geographic regions andindustry sectors, and if thesecompanies begin to devote signifi-cant resources to correcting theproblem by increasing recruiting,initiating studies, supplementingtraining, or even lobbying. Afurther indication of a shortage iswhen labor suppliers and profes-sional societies representing theworkers, not only the trade associa-tions representing the companies,become concerned.

Is There aShortage of ITWorkers?

For some readers, this is themost important question addressedin this study. It is difficult to give astraightforward answer because ofthe imprecise definitions of supplyand demand, the complexity of theIT workforce (many different kindsof jobs involving widely varyingskill and knowledge sets, anddistributed across many differentsectors of the economy), and thelack of good data. There might bea shortage in one IT occupation(such as LAN administrators), butnot in another (such as data entryclerks); or in one geographicalregion (such as Silicon Valley) andnot in another (such as a purelyagricultural area). The situation alsochanges rapidly over time. What istrue at the time this report is writtenmay not be true six months later.

In trying to answer thisquestion about a shortage, it would

33 See Barnow, Trutko, Lerman, ibid., p. 55, for a discussion of this issue. Also seeTimothy Bresnahan, "Information Technology, Workplace Organization and the Demandfor Skilled Labor: Firm-level Evidence," Department of Economics, Stanford University(http://timb.stanford.edu/research/itwold8-21f.pdf)

4 9

probably be more insightful to dosome labor market segmentation,rather than viewing the IT laborpool as one undifferentiated mass.There are barriers to geographicalrelocation of workers, which mightmake geographical segmentationuseful. These include the cost ofrelocation, the cost of living insome of the areas where IT jobsare abundant (e.g., Silicon Valley),the social upheaval to the workerand the worker's family, and thedifficulty of locating and securing ajob outside the region where theworker is currently living. Distancemay be a particular barrier in fillinglower-skill, lower-pay jobs. Thusgeographical segmentation of theIT workforce may be important tothe understanding of shortages.

Because the skills and knowl-edge required vary so much fromone IT job to the next, occupationalsegmentation may be even moreuseful than geographical segmenta-tion. It would be a good test ofthe analytical value of this report'scategorization scheme to segmentthe labor pool into conceptualizers,developers, modifiers/extenders,and supporters/tenders to segmentthe workforce when looking forlabor shortages (see table 2-2). It isclear, however, that such a segmen-tation would not provide a com-plete breakdown. For example,computer security and networkingare hot areas today, and there maywell be shortages, or at least verytight labor markets, associated withthese specialty areas. However,they cut across our four categories.The demand is uneven across evena single occupation, such as pro-grammer. There is no knowndemand for APL programmerstoday; COBOL programmers arein short-term demand for fixingY2K problems; and Java program-

mers appear to be in demand forthe foreseeable future because ofall the network applications beingwritten in Java. Unfortunately, thedata are inadequate to support asegmented analysis by eithergeography or occupational catego-ries; thus this discussion of theshortage issue will consider theentire IT workforce.

At the outset of this study, itwas clear that some trade associa-tions believed there was a shortageof IT workers, while some laborunions disputed the claim. TheH-1B debates showed that theworldviews and interests of thesetwo kinds of organizations made itdifficult to find common groundon which to evaluate the questionof a worker shortage. The studygroup was somewhat surprisedand dismayed to find a similar gulfbetween IT professionals andsocial scientists interested in thisissue. The IT professionals havewhat they believe to be over-whelming local evidence of ashortagepersonal experience ofmultiple unfilled jobs, multiple joboffers for their graduating students,high salaries for graduates in hottechnology areas, and more. Thesocial scientists, trained to discountthis kind of anecdotal evidence,demand strong statistical proof.As part of their scientific training,they tend to be skeptical aboutvirtually all of the existing statisticaldata, which is in one way oranother imperfect; and they tend todwell on the methodologicalshortcomings of the data ratherthan using them as predictorsalbeit imperfect onesof ashortage or lack thereof.

Where (most) IT profession-als see a shortage, (most) socialscientists see a tight labor market.

50 55

Is this just a semantic difference?This is unlikely. The distinctionreflects an attitude about the natureof the problem and the need for apolicy response. If there is simply atightness in the labor market, there isa belief that the costs are bearableto the market participants, themarket will eventually self-correct(or market participants will learn tolive with the situation), and nogovernment intervention is neces-sary. If there is a shortage, how-ever, the costs to the marketparticipants may be unbearably high,the ability of the market to self-correct is questioned, and govern-ment action may be warranted.

Unfortunately, no one kind ofevidence currently available providesa clear and unambiguous answer asto whether there is a shortageoreven a tightnessin the IT labormarket. Thus the strategy in thissection is to look at six differentkinds of evidence, each with itsown strengths and weaknesses:1) a direct counting of supply anddemand; 2) secondary statisticalindicators based on presumablysolid federal data; 3) data fromlimited-scope studies that seem tobe methodologically sound; 4) datafrom more general studies whosemethodologies have been ques-tioned; 5) anecdotal evidence ofemployer actions; and 6) otherqualitative evidence. The limitationsof each kind of evidence arediscussed, as well as what itindicates about the existence of ashortage/tightness.

1. Direct counting. Inorder to demonstrate that there is ashortage of IT workersthat is,an excess of demand over sup-plyone would need a clearworking definition of an informa-tion technology worker, as well as

56

good statistical data about supplyand demand. The problems offinding a good working definitionof an IT worker have already beendiscussed. The supply system iscomplicated, as chapters 5 and 6show; and anything close toadequate statistical data about thecurrent potential supply, much lessfuture supply, are nonexistent. Dataabout demand are even moreelusivethey are out of date, thesample sizes are too small, and theyrely on vague and inconsistentdefinitions of what kinds ofdemand to count as related to ITworkers. Thus it is impossible atthis time to provide meaningfulquantitative information abouta national shortage of ITworkers. In particular, there is noadequate basis for quantifying thesize of the shortage (if there isone). Collecting better data will bea major challenge, and the studygroup is not optimistic that reliablequantitative measures are likely tobe available anytime soon.

2. Statistical indicatorsbased on federal data. Even ifthe supply-demand match cannotbe measured directly, it is useful tolook at four secondary dataindicators based on federal data.These data are likely to be asreliable as any data available on thisissue because of the size and scopeof federal data sets, the objectivitywith which the government collectsthe data, and the methodologicalrigor with which they are analyzed.These secondary data indicatorsprovide only inferential, not direct,evidence of a shortage; and foreach indicator, there are somereasonable questions that might beraised about the validity of theinference. In general, these second-ary indicators are also unable todistinguish between a shortage and

Table 4-1Unemployment Rate for IT Workers. 1988 - 1996. Based on Current Population Survey (CPS) Data

Year

Operations &Computer SystemsSystems Analysts & Researchers &Scientists Analysts

ComputerProgrammers

All ProfessionalSpecialtyOccupations

All Workers 16years & older

1988 1.4% 2.2% 2.9% 1.7% 4.9%1989 1.4 2.8 1.6 1.7 4.71990 1.5 1.3 3.0 2.0 5.01991 2.6 3.2 3.5 2.4 6.21992 2.7 2.0 3.1 2.6 6.81993 3.1 1.9 2.7 2.6 6.21994 1.8 2.7 2.1 2.5 5.71995 1.9 1.5 1.8 2.5 5.21996 1.3 1.2 1.6 2.3 5.01997 1.1 1.4 1.6 2.1 4.5

Source: U.S. Department of Labor, Bureau of Labor Statistics, Employment andEarnings, Household Data, Annual Averages (published in January after each year).

a mere tightness in the labormarket. While none of the indica-tors provide conclusive evidence,taken together they offer a prepon-derance of circumstantial evidencein support of tightness/shortage inthe IT workforce.

Unemployment rates. As table4-1 shows, the unemployment ratesfor each of the three categories ofIT workers under the Bureau ofLabor Statistics (BLS) classificationsystem were all 1.6 percent or lowerin 1997. The IT unemploymentrates are about three times as lowas overall unemployment rates inthe United States-suggesting ashortage/tightness. However, toproperly interpret these numbersthey should be seen in comparisonwith some other statistics. First, theIT unemployment rates have beenconsistently low, in both absoluteterms and in relationship tonational unemployment rates, since1988; however, the claims for anIT worker shortage have only beenmade in the past several years.Why were they not made in the late1980s or early or mid-1990s?Second, it may be unfair to com-pare IT unemployment rates with

national unemployment rates, inthat professional unemploymentrates are almost always significantlylower. The overall unemploymentrate for all specialty professions isonly slightly above two percent-not that much different from theIT worker unemployment rates.But it is hardly credible that there isa shortage of all professionalworkers. Thus, while unemploy-ment rates may suggest a shortage/tightness in the IT labor market, asan indicator they are not entirelyunproblematic.

Permanent labor certificates:Certificates offered by the Depart-ment of Labor represent anotherindicator of demand unmet bydomestic workers. Aliens who wantto become permanent residentsmust have an offer of permanent,full-time work from an employer inthe United States. The employermust obtain a labor certificate fromthe Department of Labor, indicat-ing that qualified U.S. workers arenot available for the position, andthat the wages and working condi-tions are consistent with prevailingconditions for similar employmentin the United States. In 1996, about

57

58

40,000 new applications werereceived and a slightly largernumber were approved. Abouteleven percent of the approvedapplications were for IT-relatedoccupations. Three IT occupationswere among the five occupationsreceiving the largest number ofpermanent labor certificates in 1996:software engineers (ranked second,behind specialty foreign cooks, with2,238 approvals (5.5% of the total)),programmer analysts (ranked thirdwith 1,231 approvals (3.0 %), andsystems analysts (ranked fifth (aftercollege and university faculty, with616 approvals (1.5%)). Other IToccupations ranked in the top 100were: computer programmers (124approvals), database design analysts(104 approvals), computer systemshardware analysts (74 approvals),database administrators (69 approv-als), systems programmers (56approvals), and programmersengineering and scientific (45approvals).

There are several argumentsagainst using these permanent laborcertificates as a strong indicator ofIT labor shortage/tightness. Whilethe IT occupations rank high in thelist of occupations for whichpermanent labor certificates arerequested, the absolute numbersare small. Some even questionhow meaningful these statistics are.Anecdotal evidence from com-puter industry executives suggestthat many of these certificates areapplied for (and more than 90percent are approved) not to fill anunfilled job, but at the request of acurrent employee who is workingon a temporary visa (H-1B or oneof the predecessor programs) whowants to work permanently in theUnited States. The Department of

Labor's Inspector General hasreported that the certificationprogram is subject to extensivemanipulation by employers andimmigration lawyers."

Temporal), labor certificates.Department of Labor certificatesare also available for employerswho wish to hire professionalworkers under the H-1B visaprogram. Employers must certifythat they will pay the prevailingwage and that there is no strike orlockout underway. There is nolimit on the number of jobopenings that can be filed forcertification, and the Departmentof Labor approves almost all ofthem. However, there is a limit onthe number of H-1B visas that canactually be awarded in a given year.In 1996, 41.6 percent (102,422 outof 246,725) of the applicationswere for IT-related jobs. In fact,the most frequent occupations forwhich H-1B labor certificates werereceived (84,370 jobs, representingabout one-third of all jobs certi-fied), were in Systems Applicationsand Programming. The largenumber of these H-1B laborcertifications suggests a shortage/tightness in the domestic labormarket. Although many of thesecertificates are apparently awardedto foreign-owned companiesoperating in the United States thathire primarily foreign workers (e.g.,Tata Consulting, Syntel, orMastech), these companies are stillundertaking IT work for U.S.industry, helping to satisfy theoverall demand. Hence, eventhough special kinds of companiesmay dominate the requests forthese certificates does not meanthat this information should not befactored in.

34 See http://www.oig.dol.gov/public/reports/oa/1996/foreign-labor-cert.pdf.

53

Table 4-2

Median Weekly

I

% Change in Salary % Change in SalaryType of Job Earnings ($) (1997) 1988 - 1997 1996 - 1997

Computer Systems Analystsand Scientists

918 36.2 3.0

Operations Researchers/ 867 28.4 6.4Systems Analysts

Computer Programmers 840 42.9 8.8

All Professional Occupations 750 35.1 2.7

All Workers with 4 or moreyears of college

779 33.2 2.8

All Workers 16 years old orolder

503 30.6 2.7

Source: Adapted from Burt S. Barnow, John Trutko, and Robert Lerman, "SkillMismatches and Worker Shortages: The Problem and Appropriate Responses," DraftFinal Report, The Urban Institute, February 25, 1998, Exhibits 13 and 14. Based onthe Bureau of Labor Statistics data.

Wage growth. Table 4-2shows that salaries in the IT-relatedoccupations have gone up at aboutthe same rate as, or only slightlyfaster than, salaries in other profes-sional occupationsor than salariesoverall. This fact suggests there isno shortage, contrary to all of theother statistical indicators based onfederal data described above. Onepossible explanation is that BLSdata on wages are inconsistent withprivate sources of data, such asemployer salary surveys. Theprivate data sources virtually all giveannual wage growth figures that arehigher than federal dataincludingone source that shows an annual

wage growth of 18 percent in someIT occupations." The trade associa-tion Information TechnologyAssociation of America (ITAA)claims the reason for this is that"BLS numbers do not include stockoptions, signing bonuses, andreferral bonuses, which are majorparts of compensation packages inthe IT industry."" However, thecase of the systems administratorssuggests another possible explana-tion for slow wage growth. When-ever there is a rapid growth in anoccupation, the average level ofexperience is likely to fall. In someIT occupationssuch as systemsadministratorsthere are a large

35 However, some of the recent non-federal data show a flattening of salary increases.See the 12th annual salary survey by Computerworld and the related article by LeslieGoff, "Enough is Enough: The Joyride Is Over, As Corporate Managers Put the Brakes onOut-of-Control Salaries for IT Professionals," Computerworld, September 7, 1998(http://www.computerworld.com/home/features.nsf/a11/980907mgt)

36 Lauren Brownstein, "Is There a Shortage of Information Technology Workers?"Symposium Proceedings, The Jerome Levy Economics Institute of Bard College, June 12,1998, p. 5.

59

Table 4-3

IT Occupations with Anticipated High Job Growth 1996-2006

Employment (thousands)Type of Job 1996 2006 % Change

Database Administrators, ComputerSupport Specialists, and all other 212 461 118Computer Scientists

Computer Engineers 216 451 109

Systems Analysts 506 1,025 103

Desktop Publishing Specialists 30 53 74

Data Processing Equipment Repairers 80 121 52

Engineering, Science, and ComputerSystems Managers

343 498 45

Source: Bureau of Labor Statistics, Monthly Labor Review, November 1997.

number of workers at the entrylevel, apparently meeting a need atthat level; however, there may be aserious shortage of workers withthe same job title but havingadvanced skills." While the averagesalary for systems administrators didnot rise much in the period 1989 to1993, salaries for high-end systemsadministrators (at least in the SanFrancisco Bay area) doubled." Insuch cases, it might be morerevealing to track top-quartilesalaries or conduct a longitudinalstudy of salaries of a particulargroup of workers, rather than relyon average occupational salary. It

is not clear how representative thisexample of systems administratorsmay be of IT work as a whole.

Worker projections. Table 4-3shows IT occupations that ap-peared in the table of Occupationswith the Largest Job Growth,1996-2006, produced by BLS. Thejoint category of "database admin-istrators, computer support special-ists, and all other computer scien-tists" appeared at the top of the list,with an expected growth rate of118 percent. Computer engineersand systems analysts came secondand third on the list. Also appear-

37 This example is not one discussed in the ITAA report. For an interesting overview ofthe occupation of systems administrator, including a description of the job and formaleducational, commercial training, and independent study programs preparing a workerfor this occupation, see David Kuncicky and Bruce Alan Wynn, "Educating and TrainingSystem Administrators: A Survey," published by the USENIX Association for SAGE, theSystem Administrators Guild, Berkeley, CA, 1998.

38 According to Stephen Johnson, the USENIX representative on the study group.USENIX pays close attention to the work of systems administrators.

60

ing on the list were desktop pub-lishing specialists (ranked 8th), dataprocessing equipment repairers(ranked 15th), and engineering,science, and computer systemmanagers (ranked 23). Thesenumbers are not collected data asare the other federal data we haveconsidered in this section, but areonly forecasts. Thus they concernthe future, not the current situation.Why should they be considered asany kind of an indicator of thecurrent situation? The reason is thatthe projections are based in part onthe current situation, as well as onanticipated future changes that willaffect supply and demand. It isunlikely that the projections forfuture years, especially in the nearfuture, would have shown such alarge predicted growth if therewere not a current robust demand.It should also be remembered,however, that these kinds ofprojections are notoriously difficultto make and that BLS projectionshave sometimes been way off themark." For all of these reasons,these worker projections areprobably the least reliable of all thefederal data indicators of a short-age or tightness.

3. Limited-scope studies.Two studies are considered here.

Both focus on software workers,which is the area of IT work withthe greatest occupational growth.One study is national, the otherregional. The national study wasproduced by the National SoftwareAlliance, a consortium of industry,government, and academic leadersthat was formed specifically toaddress their concerns that there isan IT worker shortage." The studygives many different kinds ofevidence, some based on federaldata, many based on private data.We did not evaluate the methodolo-gies used in the collection andanalysis of these private datasources, but presumably there is arange of quality and reliabilityacross these sources. What isperhaps most remarkable about theNational Software Alliance's study,however, is the large number ofdifferent statistical analyses, fromvarious sources, that support oneanother in showing the existence ofa shortage, or at least a tight labormarket, for software workers.Here are a few examples from thereport based on private data:

The Olsten Corp., a largestaffing services company, conductsan annual staffing survey of a rangeof North American businesses. The1997 survey indicated that the high-

39 For example, in 1984 BLS projected 520,000 computer science and systemsanalysts jobs in 1995, but there were actually 860,000. BLS predicted a 53-percentgrowth in electrical engineering jobs over this period, whereas there was actually a 9-percent decline. (See John H. Bishop's commentary in the Levy Symposium Proceed-ings, p. 8.)

4° National Software Alliance, "Software Workers for the New Millennium: GlobalCompetitiveness Hangs in the Balance," Arlington, VA, 1998. This Alliance has aDepartment of Defense orientation. It has been argued that salaries and benefitpackages offered to IT workers in the defense industry are relatively poor comparedwith those offered to IT workers in other sectors; and that this disparity creates adifficult recruiting and retention situation that is reflected in the National SoftwareAlliance's view that an IT worker shortage exists, and the need to act on it for nationalsecurity reasons (Michael Teitelbaum, Sloan Foundation, personal communication,March 1999). Whether one needs to question the reliability of the National SoftwareAlliance's data, even if they are an interested party, is unclear.

61Li

tech sector had a much higherpercentage (590/0) of companiesreporting that they did not haveenough employees to meet theiroperational needs than any otherfor-profit sector.'"

Microsoft Corp. had morethan twice as many job openings inits Microsoft Certified SolutionProvider program in the UnitedStates than there were softwareworkers in all of Irelandthecountry that exports the world'ssecond largest amount of softwareafter the United States (and, hence,presumably one of the leadingpotential sources of IT workersfor the United States)."

The National SoftwareAlliance analyzed twenty-one privatesalary surveys, all of which showedthat IT worker salaries were risingfaster than inflation. A study by theDeloitte and Touche ConsultingGroup indicated an increase of7.45 percent from 1996 to 1997 forcomputer networking professionals.Computerwodd's 1997 salary surveyrevealed average annual salaryincreases of more than 10 percentin 42 percent of the twenty-sixoccupations tracked."

Starting salaries for com-puter engineers with a newbachelor's degree increased by 5.8percent to $39,722 from 1996 to

1997." (This increase is well abovethe national inflation rate, but lowerthan increases in a few otherprofessional occupations.)

The insurance, automotive,computer software and hardware,telecommunications, and pharma-ceutical industries all have annualturnover rates of between 15percent and 19 percent for softwareworkers .45

The second study, conductedby the Washington SoftwareAlliance, looks at software workersin the State of Washington.46 As theAlliance itself notes, the situation inWashington is not entirely represen-tative of the nation as a whole. Theaverage software wage in the state isreported to be $66,752, the highestof any state in the nation. They alsonote that growth in the softwareindustry in the state is outpacing thenational average, with a growthfrom 1990 to 1996 of 17.8 percentcompared with a national growthof 9.8 percent. The state has twomajor employers, Microsoft andBoeing, that may make its situationsomewhat different from the nationas a whole. Given these caveats, it isuseful to consider the report'sstatistics, which suggest a seriousworker shortage.

The Alliance's survey includesall positions in the Washington State

41 "1997 Olsten Forum on Human Resource Issues and Trends: Staffing Strategies,"William Olsten Center for Workforce Strategies, Melville, NJ, 1997.

42 "Ireland: The Software Capital of Europe," National Software Directorate, Forbairt,1997; Christina Torode, "Closing the IT Skills Gap," Computer Reseller News, December1, 1997, Issue 766.

43 See National Software Alliance, op. cit., page 2-21 for citations to the original studies.

44 Based on data from the National Association of Colleges and Employers, "SalarySurvey," July 1997.

45 Based on 11th Annual Salary Survey, Computerworld, September 1, 1997.

46 See http://www.wsatorg/gotwork/wasoft.htm

62

software industry, technical andnon-technical. The industry includes2,500 companies, with $20 billion inannual revenue and 47,000 employ-ees. There are 7,300 currentvacancies (15.5% of all positions),with 64,000 total desired hires in thenext three to four years. A $12.8billion three-year revenue gain isprojected if these positions can befilled. Three-quarters of all jobsrequire a bachelor's degree orhigher. The greatest need (largestnumber, hardest to hire, drivingforce for the industry) is for juniorand senior developerspositionsthat require a BS or MS in com-puter science or computer engineer-ing. There are eight jobs for everyrelevant in-state bachelor's graduate,and four jobs for every relevant in-state associate's (two-year) degreegraduate. Unlike the views of anumber of other industry represen-tatives, the Washington Statesoftware employers are verysatisfied with the technical, social,and English-language skills ofthose they hire. Their problem,they claim, is a numbers gap, not aneducational quality gap.

In addition to consideringhow representative the WashingtonState situation is of the nationalscene, determining what the statisticsactually mean also needs to beaddressed. It is difficult to projectaccurately the number of newpositions that will exist or beneeded in the future, or the dollarvalue added by additional filledpositions. While Microsoft andBoeing may hire many of theirworkers from in-state schools, theyboth recruit nationally, making it

47 " He I p Wanted: The IT Workforce Gap at theTechnology Association of America, Arlington, VCollaborative Action for the New Millennium,"America and Virginia Polytechnic Institute and

somewhat misleading to calculatestate-graduate-to-open-job ratios.The problems of using vacancies asan indicator were described earlier.

4. Methodologicallychallenged national studies Thefirst ITAA report initiated thenational debate over the IT workershortage. This section will considerit and its follow-up report." Criti-cisms concerning the ITAA's meth-odology and results were discussedearlier, but are briefly reviewed here.

The main criticism was thedirect count of supply and demandthat ITAA tried to make. The firststudy used undergraduate gradua-tion rates in computer science tomeasure supply, when this is onlyone of many supply sources of ITworkers. The report also did notpresent available data that indicatedthe number of undergraduatemajors in computer science wasincreasing after a long decline. Onthe demand side, a very broaddefinition of an IT worker wasused, and the telephone survey usedto collect the data had a very lowresponse rate. Because of the lowresponse rates, the results arestatistically questionable. Forexample, companies that needed ITworkers may have been morewilling to respond to the surveythan companies that did not. Forthese reasons, there are real ques-tions about the reliability of thepredictions in the two ITAAreports, respectively, that there are190,000 and 346,000 unfilled ITpositions in the United States. Buteven if the results do not have theweight of scientific evidence, they

Dawn of a New Century," InformationA, 1997; "Help Wanted 1998: A Call forInformation Technology Association ofState University, March 1998.

J 63

64

suggest that a number of compa-nies are having difficulty fillingpositions.

Other kinds of evidence arealso cited in the first ITAA study tosupport a worker shortage. Hereare some samples:

82 percent of the companiesthat responded to the ITAA surveyindicated that they planned toincrease the number of IT workersthey employed in the coming year,while only 2 percent indicated theyplanned to decrease the number.

83 percent of the companiesindicated that they thought thedemand for IT workers was higherthan the demand for other kinds ofworkers.

William M. Mercer, whichconducts a compensation study forITAA, noted an increase in averagehourly compensation of nearly 20percent from 1995 to 1996 foroperating system software architectsand consultants, about five times thenational average.

The second ITAA report, inaddition to its information aboutvacancies, provides a combinationof anecdotal information andstatistical data from other sources.All of this additional evidence theycited suggested a shortage/tightness.(However, it should be remem-bered that ITAA is an advocacygroup and that they presumablychose only examples that supported

their position.) Here are some ofthe examples ITAA cites:

The high IT vacancy rates oftwo companies are noted: 14percent at Booz-Allen & HamiltonInc., and 16 percent at Mary Kay Inc.

A quarter of the IT manag-ers reported that employees whodeparted voluntarily in the previousyear was in excess of 10 percent oftheir organization's total program-ming staff, according to anInformationWeek survey."

Some companies werepaying employees up to $15,000 forreferrals that led to critical IT hires;company stock options werereported as being routinely offered;benefits and amenities such as gymsand exercise facilities, day care,restaurants, medical assistance,flexible work schedules, andincreased vacation time wereincreasingly common." (But thereport did not reveal how oftenthese perks and benefits wereoffered.)

5. Anecdotal evidenceabout employer actions. Table 4-4 provides a list taken from theBarnow, Trutko, and Lerman paperon what a company might do inreaction to a labor shortage. Thereis anecdotal evidence that every oneof these strategies has been adoptedin connection with IT workers.

Recruiting. There are manysigns of increased recruiting. More

48 E. Cone, -staffing: Short Supply," InformationWeek, November 1997 (http://www.techweb.cmp.com/iw)

48 "Good Help Is Hard to Find," Computerworld, October 1997, and "Good Help Is Hard toFind," Computerworld, November 1997both at http://www.computerworld.com; "TechCorporate Culture Cozy, Creative," USA Today, November 1997 (http://www.usatoday.com); E. Cone, "Staffing: Short Supply," InformationWeek, November 1997(http://www.techweb.cmp.com/iw).

5 9

Table 4-4

Company Reactions to a Worker Shortage

Increase recruiting effortsIncrease use of overtimeReduce minimum qualifications forthe jobRestructure work to use current ornew employees in otheroccupations

Substitute machinery andequipment for laborTrain workers for the jobsImprove working conditionsOffer bonuses to new employeesImprove wages and fringe benefitsContract out the workTurn down work

Source: Burt S. Barnow, John Trutko, and Robert Lerman, "Skill Mismatches andWorker Shortages: The Problem and Appropriate Responses," submitted by the TheUrban Institute to the Office of the Assistant Secretary for Policy, U.S. Department ofLabor, draft final report, February 25, 1998, pp. 22-31.

ads are being placed in both thetraditional print sources and on theInternet; and display ads, whichattract more attention and are moreexpensive, are on the rise. Recruit-ing on campus appears to be wayup. Georgia Tech, for example, hasseen the number of employersattending their IT career daysdouble each year for three years,until meeting space capacity wasreached. Job fairs are being heldwith increasing frequency, if ads inthe local and national newspapersare any indication. Booklets listingavailable IT jobs have started to bedistributed free of charge in videostores and other retail establishmentsthroughout Northern Virginia, nextto the Apartments to Rent booklets.It is apparently becoming morecommon for companies to paysignificant recruitment bonuses tocurrent employees who can recruitnew workers. Companies are alsoincreasingly targeting specificschools in their recruitment efforts,and are sending their corporateexecutives and highest profileresearchers to these campuses toraise the company's visibility withthe students.

Overtime. There have beensome reports of extra overtime forIT workers. IT faculty and ITmanagers in industry are typicallyexempt employees who work longhours without extra compensation.IT workers in computer startupfirms may also work long hourswithout extra compensation becausethere is the expectation of totaldedication in return for a cut of theaction. But overtime is generallypaid to IT workers by mostcompanies outside the IT sector,and by many of the larger andmore established computer andsoftware manufacturers. Becauseof the prevalence of long hours,employers have implemented anumber of standard practices, suchas feeding employees who worklate or taking the development teamon expensive vacations after aproject has been successfullycompleted. However, hours canonly be increased so much and forso long before the strategy be-comes counterproductive, leadingto poor morale and shoddyworkmanship, for example.

6 0 65

Reducing minimum qualifica-tions. There is a long history ofcompanies hiring people for ITjobs who possess fewer or differ-ent skills from those outlined in thejob description. Perhaps this isbecause the unemployment rate forIT workers has been so low forsuch a long time. There is someevidence that today companies arerecruiting in more places, andhiring students graduating fromschools that are less prestigiousthan they would have considered inthe past. As discussed elsewhere,hiring underskilled employees canbe a dangerous practice becauseproductivity and quality levels varyhighly from worker to worker inmany IT occupations, especiallyones involving programming.

Restructuring work. IT workis being restructured so that humansubstitution can occur. Clericalstaff are being trained to work aslocal area network or databaseadministrators, Web designers, andsoftware installation experts.Companies are trading off agreater amount of general andfoundational education (such aswould be gained from a bachelor'sdegree and some course work incomputing) for training in aspecific, current technology (highschool graduates with a six-monthcertificate program to become aNovell or Microsoft technician). Insome cases, companies contract outthe technical aspects of the workand retain only a small IT staff ofone or two leaders and somemaintainers.

Substitution of machiner y forlabor. There have been many effortsto substitute machinery for labor,but the long-term effect is un-known. Companies relying onsoftware applications are rapidly

66

introducing automated softwaremaintenance programs that replaceworkers, but, in many program-ming environments, technology isexpected to reduce labor by nomore than ten percent.

Training. Companies seem tobe more willing to retrain currentemployees than to train new work-ers. In a highly competitive market-place with narrow windows ofopportunity to introduce a newproduct, such as an Internet-basedproduct, it is not surprising thatcompanies would be reluctant to hireworkers who needed six monthsbefore they could be fully produc-tive. However, some companies,such as Andersen Consulting andPeopleSoft, have a successful practiceof hiring bright college graduates,independent of their undergraduatemajor, and giving them intensivetechnical training when they join thecompany. It is unlikely, however, thatthis practice would work for allcompanies.

Working conditions. Companiesin the IT industry have a long-standing record of providingwhatever is needed in the way ofbenefits, improved offices, childcare, flexible working hours, andother non-wage perquisites toattract and retain workers. It issomewhat harder for companies insome other industrial sectors to givepreferential treatment of this kindto their IT workers and not to theirother workers.

Bonuses. Sign-on bonuses arebecoming increasingly common forIT workers.

Wages and Benefits. This issuewas discussed earlier. Anecdotalevidence suggests that IT workersare sensitive to market salaries and

benefits and are willing to switchjobs on this account. Organizationsthat have restrictions on their abilityto meet market conditions, such assome colleges and governmentorganizations, often have troubleattracting or retaining IT workers.

Contracting. The amount ofIT work that is contracted out isalready substantial and is growing.The rapid growth of IT consultingfirms is evidence of this trend.

Refusing work It is commonfor IT departments to tell theirinternal customers that they cannotmeet their requests. It is difficult todetermine how much external workis refiised.

6. Qualitative evidence.There is also an abundance of non-statistical evidence. This does nothave the same persuasive force asmethodologically sound quantita-tive data because it is hard todistinguish the special case from thegeneral rule, but qualitative evi-dence can be both revealing andsuggestiveespecially if there is alot of it, of various kinds, and fromvarious sources. The preponder-ance of this anecdotal informationsupports either a shortage or atightness in the IT labor market,although none of the evidenceavailable can distinguish between the

two. The third bullet belowpresents one of the few examplesthat speak against there being eithera shortage or a tightness." Here area few examples:

o Every company our studygroup heard about, both in the ITsector and in other sectors, claimedit was experiencing difficulty withIT recruitment (either worse-than-expected results in hiring, or havingto work much harder than in thepast to achieve hiring quotas).

o The study group cameacross various professional societ-ies, state government agencies,regional economic trade associa-tions, private foundations, indi-vidual colleges and universities,university systems, and chambersof commerce that were concernedabout and studying the IT laborshortage issue, or that had devel-oped programs to develop ITworkers." Most of these organiza-tions would not have devoted thetime or effort unless they wereconvinced there was a problem.

o A small body of non-statistical evidence speaks against aworker shortage (or even tightnessin the market). This includesanecdotal evidence from oldertechnical workers unable to obtainIT jobs, for example in the electrical

50 For some examples of articles questioning the existence of a shortage, see MargieWylie, "The Skills Shortage That Isn't," CNET news.com, February 4, 1998 (http://www.news.com); also Dominique S. Black, "Taking Shots at the Labor Shortage," ITCareers, March 23, 1998.

51 For example, the Society for Information Management (SIM), a professional associa-tion of 2,700 managers of information technology and its applications to business, hasdrafted a position paper on the worker shortage. "SIM believes this labor shortage isthe most severe in the 50-year history of computing, and will continue well into the nextmillennium...[in part because of] a fundamental shift in investment economics favoringincreased use of IT." ("Addressing the Information Technology Workforce Shortage,"Position Statement, Society for Information Management, October 1998, Chicago, IL.) A

Web site maintained by the Department of Commerce profiles 170 IT worker-develop-ment programs throughout the country. See http://www.ta.doc.gov/go4it/

667

engineering field. One wouldassume that experienced electricalengineers would have some com-puter skills, mathematics, scienceand engineering background,problem-solving and communica-tions skills, and a knowledge ofindustry that would make themdesirable to companies needing ITworkers; but electrical engineersindicate some trouble being hiredfor these IT jobs.

Conclusions:There is no way to directly

answer the question of whether thereis a shortage of IT workers becausethere are no adequate definitions oradequate data to directly counteither supply or demand.

Other sources of informa-tion are inferential and less reliablethan the direct counting approach.Indeed, there are credible reasonsfor doubting virtually any piece ofevidence that is currently available.

The inferential evidence doesnot easily allow one to distinguishbetween a shortage and a tightnessin the IT labor market. (Theindicators would look about thesame whether there was a shortageor simply tightness.)

The statistical indicatorsbased on federal data, the regionaland occupation-specific data studies,the methodologically challengedadvocacy studies, and the qualitativeevidence almost all suggest either atightness or a shortage.

It is likely that there are spotshortages, both in specific geo-graphic regions and in specificoccupations. In a field experiencingrapid growth and rapid technologi-cal change, it would be surprising ifthere were not such shortages.

68

Discussion of supply anddemand of IT workers would bemore insightful and useful if themarket could be segmented bygeography and/or occupation, butdata do not exist to carry out thisanalysis.

Where Are ITShortagesOccurring?

There is only anecdotalevidence to answer this question.The federal data categories are toocoarse and the data too old to beof much help. The software areaappears to be experiencing signifi-cant shortages at this time, especiallyfor applications relating to network-ing, databases, and Internet-basedapplications. These shortages tendto be concentrated in niches and injobs requiring higher skill sets. Theproblem is worsened by the factthat software workers are notalways qualified to move from onesoftware job to another. A pro-grammer experienced in oneprogramming language may not beeffective in another. It is not simplya matter of learning the grammarand vocabulary of the language.More critical is learning the underly-ing methodology used with thatprogramming language to attackprogramming projects, and thismethodology varies considerablyacross programming languages. Forexample, programming in COBOLis very different from programmingin Java; and it has been hard toretrain COBOL programmers(who received their experience onmainframe computers and arebriefly in demand today because ofthe Y2K problem) to be Javaprogrammers (who are likely to bein great demand well into the new

6 3

millennium to program applicationsfor use on the Internet).

Based on anecdotal evidenceonly, the following points representthe study group's consensus aboutoccupations where either demandoutstrips supply or many positionsare filled with underskilled workers:

programmers, but especiallythose who are familiar with Oracle,SAP, BAAN, and ERP;

programmers and designerswith object-oriented and Javaexperience;

Web and e-commercespecialists;

Network designers;

Problem-solvers withenough IT background to be ableto work as consultants for themajor accounting and consultingfirms (Ernst & Young, AndersenConsulting, PriceWaterhouseCoopers,EDS);

Faculty, especially in highschools and some two- and four-year colleges;52 and

Managers and project leaders.

52 A small survey on faculty hiring carried out by the ACM Special Interest Group onComputer Science Education, to which 64 institutions responded, suggests that facultyrecruitment is more difficult in non-research institutions. When asked how difficult oreasy faculty recruitment is, the survey received the following responses: at schoolsthat grant the Ph.D. in computer science: easy 12.5%, moderate 31.3%, difficult 56.2%;at schools that grant the master's degree as their highest degree in computer science:easy 4.7%, moderate 28.6%, difficult 66.7%; for schools that grant the bachelor'sdegree in computer science as the highest degree; easy 3.7%, moderate 3.7%, difficult92.6%. Preliminary results, July 27, 1998. For more information, contact ProfessorPaul Myers, Department of Computer Science, Trinity University, San Antonio, TX.

69

Chapter 5. SupplyThe Degree Programs

What Are theSources ofIT Workers?

The traditional, formaleducational system remains criticallyimportant to the training of theinformation technology (IT)workforce. Different kinds of jobswithin the IT field require verydifferent skill sets and levels ofknowledge, and thus different ITjobs vary greatly in the kind andlevel of education they require.Formal programs leading toassociate's, bachelor's, master's, anddoctoral degrees in IT-related fieldsall have their place in the supplysystem. An associate degree willtrain a person for certain kinds ofentry-level positions that mayinvolve maintaining or tendinginformation technologies, whereas adoctorate might prepare a personto create new information technol-ogy. None of these relationsbetween training and particular IToccupations are hard and fast,however.

The associate's and master'sdegree programs are importantsupply sources for IT workersbecause they tend to be morevocationally oriented than bachelor'sor doctoral programs. However,the bachelor's degree programsproduce the largest number ofgraduates for the IT workforce, andthe doctoral programs are critical inthe production of trained workersfor both the occupations involvingconceptualization and advanceddevelopment, and for facultypositions that will educate the nextgeneration of IT workers. The mostpopular IT-related majors arecomputer science, followed bycomputer engineering and manage-ment information systems. How-ever, one author has recentlyidentified twenty IT-related degreedisciplines offered in the UnitedStates (see table 2-1 in chapter 2),and new ones are being created allthe time." What these programsteach is more important than whatthey are called. Indeed, programnames are often confusing, and itcan be difficult to establish similari-

53 See Peter J. Denning, "Computing the Profession," Educom Review, November 1998;also Denning, "Information Technology: Developing the Profession," unpublisheddiscussion document, George Mason University, December 4, 1998.

6 5 71

ties or differences between programssolely on the basis of their names.

One of the least known andmost important facts is that the vastmajority of IT workers do notobtain formal degrees in IT-relateddisciplines. Perhaps the mostcommon of the many differenttraining paths to an IT career is abachelor's degree in some technicalfield unrelated to informationtechnology, accompanied by somecourse work in either an IT subjector in closely related preparatoryfields such as mathematics, electricalengineering, or business.

At the same time, certain IToccupations do demand a particularkind of formal education. Ad-vanced researchers, such as facultymembers in research universities orprincipal scientists in industrialresearch laboratories, almost alwayshave a doctorate in an IT-relateddiscipline, usually computer scienceor computer engineering (oroccasionally in a closely related fieldsuch as physics, mathematics, orelectrical engineering).

Over the past decade, therehave been vast changes in thecharacteristics of IT work andpreparation for it. Traditionally,higher education served as the basisfor one's career, although some ofthe larger IT companies hadtraining programs for their employ-ees. Today, higher education is anentry ramp into a job, but it is notexpected to carry one through acareer. Taking advantage of on-the-job experience and variouskinds of continuing education, theIT employee is today expected toengage in a life-long retrainingeffort, which is intended to keepthe worker up to date in this rapidlychanging field.

72

Many different groups supplythis continuing education. Thehigher educational system has amajor role, offering seminars, shortcourses, and groups of courses thatlead to certificates in specializedaspects of information technologysuch as network administration orbiocomputing. Universities oftenattract the mid-career employeewho goes back to earn an addi-tional degreeperhaps a computerscience degree for someone whomajored in humanities or a master'sdegree in business administrationfor the computer engineer. Everylevel of the higher education systemparticipates, but the training is mostlikely to come at the associate ormaster's level.

Others provide continuingeducation and retraining as well.For-profit educational companies,such as DeVry or the University ofPhoenix, offer formal degreeprograms and certificate programs.Private consultants and privatetraining companies offer specializedseminars, as well as customizedtraining programs for individualcompanies. Companies themselvesare getting in to the training businessfor their own employees (andsometimes for others), forming so-called "corporate universities" thatthey may develop on their own orin partnership with one of the othertraditional or for-profit suppliers.These corporate universities teachnot only IT technical materials, butalso develop communications andinterpersonal skills, and impartknowledge of business and industrypractices.

The need for continuousretraining has made it necessary torethink the way in which educationis delivered. It has to be madeavailable at a time, place, and in a

6 6

Figure 5-1

IT Worker Traditional

Formal

Career Path

Career

RetirementHS Job

1

Job

2

Job

32 yr.

4 yr.

Master

Doc

Source: Computing Research Association, Intersociety Study Group on InformationTechnology Workers, April 1999.

style that accommodates theemployees' work and personal lives.This requires more frequent offer-ings on evenings and weekends,scattered around many differentgeographical locations. Even better,in some cases, is the use of teachingmethods that free the student froma specific time and place. Com-puter, Internet, and broadcasttechnologies are being used todevelop various kinds of distancelearning, some of which can beengaged by the student asynchro-nouslythat is, when the studentwants the training, not when ateacher is scheduled to give alecture. These technologies areavailable in a rudimentary formtoday, but they continue to bedeveloped and are being put intopractice at a rapid pace.

..

How HaveCareer Pathsfor IT WorkersChanged OverTime?

Information technologyworkers are pursuing a training andcareer path today that is differentfrom that practiced by mostmembers of this profession asrecently as a decade ago. Thetraditional career path can berepresented by a linear model, asshown in figure 5-1. The prospec-tive IT worker prepares for theworkforce through formal (non-profit) education, gets a job, andmoves up through the worker andmanagement ranksworking forone or a very small number ofemployers until retirement. Thereare various exit points in the formaleducational system, represented bythe awarding of a degreethe high

73

Figure 5-2

Current IT Worker Career Path/Formal Degree Programs

HS, Associate, Bachelor's,Master's, Doctorate

Non-degree ProgramsDistance Education,Employer Training,

Self-study

Source: Computing Research Association, Intersociety Study Group on InformationTechnology Workers, April 1999.

school diploma, associate's degree,bachelor's degree, master's degree,or doctorate. The place at whichone exits the formal educationalsystem largely determines thestarting point in the workforce, andit also restricts the range of careeropportunities. It is uncommon, forexample, for a person to rise tosenior management without holdingat least a bachelor's degree, notwith-standing the many journalisticaccounts of high school and collegedropouts becoming millionaires inthe computer industry.

The current career model, asrepresented by figure 5-2, is morecomplicated and less linear. Thereare multiple opportunities foreducation and training: various levelsof school-based formal degreeprograms (as before), non-degreeprograms produced by these samesuppliers, distance educationprograms, employer-based and for-

74

profit training organizations, andself-study. Unlike the traditionalcareer path, where one's educationwas completed before entering theworkforce, the current model hasworkers moving back and forthbetween the educational system andthe workforce, or participating inthem concurrently, as they continu-ously, or at least periodically, retrainthroughout their careers.

In this new model, significantlinearity remains in the college anduniversity degree programs. Forexample, a bachelor's degree isexpected before entering into adoctoral program. But whenworkers seek additional training,even from the traditional collegeand university system, there is nomost common path. IT workerswith a bachelor's degree, forexample, might enroll in: 1) acommunity college for a certifica-tion program in networking

technology; 2) a four-year college totake mathematics or science coursesto bolster their basic knowledge;3) a few courses or a degreeprogram at the master's level ininformation systems or business; or4) a doctoral program in computerscience. The purpose of thisformal training might be to helpthem to do their current job better,or perhaps to prepare them for abetter job. However, there aremany paths to a particular IT career.Box 5-1 presents six examples ofnon-traditional career paths of ITworkers. All of the people in theseexamples are personally known tomembers of the study group.

Company training is, ofcourse, not a new phenomenon.For many years, companies havebeen providing some training totheir employees. A few largecompanies, such as IBM, AT&T,and Motorola, have long placed anemphasis on this kind of training.However, as the discussion belowon corporate universities indicates,there has been a very pronouncedincrease over the past decade inboth the amount of training and theexpectation that all employees(rather than a few select groups ofemployees) will engage in continu-ous or periodic retraining.

The attitude of employeestoward jobs and employers haschanged as well. Given that lifetimeemployment with a single employeris increasingly rare, workers nolonger regard the particular job theyhold as a rung on a ladder they areclimbing within the company. Theyare less likely to accept just any jobthe company management wants toassign. Jobs are now regarded asanother element of the trainingprocess, of learning by doing, andemployees move from job to job

to gain new skill sets and experi-ences rather than assume they willstay with a particular company forlife. Acquiring new skills allowsthem to move within the entire ITwork community for opportunities,rather than solely within a particularcompany.

The following sectionsdescribe formal degree programs,followed in chapter 6 by otherforms of training. Although thedescription of the non-formalmethods of training is briefer andthe statistics more meager, this formof training appears increasingly tobe prevalent and a principal meansof obtaining many IT skills.

Mae llzs Ovaffolle a lapzeilvoollz filveThe Zauppolly

Zyzeopm?

It is during the high schoolyears that people typically receivetheir first knowledge and skill sets ininformation technology. Some ofthis learning occurs in formalprograms in the schools, but formany students it is likely that evenmore learning occurs throughpersonal exploration and interactionwith peers. As personal computersbecome more common in homesand schools, students are gainingexposure at progressively earlierages; and this will no doubt lead toprogressively earlier introduction ofcomputers into the curriculum.While it is undoubtedly valuable tofind ways to enhance informallearning, this discussion focuses onformal programs in the highschoolsboth vocational programsthat prepare the student directly to

. 6 9 75

76

Box 5-1

Non-traditional Paths to and from IT Careers

AAA is a 56-year-old Ph.D. in Astronomy. He worked for nearly twenty years in anumber of large and small companies, the last ten years in the Aerospace industry.When that industry downsized in the early '90s, he was laid off. Living off savings,his wife's salary, and his severance package, he returned to school full time andearned a master's degree in Computer Science. He quickly found work with asemiconductor company.

BBB is a 25-year-old college graduate who majored in Psychology, and worked in adaycare center after graduation. As a result of both home and college environments,she had basic computer literacy, and, discouraged by the low salaries in childcare,took a temporary job in IT doing quality assurance. She enjoyed this work, andnow works full time doing quality assurance for a company involved in Internetcommerce.

CCC is a 38-year-old with a master's degree in Russian economic history. Whileworking on his masters, his roommate got a PC, and CCC became intrigued with it.He started reading books on computing. He landed a job with a company thatmade systems software, and ended up moving to a job where he is one of themajor software resources of a small company.

DDD was a poetry major in college, working on her master's. She took a summerjob at a bank as a technical writer when the bank was just introducing ATMmachines. She became interested in the technology, dropped out of school, andended up working for five years with the bank, by the end of which she was doingsoftware project management. She joined a start-up company that eventuallybecame a major producer of computers, and ran their software delivery system formany years.

EEE was a COBOL programmer for a major automaker. His department hadmore than 200 workers in his division. After a successful pilot project the companydecided to move to Java and object-oriented programming. They used a rapid re-skilling program designed by one of the largest IT consulting companies. Aboutforty co-workers passed the certification tests and are working on other projects.EEE is working at the local McDonald's. The company is busy recruiting newcollege hires and experienced object programmers to fill the void.

FFF was married at 18 to a Navy man, who left her with two small children andonly a high-school education. She got work at a bookbindery. After several yearslearning the printing business she made the jump to electronic publishing, joining astart-up company in this field. Being willing to "tackle anything," she rose inmanagement until she was assistant to the President and Chairman.

Source: Computing Research Association, Intersociety Study Group on InformationTechnology Workers, April 1999.

7 0

enter the workforce, and academicprograms that prepare the studentfor college study.

High school vocationalprograms train students for manydifferent occupations, but only afew of them, such as data entry andelectronics technician, are IToccupations." The horizons arelimited for people who enter the ITworkforce by this route, unless theyare willing to seek additional educa-tion over time. This is true even ofthe bright high school studentsreported in the national press whohave been hired before completinghigh school to be Web designers orother kinds of programmers."Unfortunately, many students whoenter the vocational high schooltracks will not have the foundationalskills for further education and careeradvancement. A few model pro-grams scattered around the countryhave tried to address this issue. Theseprograms blend academic rigor withvocational objectives. Students areattracted to the programs becausethey are guaranteed jobs upongraduation. The programs generallyhave a local focus, with the highschools partnering with industry andtwo- and four-year colleges anduniversities in their region. While notall students from these programsactually attend college, the educa-tional requirements are consistentwith the university-bound curricu-lum. This enables the participant tobetter handle a first job and to moreeasily pursue further education later.

Anecdotal evidence suggeststhere are several challenges for the

high schools in preparing evencollege-bound students for ITcareers. Because many parentsknow less about the computingprofessions than the more estab-lished professions, it is incumbenton high school guidance counselorsto provide information aboutcomputing careers. Counselorsthemselves need to know muchmore about the career opportunitiesand the training needed to enterthese professions. Because of therapid growth of informationtechnology, high schools are havingdifficulty attracting qualified teachersand helping them to keep theiracademic preparation current.There are also particular problemsof attracting female and minoritystudents to IT careers (see chapter 7.)

Two kinds of programsappear to be succeeding in prepar-ing high school students for ITcareers. One is directed primarily atstudents who are already on theacademic track. It offers college-level courses in computing (some-times conferring college credit),taught by high school teachers orfaculty from local colleges. Thesecond type of program tries tomake technical (but not necessarilyIT) careers attractive to able stu-dents who might not otherwiseenter college and get professionaljobs. It is designed for at-riskstudents who are more likely to beminorities or come from poorcommunities. An example is theEngineering Vanguard Program ofthe National Action Council forMinorities in Engineering, Inc.(NACME). This program conducts

54 One example of an IT program for high schools and two-year colleges is the CiscoNetwork Academy. See http://www.cisco.com/edu/academies/index.html

55 See, for example, Ethan Bronner, "Computer Industry Luring Students Into DroppingOut," The New York Times, June 25, 1998.

7 77

its own assessment of high schoolseniors (rather than relying onschool grades and evaluations),provides intensive academic enrich-ment, and sends students to partici-pating colleges on full scholarships.The NACME program has shownexcellent retention rates.

llz eggsffogsnag coililagePirOgYEIEW flog

Ma arm*Sy.seam?

Both of the major degreetracks (transfer and non-transfer) intwo-year colleges are a source of ITworkers. Many students enroll intransfer programs, where theobjective is to prepare students totransfer to a four-year school tocomplete a bachelor's degree uponcompletion of their two-yearassociate's degree. The educationand training they receive, which isroughly an equal mix of generaleducation and discipline-specificcourses, is similar in depth andbreadth to what students wouldreceive in the first two years atmany four-year colleges. Thisalternate route to a bachelor'sdegree is important for studentswho have financial or geographicrestrictions, such as needing to liveat home, or who do not haveconfidence in their academicabilities at the time of high schoolgraduation. Thus the transferprograms are the beginning of oneeducational path for higher-level IToccupations.

Also popular are non-transferprograms, which are designed toprepare graduates for immediate

7 8

Yews 3-1.mr-osw.

Colleges Awarding Computernformation:Science Degrees

Year Number of Institutions

1989-1990 632

1990-1991 625

1991-1992 696

1992-1993 697

1993-1994 709

1994-1995 727

Source: National Center for EducationStatistics, Digest of Education Statistics,using preliminary data for 1990-91 andfinal data for all other years.

employment. These programs tendto have a high concentration ofdiscipline-specific coursesgenerally 75 percentcombinedwith general education courses.Non-transfer programs preparestudents for various kinds of ITwork, such as network installation,Web development, or computersupport services. The knowledgeand skill sets acquired tend to focusmore on a specific subarea ofcomputing, such as Web develop-ment, and less on the general theoryand concepts that are emphasized ina bachelor's degree program.Compared with those holdingbachelor's degrees, therefore, thesestudents tend to be better preparedto immediately begin performing ajob in the specific subarea in whichthey have been trained; but they areless well prepared to use their skilland knowledge base to transfer toother kinds of IT work.

Table 5-1 shows the numberof two-year colleges awardingdegrees in computer or informationsystems. The number increased byabout fifteen percent during thefirst half of the 1990s. Assuming

1 4.

Table 5-2

Associate Degree Production inInformation Technology

YearNumber of Degrees

Awarded

1992-1993 9,196

1993-1994 9,301

1994-1995 9,152

Source: National Center for EducationStatistics, Digest of Education Statistics

the increase continued at approxi-mately the same pace during thesecond half of the decade, thenumber of two-year collegesawarding IT degrees numbersbetween 800 and 850 today. In1994-95 there were 2,184 two-yearcolleges in the United States.% Thusin that year only about one-third ofthe nation's two-year collegesoffered programs in informationtechnology. This suggests there aresignificant growth opportunities inthis part of the supply system.

There are no reliable statisticson enrollments in IT courses at thetwo-year college level." Somewhatbetter data are available for associ-ate degrees awarded in informationtechnology. Table 5-2 shows thatthere were between 9,000 and10,000 degrees awarded in each ofthree years during the mid-1990s.These numbers include all degreesawarded in the areas of computer

and information sciences, computerprogramming, data-processingtechnology, information science andsystems, and computer systemsanalysis. These statistics underre-port, at least slightly, the number ofIT workers being trained in thetwo-year colleges, in that studentsalso prepare for IT careers inseveral other majors that graduate amixture of IT and non-IT workers.Two examples are electronics andgraphic arts programs. Electronicsprograms are diminishing nation-wide, due to lack of students.Graphics arts programs, however,are on the rise because of theinterest in computer graphics andWeb design. It is hard to estimatethe exact number of associatedegrees awarded in IT areas in themid-1990s, much less today, butthe number appears to be on theorder of 10,000 per year.

Two occupations, electronicstechnicians and Web designers,illustrate some of the dynamics thatare occurring in the industry thathave a bearing on the kind oftraining needed. Many of theroutine tasks that electronicstechnicians graduating from associ-ate degree or high school vocationalprograms used to do are now doneby machine rather than by a techni-cian. As a result, the skill set in thetechnician's job is changing, requir-ing the technician to think moreabstractly, have a greater basic

56 U.S. Department of Education, National Center for Education Statistics, Digest ofEducation Statistics.

57 The only relevant data found were from the National Center for Education Statistics,as reported in the Digest of Education Statistics. These statistics showed a halving ofenrollments from the 1992-93 academic year to the 1995-96 academic year, from583,000 students enrolled to 275,000 enrolled. The data were extrapolated fromsamples that may have been too small to be meaningful, they combine associatedegree enrollments with certificate program enrollments, and they differ greatly fromthe experiences of the two-year college personnel consulted for this report. For thesereasons, there is little confidence in the reliability of these statistics.

79

80

knowledge of the technology, andwork with more complex systems.

The situation with the job ofWeb designer is similar but morecomplicated. Several years ago,Web sites were rudimentary;nevertheless, significant program-ming ability was required to writethe HTML code that is used to laythem out. Over the past severalyears, technology has been de-signed to automate many of theprogramming aspects of thelayout. Thus if one is doing asimple Web design today, it ispossible to use automated soft-ware that does not require pro-gramming skillsin fact, the task issimilar to using word-processingsoftware. However, Web pageshave become much more elabo-rate and powerful, and high skilllevels are still required to lay outpages that meet these new stan-dards. Some of these skills involveprogramming, but others involveknowledge of design or human-computer interfaces. As these twoexamples illustrate, the trainingsystem has to be sensitive to theserapid changes in the requiredknowledge and skill sets of ITjobs in order to train usefulworkers.

A number of qualitativeconcerns about impediments totwo-year college production of ITworkers were expressed during thecourse of this study:

Inadequate career counsel-ing at the middle school and highschool levels leaves students intwo-year colleges ill prepared tomake decisions about their degreeprograms.

Weakness in the academicpreparation of entering studentshas created a need for remedialeducation in basic mathematics,reading, and English. This processdelays the production of gradu-ates.

The availability of jobsprior to graduation results instudents not completing theirdegree requirements, which mayhave long-term negative impact ontheir ability to advance or changecareers.

Four-year colleges some-times impose artificial barriers tothe acceptance of transfer creditfrom two-year colleges." Thesecan result in the unnecessaryrepetition of course work or candissuade students from matriculat-ing in a bachelor's program.

There are similar difficultiesin transferring credits between thetransfer and non-transfer tracks,even within the same two-yearcollege. This can box in studentswho begin in a non-transferprogram and decide subsequentlythat they want to pursue abachelor's degree.

Inadequate availability oftrained faculty, good facilities, andother resources reduces thenumber of students who can betrained.

Cultural attitudes learned inhigh school or earlier reduce poolof students electing to major inprograms leading directly to ITemployment. These factorsparticularly affect women andminorities.

58 We understand there are also legitimate reasons, involving both quality and content, fornot allowing transfer of credits from one school to another, or from one program to another.

7 4

What Is theRole of Four-Year CollegePrograms inthe SupplySystem?

Four-year colleges are theprimary supply sources for ITworkers. The first two years of thebaccalaureate program typicallyoffer general education and intro-ductions to various disciplines. Thefinal two years include substantialcourse work in a single majordiscipline. It is these final two yearsthat is the principal focus of thissection. In most cases, coursestaken in the final two years enhanceexisting IT knowledge and skills,rather than providing first access tothem."

Three classes of undergradu-ate students tend to enter careers ininformation technology:

Those majoring in a degreeprogram focused specifically onsome aspect of informationtechnology (e.g., computer science,computer engineering, or informa-tion systems);

Those majoring in a degreeprogram in a closely related field(electrical engineering, mathematics);and

Those who major in adiscipline largely unrelated toinformation technology (psychol-ogy, management), but who take acluster of IT-related courses asdistribution requirements, electives,or a minor."

59As this report was going to press, a monograph came to our attention on "change inthe nature and extent of college students' study of computer science over the period,1972-1993, the occupational destinations of students with computer science back-grounds, and the forces that shape the path from higher education to the labor market."(Executive Summary). There was not time to incorporate the findings of the monographinto this report, but the citation is provided for those interested in further examinationof these issues: Clifford Adelman, -Leading, Concurrent, or Lagging? The KnowledgeContent of Computer Science in Higher Education and the Labor Market," Office ofEducational Research and Improvement, U.S. Department of Education, May 1997.

60 One might want to divide up this third category, and add a fourth category of studentswho major in disciplines far removed from information technology (e.g., English), but whoget on-the-job training working as user support specialists in the university. BrianHawkins, president of EDUCAUSE, described this class of IT workers to us: "On collegeand university campuses this process of internal training and development is theprimary vehicle for finding user support specialists, IT support personnel in academicdepartments, etc. In universities, the internal consulting program, and the skillsdeveloped in course, as well as informal acquisition of skills are primary developmentvehicles, transforming English majors and other non-IT fields of study into quite capablesupport personnel. I would estimate that well over 2/3 of the young staffers in theuser support arena who are hired in universities are humanists and social scientists.These young people developed appropriate skills in their 4 years, developed a liking forthis work, and find the job opportunities better and higher paying. These individualsalso have a better working knowledge of supporting other academic units, since theyoften bring their "educational training" to bear on their IT support functions. Many ofthese people have gone on to professional IT positions outside of the university, basedupon their experience and OJT [on-the-job training]. None of these processes wouldshow up in professional surveys as a way of preparing IT professionals." Personalcommunication, March 8, 1999.

3 81

Box 2-1 (see chapter 2) showsthe continuum of degree programsthat focus on information technolo-gies, as reported in a classificationscheme developed by a NationalResearch Council (NRC) committeein 1993. The objectives range fromtraining students to develop hard-ware or software, maintain infor-mation systems in organizations, orprovide information services.There is considerable overlap inthese descriptions, and sharpboundaries cannot be easily drawn.Newly emerging computing areas,such as computer support servicesand Web designers, are not wellcovered by this classification; andcolleges are not uniform in the waythey name their programs. Thedescription given for InformationScience may better apply to pro-grams in Library and InformationScience. But the NRC classificationremains a useful description of thekinds of IT programs offered byfour-year colleges.

The professional societies andprofessional accreditation organiza-tions provide guidelines on the skillsand knowledge expected ofgraduates in several IT-relatedbachelor's degree programs. Modelcurricula for computer science andcomputer engineering are publishedjointly by the Association for

Computing Machinery and theIEEE Computer Society, whilemodel curricula for informationsystems are published jointly by theAssociation for Computing Ma-chinery, the Association for Infor-mation Systems, and the Associa-tion of Information TechnologyProfessionals.° Accreditation isprovided for computer science bythe Computing Sciences Accredita-tion Board and for computerengineering by the AccreditationBoard for Engineering Technology(ABET).62 Software engineeringaccreditation criteria were approvedin October 1998 by ABET. Licens-ing of software engineers is cur-rently being introduced into theUnited States by the State of Texas,where the state licensing board isworking with the professionalsocieties to develop standards andrelevant examinations.

Although some data exist onthe production of IT workers byfour-year colleges, it has not beencollected in a way that conforms tothe NRC classification scheme. TheNational Science Foundation (NSF)is the primary source of data.Under its classification scheme,computing programs tend to beaggregated under computer science,computer and information sciences,or computer programming; while

61 See, for example, "IS '97 Model Curriculum and Guidelines for Undergraduate DegreePrograms in Information Systems," The DATA BASE for Advances in Information Systems,Vol. 28, No. 1, Winter 1997, ACM Special Interest Group on Management InformationSystems (SIGMIS). The struggle to define an educational philosophymuch less acurriculumin the field of information systems, which is beset by extremeinterdisciplinarity and rampant technological change, is the topic of an editorial byHenry H. Emurian, "Information Systems: An Interdisciplinary Perspective," InformationResources Management Journal, Fall 1998, pp. 3-4.

62 Virtually all of the departments in IT-related disciplines see the value of modelcurricula and pay attention to them, even if they do not adopt them as a whole. However,there is not universal adoption of accreditation. Many of the top-ranked computerscience departments, for example, refuse to have their programs evaluated foraccreditation because they believe the value derived is far outweighed by the difficultyof the accreditation process.

82 76

Table 5-3

I I

Year Number of Institutions

1989-1990

1990-1991

1991-1992

1992-1993

1993-1994

1994-1995

1,059

1,042

1,036

1,039

1,042

1,068

Source: National Center for EducationStatistics, Digest of Education Statistics,using preliminary data for 1990-1991and final data for all other years.

computer engineering tends to belisted with electrical engineering in away that is hard to disaggregate.Programs in business areas, such asinformation systems and manage-ment information systems, tend tobe excluded." However, the maincomputing programs in businesscan be captured reasonably wellfrom the Digest of Education Statistics,which breaks out business informa-tion systems as a category ofbusiness degrees, and which has asubcategory for managementinformation systems and dataprocessing under the businessinformation category.

Given these caveats aboutdata, let us first consider thenumber of four-year collegesoffering degrees in computer andinformation sciences. Table 5-3shows that the number of schools

offering four-year degrees incomputer and information scienceswas relatively constant at just over1,000 throughout the first half ofthe 1990s. The total number offour-year degree-granting institu-tions in the United States was 1,855in 1994-95." However, some ofthese institutions specialize in asingle discipline, such as theology,art, music, or design; and othersmay already be teaching computingas part of some other activity, suchas mathematics. This same data setindicates that there were 1,145schools granting bachelor's degreesin mathematics and 1,248 schoolsgranting bachelor's degrees inEnglish. Business management andadministrative services was themajor offered by the mostschoolsand this number was only1,383. These facts suggest that thenumber of degree programs incomputer and information scienceat four-year colleges could not beincreased by more than 20 percent.Of course, the number of studentsgraduated by each of these pro-grams could possibly be increased,and it is probably more efficient toincrease the size of programs thanthe number of programs.

Table 5-4 shows degreeproduction in IT fields at four-yearcolleges. The largest number ofdegrees is given in computerscience, followed by managementinformation systems. It is hard todetect any trend from these figures.However, these data are old and donot reflect the dramatic recent

63 See Joanna Fortuna and Rob Kling, "Information Systems Data Missing in KeyDebates about the IT Worker Shortage: Data About the Education and Employment ofOrganizationally Insightful IT Workers," Working Paper, Center for Social Informatics,Indiana University, November 13, 1998.

64 U.S. Department of Education, National Center for Education Statistics, Digest ofEducation Statistics.

(7 83

84

Table 54

Number of Bachelors Degrees Awarded in Information Technology Fields 1992-1995

AcademicYear

ComputerScience

ComputerEngineering

ManagementInformation Science

and DataProcessing

Other BusinessInformation

Systems1992-1993 24,000 6,174 3961993-1994 25,200 2,237 5,434 4051994-1995 24,404 2,345 5,788 378

Source: National Center for Education Statistics, Digest of Education Statistics.

changes that seem to be occurring.The only data set that gives morerecent data is the Taulbee Surveyproduced by Computing ResearchAssociation. The Taulbee surveys infall 1996 and fall 1997 each col-lected data from more than eightypercent of the departments in theUnited States that grant doctoraldegrees in computer science andcomputer engineering. Thesedoctorate-granting departmentsrepresent approximately one-thirdof the national production ofbachelor's degrees in computerscience and computer engineering.The 1996 Taulbee survey showedan increase in students declaringmajors in computer science of 40percent over the previous year. The1997 Taulbee survey showed anincrease of 39 percent in declaredmajors over 1996. Thus, com-pounding the changes over the pasttwo years, the number of studentsdeclaring majors in computerscience and engineering effectivelydoubled (1.40 x 1.39 = 1.946).These figures may roughly hold truefor all schools granting four-yeardegrees, not only for those that alsogrant doctorates in computerscience and computer engineering.Of course, it takes several yearsfrom the time a student declares themajor until graduation, but presum-ably there will eventually be compa-rable percentage increases in the

number of students graduating withfour-year degrees in computing.

Table 5-5 is based on percent-ages of those with bachelor'sdegrees who said that their workwas not related to their degree. Itshows that undergraduate majors incomputer science are more likely topursue and continue in IT jobs thanengineers are to pursue and continueengineering careers, or scientists areto pursue and continue scientificcareers. It should not be inferredfrom this data, however, thatpushing a larger number of studentsthrough an undergraduate major incomputer science would result in ITworkers with similar levels of careerfaithfulness to those shown in thesestatistics. It is hard to determinewhether the additional recruits tocomputing will have the same skillsor enthusiasm for the field as those

Table 5-5

Career Faithfulness

Percentage Working In that FieldUndergraduate

After 1-5 Years After 20 YearsMajorComputer Science 70 70

Engineering 50 50

Physics 50 40

Mathematics upper 40's 35

Life Sciences upper 40's 35

Source: National Center for EducationStatistics, Digest of Education Statistics.

7 3

Figure 5-3

Degrees and Occupations in Computer Science 1992-1993

Mathematics10%

Other22%

Computer andInformation Sciences

33%

Engineering (includingComputer Engineering

11%

BusinessManagement

24%

Source: National Science Foundation, National Survey of College Graduates.

who might self-select computingwithout incentives.

Figure 5-3 shows that in 1992-93 only about one-third of thepeople in computer science orprogramming jobs had graduatedwith computer and informationscience degrees, according to theNational Survey of College Gradu-ates. The majority of the other two-thirds held degrees in businessmanagement, engineering, ormathematics. This is bad news foremployers of IT workers, given thatbusiness and mathematics enroll-ments have been dropping andengineering enrollments have been

flat." A number of other academicdisciplines are also common paths toan IT career, as box 5-2 shows.Moreover, many people notedduring this study that other attributes,such as the ability to work on a teamand communicate effectively, are asimportant as technical training.66

People also enter the ITworkforce after working first inother occupations, especially inclosely related ones such as engi-neering, science, and business. Onestudy found that almost half of thepersons employed in the ITworkforce in 1993 who hadgraduated from college at least four

NCES data showed that business enrollments dropped from 1,766,000 in 1992-93to 1,233,000 in 1995-96, and bachelor's degrees in business dropped from 257,000in 1992-93 to 234,000 in 1994-95. The number of juniors and seniors majoring inmathematics dropped from 67,000 in 1994-95 to 60,000 in 1996-97. Engineeringtotal enrollments and total bachelor's degrees in engineering have been unchanged atabout 325,000 and 65,000, respectively, between 1994 and 1997, according to theAmerican Association of Engineering Societies. The study group has heard anecdotalevidence, but has not been able to confirm with statistical data, that although thenumber of bachelor's degrees in business is dropping, the percentage of these degree-holders concentrating in management information systems is increasing.

66 See Michael C. Mulder and Doris K. Lidtke, co-chairs of a Collaborative Academe/IndustryTask Force, "An Information Systems-Centric Curriculum '99: Educating the Next Generationof Information Specialists, in Collaboration with Industry," draft report, January 1999.

r) 85

Box 5-2

Academic Disciplines Other than the Computing Disciplines ThatOffer Strong fraining for IT Careers

Mathematics - graduates ususally have an excellent background in logic andanalysis, and often a strong background in scientific programming andmodeling.

Statistics - graduates are usually very familiar with computer usage, relyingheavily on statistical packages, and are comfortable with data analysistechniques.

Engineering (other than computer engineering) - graduates generally havegood mathematics and science backgrounds that involve at least somecomputing. They are also trained in design and analysis.

Physics - graduates generally have a strong mathematics and sciencebackground and are usually familiar with computer hardware and someprogramming.

Chemistry - graduates have strong science and mathematics backgrounds,and they frequently have used computers.

Philosophy - graduates have strong logical thinking ability and may havetaken courses in mathematical logic that provide good training for computertheory.

Business - graduates have knowledge of the organizational characteristicsand issues involving the private sector, and modern business programsintegrate computing into their courses to give graduates competence incomputing.

Music - graduates have learned about the manipulation of patterns andthemes within constraints, which often serves as good background forprogramming.

Instructional design - graduates are familiar with many aspects of computersas users and developers.

Graphics arts and industrial design - graduates are familiar with userinterface and human-computer interaction issues.

Source: Computing Research Association, Intersociety Study Group on InformationTechnology Workers, April 1999.

86

years earlier were employed in orcompleting training for otheroccupations in 1989-20 percentwere in business, 10 percent were inengineering fields other than com-puter engineering, and 17 percentwere in fields other than science orbusiness." Yet another study, by theNational Software Alliance, indicatedthat 55 percent of electrical engineer-ing graduates had moved intosoftware-related jobs.

There are, unfortunately, nogood data available about the thirdcategory of IT workers graduatingfrom four-year colleges: studentswho major in unrelated areas, butwho may have taken substantialamounts of coursework in com-puter and information science.These data would be useful,inasmuch as the numbers involvedappear to be substantial.

A number of qualitative issueswere raised in the course of thisstudy about the four-year colleges assuppliers of IT workers:

The growth in student majorsand enrollments is stretching thin theabilities of colleges to provideadequate computing facilities."

In any academic field, theintroductory course in the field has agreat effect on the recruitment ofstudents to enroll in additionalcourses and become majors. Butfirst courses in computer sciencetend to have high attrition rates.Partly this is because students takingthese courses have a wide variety of

backgrounds in computing, and it ishard to present the course in a waythat does not bore some studentsand go over the head of others. Asecond reason is that the introduc-tory computer science coursetypically focuses on teachingprogramming skills. If it were toteach a sampling of the subjects thatare covered as part of the majorin much the way that chemistry orphysics departments commonlyorganize their introductory coursestudents might be more interestedand have a more informed view ofwhat constitutes IT work. A thirdreason is that some departmentsmake this introductory courseparticularly challenging (a rite-of-passage course) as a way to reduceenrollment pressures in the depart-ment so that the number ofstudents does not swamp the sizeof the faculty and the computingfacilities available.

Colleges are having difficultyattracting qualified instructionalpersonnel. The recent doubling ofnewly declared majors and the similarincrease in course enrollment willrequire many additional instructionalstaff. However, excellent opportuni-ties in industry, together with the lackof an increase in the number of newdoctorates being awarded in com-puter and information science, ismaking it difficult for schools torecruit new tenure-track faculty.

There are many highlyqualified IT workers in industrywho would enjoy a chance to teachcomputer science or information

67 Burt S. Barnow, John Trutko, and Robert Lerman, "Skill Mismatches and WorkerShortages: The Problem and Appropriate Responses," Draft Final Report, The UrbanInstitute, February 25, 1998.

68 For a discussion of various issues related to IT support on campus, including compen-sation, recruitment, and retention of IT staff, see www.educause.edu/issues/hrit.html

81 87

88

systems courses in the colleges part-time, in early retirement, or as acareer change. Universities alreadyemploy a large number of adjunctfaculty, especially to teach lower-levelcourses. There remain, however,considerable additional opportunitiesto use adjuncts from industry toteach in all parts of the curriculum.University regulations against long-term teaching appointments outsideof the tenure system and low adjunctpay scales, as well as companypolicies, are sometimes impedimentsto such teaching arrangements.

The number of women andminorities, other than Asian Ameri-cans, becoming majors or evenenrolling in entry-level computingcourses is small. There is anecdotalevidence that, of all the IT-relateddisciplines, women and minoritiesare most willing to enroll in informa-tion systems courses and majors, butnot in percentages that reflect theirnumbers in the general population orin the university student population.Possible reasons for the relativelygreater attractiveness of informationsystems are fewer mathematics andscience requirements than in com-puter science, as well as the percep-tion that information systemsinvolves more teamwork and ismore oriented towards applications.

Anecdotal evidence suggeststhat part-time students are enrollingat increasing rates, but no data arecollected on part-time students in ITdisciplines. Colleges need to offercourses at times that are moreconvenient to this student popula-tion, who are likely to be workingduring the day. One method fordoing this is to make the instructionasynchronous, which is the mode ofsome of the distance-learningprograms today.

There is also anecdotalevidence that students in computingprograms are increasingly likely toleave school and enter theworkforce before completing theirdegrees. While there may be short-term gains from early employment,educators are worried that theseworkers will not have the basicconceptual knowledge to keep upwith the rapid changes in the ITindustry and that emotional orfamily reasons will make it moredifficult for them to return toschool a few years later in order tolearn those basic skills.

What Is theRole ofGraduatePrograms inthe SupplySystem?

The main graduate degreeprograms lead to either the master'sor doctoral (doctor of philosophy)degrees, although a small numberof students receive a degree knownas the doctor of engineering degree.Graduate programs in computerscience will be covered first,followed by graduate programs inother computing disciplines such ascomputer engineering and informa-tion systems.

The master's program pre-pares graduates for higher-level jobsin information technology. Thereare two common tracks: profes-sional master's degrees, which aregeared to meeting the needs ofworking professionals and preparethem for immediate entry into thepractice, and research-preparatorymaster's degrees, which are designedto prepare students for study at the

8 2

doctoral level. A typical master'sprogram in computer scienceinvolves the completion of 10 to 12semester-long courses. Somespecialization or concentrationwithin a broad subfield of comput-ing, such as software engineering ornetworking, is possible and is oftenexpected. Master's programs allowa student opportunities to conductindividual research (in the form of athesis), enroll in advanced topicscourses that explore boundaries ofthe state of the art and practice ofthe discipline, and participate insmall group seminars that provideexperience with projects, presenta-tion, and teamwork.

For many employers of high-level information technologyworkers, it is the master's degree thatadds the greatest value. The educa-tion in professional and research-preparatory master's programs canbe very similar, but professionalmaster's programs tend to empha-size a balance between foundationsand practice, whereas the research-preparatory track gives greateremphasis to foundations. Theprofessional master's programssometimes has a greater emphasis oninteraction with industry and localbusiness; and, especially in informa-tion science departments, theprofessional-track faculty tend to befamiliar with business environments.

Doctoral programs areintended to produce high-endinformation technology workerswho possess cutting-edge knowl-edge of some particular technologyarea and are trained to carry outresearch. They involve additionalcourse work beyond the master'sprogram and rigorous examinationsensuring competency in a broadrange of topics. This breadth ofknowledge is complemented in the

later years of the doctoral programby gaining depth of knowledge in anarrow subfield, such as softwareengineering metrics or verification ofnetworking protocols. Specializationat the doctoral level may also involveother disciplines beyond the coreareas of computer science, such ascognitive science or bioinformatics.Emphasis is placed on learning theskills necessary to identi& importantunsolved problems within the area ofspecialization, defining and framingthem for solution, and discovering asolution through organized research.A substantial dissertation project thatresults in extending knowledge of thefield is a critical element of the degreeprogram.

Between the master's degreeand the doctor of philosophydegree in level of training is thedoctor of engineering degree inengineering schools (sometimescalled the 'doctor of arts' degree inliberal arts schools). It is similar tothe doctor of philosophy degree inrequiring the breadth of knowledgethrough course work and a com-prehensive exam, and depth ofknowledge through additionalcourse work and study. However,the doctor of engineering degreedoes not typically require a disserta-tion. It requires less time to com-plete than the doctor of philosophydegree, but still provides some ofthe same advanced training features.If there is continued high demandfor high-end information technol-ogy workers by industry, it may bedesirable to increase the number ofdoctor of engineering degreesawarded. Today this degree isrelatively uncommon; it is usedprimarily to give students in doctorof philosophy programs credit fortheir accomplishments if theydecide to leave school after their

: 3 89

90

Table 5-6Graduate Programs in Computer Science

Number of Number of

Academic Year Master's Doctoral

Programs Programs

1989-1990 311 100

1990-1991 314 100

1991-1992 319 103

1992-1993 316 112

1993-1994 325 117

1994-1995 339 119

Source: National Center for EducationStatistics,Digest of Education Statistics,using preliminary data for 1990-1991and final data for all other years. Nostatistics are available that break outprofessional from research-orientedmaster's programs.

course work is completed, butprior to writing a dissertation.

Academic quality and specifictraining offered are typically theforemost considerations when astudent selects a graduate school,especially at the doctoral level. Manystudents entering professionalmaster's programs are alreadyworking, and they continue to do sowhile they go to school. Thesestudents will often choose the bestacademic program close to theirworkplace, rather than the bestacademic program overall. It iscommon for graduate students toreceive financial support in the formof fellowships, teaching assistant-ships, research assistantships, orcompany-paid tuition. These

assistantships are an important partof the educational experience,providing valuable training inworking with groups, makingpresentations, writing proposals,and gaining practical knowledge.Because stipend levels in graduateprograms are significantly lowerthan industry salaries, some studentsfind it difficult to justify going tograduate school, especially for thedoctorate.

Table 5-6 shows the numberof master's and doctoral programsin computer science in the UnitedStates. During the first half of the1990s there was a 9-percent growthin the number of master's programsand a 19-percent growth in thenumber of doctoral programs. Ifthose growth trends have continued,today there are probably about 350master's programs and 140 doctoralprograms." To put these numbersinto context, in 1994-95 there were1,351 institutions awarding master'sprograms (in at least one discipline)and 482 institutions awardingdoctoral degrees (in at least onediscipline). Thus, graduate degreesare given in computer science at onlyabout 30 percent of the institutionsof higher learning that award at leastone graduate degree at the master'sor doctoral level. This apparentopportunity for growth must betempered by the recognition that itrequires significant numbers ofhighly trained IT workers to staffsuch programs, and that the kinds

69 The Computing Research Association maintains the Forsythe List of Ph.D.-grantinginstitutions in computer science, computer engineering, and closely related disciplines.The fall 1998 list counts 175 departments in the United States, as follows: computerscience 111; computer engineering 16; computer science and engineering 16; computerand information science 12; electrical and computer engineering 6; electrical engineeringand computer science 6; and one each for computer science and electrical engineering;electrical engineering and computer engineering; electrical and computer engineering andcomputer science; electrical engineering; department of engineering-systems; informationscience; management information systems; and information technology and engineering.

8 J

Table 5-7

Graduate School Enrollment inComputer Science

Academic Year

August 1990

August 1991August 1992

August 1993August 1994August 1995

August 1996

Number ofStudents Enrolled

34,00034,60036,30036,20034,10033,40034,600

Source: National Center for EducationStatistics, Digest of Education Statistics,using preliminary data for 1990-1991and final data for all other years.

of people who would make highlyqualified faculty members are thosewho are already in high demandfrom industry and other universities.Most newly formed graduateprograms in computer science arequite weakand continue to beweak for some years thereafter.Perhaps the greatest opportunity isfor starting professional master'sprograms in geographical regionswhere there is industry demand andwhere industry might be able toprovide some of the instructors onan adjunct basis.

The most reliable statisticsabout graduate enrollments incomputer science come from NSF.Table 5-7 shows that graduateenrollment in computer science hasbeen fairly steady throughout thefirst half of the 1990s, at about

35,000. The ratio of full-time topart-time students is about 1:1 andhas not changed much throughoutthis period. There are no statisticsthat distinguish master's students inthe professional track from those inthe research-preparation track, ormaster's from doctoral students.The Computing ResearchAssociation's (CRA's) Taulbee Survey,which is more up to date than datafrom NSF, shows new master'sdegree full-time enrollments (at thePh.D.-granting schools) of 3,400 inboth August 1996 and August 1997.This suggests that the steady master'senrollment is continuing. However,the Taulbee Survey shows 1,300 newfull-time doctoral students enrolledin August 1996 (an increase of 25percent over the previous year) and1,400 new doctoral students (an 8percent increase over the previousyear) in August 1997. Whether theseincreased enrollments in doctoralprograms will result in increaseddoctorates will depend on howsuccessful the universities are atretaining their students in the face ofa very good job market for ITworkers with advanced skills.

Computer science graduateprograms include enrollments oflarge numbers of foreign nationals.The National Software Alliancereported that, in 1994, 37.5 percentof the master's students in computerscience were foreign nationals, aswere 44.8 percent of the doctoralstudents. The CRA Taulbee Surveyfor that year showed that at least 41

70 The CRA Taulbee Survey showed 41 percent full-time and 21 percent part-time foreignnational students. However, the counts in the category for Asian-Pacific Islanders, whichwas supposed to include only domestic students of these backgrounds, lookedunusually large to the computer scientists responsible for managing the survey. Thereis a belief, which is impossible to support or refute, that some of the students reportedin the survey as Asian-Pacific Islander domestic students were actually Asian-PacificIsland foreign nationals on visas. If that is indeed the case, the percentages of foreignnationals are higher than the 41 percent and 21 percent reported.

85 91

92

Table 5-8Non-U.S. Employment of ComputerScience Doctorates Awarded in theUnited States

Survey Year Percentage

1994 18.0

1995 15.9

1996 9.0

1997 5.5

Source: Computing Research Associa-tion, Taulbee Survey.

percent of full-time and 21 percentof part-time students in graduateprograms in computer science andcomputer engineering were foreignnationals.'° There are indicationsthat this percentage of foreignnationals is slowly increasing. If alarge percentage of foreign nation-als return to their home countryupon graduation, this would implythat a significant fraction of thedoctorates produced in the UnitedStates would find employmentabroad. However, as Table 5-8indicates, there is a sharply decreas-ing trend of recent computerscience doctorates going abroad.Clearly, many of the foreignnational students are remaining inthe United States to work aftercompleting their studies.

The study group receivedqualitative reportsunsupported byany statisticson changes overtime in the employment practicesof foreign students educated in theUnited States. In the 1970s it wascommon for these students tomake concerted efforts to remain inthe United States upon graduation,given that most other countries didnot have good career opportunitiesfor them. In the 1980s a numberof countries, such as Taiwan andSingapore, sponsored nationalinitiatives to build up indigenous IT

8 6

industries. This led to a change inthe career patterns of foreignstudents in U.S. IT graduate pro-grams. The students usually re-mained in the United States forfour or five years to gain valuableon-the-job experience to supple-ment their formal education, andthen they would return home totake up positions of leadership. Inthe 1990s, with the greater entrepre-neurial opportunities in the UnitedStates than in most other countriesand the economic problems in Asia,there appears to be some swingback to a desire among thesestudents to remain in the UnitedStates for the long run.

Critics of the large number offoreign students trained in U.S.graduate programs often complainthat they are taking places awayfrom U.S. students, and that the U.S.government and American universi-ties subsidize the training of work-ers for other countries. There maybe some element of truth in thesecriticisms, but there are also goodreasons for the United States tocontinue this practice. Some ofthese foreign students do remain inthe United States and become partof the educated professionalworkforce of our country. Theseworkers give U.S. companies withglobal markets a competitiveadvantage through their knowledgeof the culture and the personalcontacts in their home countries. Ifthey return home to take up seniorpositions after completing theirformal education or after working afew years in the United States, theycan also be helpful to Americancompanies because they haveAmerican contacts a familiarity withU.S. practices. This increases theireffectiveness in working with U.S.companies wanting to enter theirhome country's market.

Table 5-9

Number of Master's and Doctoral Degrees Awarded in IT Fields,1992-1995

Master's Degrees

AcademicYear

ComputerScience

Managementinformation Science

Computer and DataEngineering Processing

Other Businessinformation

Systems1992-1993 10,163 1,592 2081993-1994 10,416 1,071 1,877 2631994-1995 10,326 1,040 2,012 394

Doctoral Degrees

AcademicYear

ComputerScience

ManagementInformation Science

Computer and DataEngineering Processing

Other Businessinformation

Systems1992-1993 805 0 0

1993-1994 810 123 0

1994-1995 884 140 3 3

Source: National Center for Education Statistics, Digest of Education Statistics.

Turning from enrollments todegree production, table 5-9 showsthe number of computer sciencemaster's and doctoral degreesproduced earlier this decade. Atthat time, master's production wasfairly steady, at slightly more than10,000 degrees per year. Morerecent data from the CRA TaulbeeSurvey suggest that this level ofproduction continued into 1997,possibly with a small increase (lessthan 10 percent in 1997).71 How-ever, Taulbee data suggest a de-

crease in Ph.D. production of about10 percent since the peak produc-tion in 1992.7' The study group hasestimated that about 70 percent ofstudents who enter graduate schoolin computer science eventuallygraduate with either a master'sdegree or a doctorate!'

Graduate degree programs inother areas of information technol-ogy (computer engineering, man-agement information systems, andother business information systems)

71, Extrapolation is needed to figure out trends from the Taulbee data. The Taulbeenumbers had to be increased to include the 10 percent to 20 percent of the PhD-grantingdepartments not responding to the survey, scaled up to consider master's degreesproduced by other than PhD-granting departments since the latter only produce about one-third of the master's, decreased for inclusion of computer engineering degrees in thestatistics, and further decreased for degrees awarded by Canadian schools.

72 CRA Taulbee statistics and NSF/SRS statistics track in close parallel with Depart-ment of Education statistics, but are about 10 percent higher.

73 Assume two years for a master's degree, and the initial two years plus an additionalfour years for the doctorate. Data show about 11,000 graduates per year and 35,000enrolled. Yield = (21,000 (2 x 10,500) + 3,200(4 x 800))/35,000, which equalsapproximately 0.69.

8 7 93

together produce about one-third asmany master's degrees as doescomputer science, and about one-eighth as many doctorates. There isevidence of steady growth in thenumber of master's degrees inmanagement information systems,but data are not available to deter-mine if this trend has continued inthe past several years. There isanecdotal evidence that manystudents, both those with technicaland those with non-technicalundergraduate training, chooseMBA programs for graduate study.In this case, their supplementary ITeducation is more likely to be in theform of continuing education,tutorials provided by professionalorganizations, or company training.

A number of qualitative issuesconcerning graduate education aroseduring the course of this study:

Many universities are havingdifficulty attracting qualified stu-dents, especially U.S. students, forgraduate study in computing fields.The study group estimates that only11 percent of those who receivebachelor's degrees in computerscience in the United States attendcomputer science graduate school inthis country."

Stipends for graduate studyare low, given the amount of timerequired to complete the require-

ments and the salaries available inindustry.

Traditionally, qualifying forfinancial aid has been based on theexpressed willingness of students tocontinue on for the doctorate. Therationale is that faculty preferdoctoral students over master'sstudents because they are likely to bebetter colleagues and be available towork on research projects forlonger periods of time with greaterinvolvement and responsibility.However, this preference maynegatively affect the ability to buildstrong professional master's pro-grams, which industry particularlyvalues. (This distribution pattern forfinancial support is still in force inmany computer science and com-puter engineering programs.However, in some professionalschools where there is a strongemphasis on master's programs,different sources and criteria havebeen established for providingfinancial aid to master's and doc-toral students.)

Anecdotal data suggest thatuniversities are having an increasinglydifficult time retaining their doctoralstudents to complete the degree, butwithout statistical evidence it is hardto assess the extent of the problem.This phenomenon appears to betied to the attraction of industrialpositions with high salaries, good

74 This is an estimate based on the following argument: Assume that 90 percent of theforeign nationals who do graduate study in computer science in the United States didtheir undergraduate study outside the United States, and that 100 percent of U.S.students in graduate school in computer science did their undergraduate work in theUnited States. The fraction of foreign nationals in graduate school in computer scienceis approximately 50 percent. These statistics imply that only about 55 percent ofthose in graduate school did their undergraduate work in the United States. In 1995,the number of computer science bachelor's degrees produced was 24,769 (NSF 98-307,Table 46). In 1996, there were 4,908 full-time computer science students in graduateschool for the first time (NSF 98-307, Table 26). The fraction of U.S. bachelor'sstudents going to graduate school is then approximately .55 x 4908/24769 = 0.11).

94 88

Figure 54

Career Choices of Computer Science Ph.Ds, 1994 and 1996

Post-docs13%

Other8% Industry

43%

Faculty Posts36%

Post-docs18%

Other6%

Faculty Posts26%

Industry50%

Source: National Center for Education Statistics, Digest of Education Statistics

facilities, and interesting work; andalso to the fact that academicresearch has taken an increasinglyshort-term focus, so that it is not asdifferentiated from industrialresearch as it used to be. Histori-cally, computer science graduateprograms attracted many ablestudents who wanted to work onmajor research problems that weremore likely to have long-term thanshort-term payoffs.

IT faculty are being over-loaded with work, which has anegative effect on their ability tospend time mentoring their doctoralstudents. Multiple factors contributeto the overload: the rapidly increas-ing undergraduate enrollments incomputer science; the increaseddemand on computer sciencefaculty to help their university tointegrate information technologyinto its own central managementand enterprises; the increaseddemand to help other departmentsincorporate information technologyinto their academic programs; andthe increased pressure to obtain

sponsored research. The overloadalso makes it more difficult torecruit and retain good faculty. Inaddition, faculty may be morereluctant to participate in cross-disciplinary programs, which mightbe good for computer science andfor the university, for fear that thissimply invites additional demandson their time.

Several people reported thatthe attractiveness of being a facultymember is rapidly diminishing, andthat many of the brightest studentsare choosing not to enter academiccareers once they witness first-handthe demands on their facultymentors. As figure 5-4 shows,there was an increasing tendencyfrom 1994 to 1996 for computerscience Ph.D.s to choose industrialcareers fresh from graduate school.The 1997 CRA Taulbee Surveyindicates that this trend is continu-ing. It shows that 53 percent ofthose whose post-Ph.D. employ-ment was known accepted indus-trial positions, compared with 48percent the year before. Table 5-10

8 9 95

96

Table 5-10Faculty Flight to Industry from Ph.D.Granting Departments of ComputerScience and Computer Engineering

Year Number

1993-1994 401994-1995 441995-1996 441996-1997 53

Source: Computing Research Associa-tion, Taulbee Survey.

shows a similar trend of facultyleaving their posts in increasingnumbers in recent years, althoughthe numbers are small.

DARPA officials haveindicated that in certain key appliedareas, such as software engineeringand database systems, universitieshave reported difficulties retainingfaculty and support staff, includingprincipal investigators in charge ofDARPA contracts. If this problemworsens, it could have a detrimen-tal affect on universities carryingout mission-critical research for theDepartment of Defense and othergovernment agencies.

Base faculty salaries tend tobe low compared with base

industry salaries (see figure 5-5),although faculty members do havea chance to obtain summer salariesand earn extra income by consult-ing. If faculty are able to obtainsummer salaries and a moderateamount of consulting revenue(amounting to a not-unusual 43percent of their base salary), thenthe total salary for an academiccomputer scientist is roughlycomparable to the total salary(base salary plus incentive bonuses,stock plans, and other variablesalary) of a computer scientist inan industrial research laboratory.However, it may not be easy for abeginning faculty member tosecure summer and consultingincome. Starting nine-monthfaculty salaries were mostly in the$50,000 to $65,000 range for the1998-99 academic year." Whilethese are good salaries, they arenot outstanding, given the talentand amount of training a newgraduate in computer and infor-mation science has acquired.There is some salary compression,so that mid-career faculty (associ-ate professors) often earn onlyslightly more than newly hiredassistant professors. This com-pression causes morale problemsand makes industrial alternativesappear that much more attractive.

75 The computer science faculty salaries are taken from the annual CRA Taulbee Survey,which is published annually in the March issue of Computing Research News. Seehttp://www.cra.org/statistics/. The Association for Information Systems, IS Worldnet,and the University of Pittsburgh have begun an annual salary survey for MIS faculty.The 1998 results can be found at http://www.pittedu/-galletta/salsurv.html

90

Figure 5-5

Twelve-Month Industrial Salaries Compared to Nine-MonthAcademic Salaries

02

7'.3U)

190,000

170,000

150,000

A130,000

P3110 000

'

90,000

70,000 dilli11111111"---W------''.;

50,000

0 5 10 15 20 25

Post Ph.D. Experience (Years)

Industrial-Min M Industrial-Mean A Industrial-Max0 Academic-Min 0 Academic-Mean 6 Academic-Max

Source: The academic salary figures are taken from the 1997-98 CRA Taulbee Surveyfor the twelve top-ranked computer science departments, assuming that an assistantprofessor has an average of three years' experience, an associate professor has anaverage of eight years' experience, and a full professor has an average of twenty-twoyears' experience. The industrial salary figures are taken from the 1997 CRAIndustrial Salary Survey.

9 19 7

Chapter 6. SupplyNon-Degree Programs

When people think of highereducation, they generally think ofstudents enrolling in colleges anduniversities to earn traditionalformal degrees. As the previouschapter discussed, there is a well-established system of formal highereducation for information technol-ogy (IT). The explosion of non-degree programs in informationtechnology, however, is less wellknown. Figure 6-1 shows thatwork-related training is offered tonearly three times as many people asare enrolled in traditional post-secondary education. Figure 6-2shows that even among those whoare in the post-secondary educa-tional system, less than one-third ofthose enrolled are traditional frill-time students.

Non-degree programs takemany forms: certificate and enrich-ment courses taught by colleges anduniversities at every level; trainingprovided by private educators,ranging from individual consultantsto large commercial educationalfirms; training associated withspecific IT products; businesses thattrain their own workforce; andcourses offered via distance educa-tion. The length and purpose of

these programs also vary widelyfrom the half-day course or shorterlecture and seminar that give anoverview of a specific topic, to thetwo-week course that provides aworking knowledge of a focusedtopic, to the six-month certificationprogram that offers in-depthknowledge of a focused topic, tothe apprenticeship that may lastseveral years and result in significantcareer progression.

These forms of educationappear to be growing very fast.However, because they are sovaried and non-traditional, there is alack of reliable data with which toevaluate their growth rates. For themost part, these forms of trainingprovide IT workers with knowl-edge and skills required to meetspecific vocational needs. Trainingtime is usually relatively short andcosts generally quite low. Thetraining is unlikely to providelonger-term foundational knowl-edge that would support a life-longcareer. People sometimes enroll inthese programs at their owninitiative to enhance their careers,and often because their companieshave asked them to acquire newskills and knowledge. Typical

9 99

Figure 6-1

Adult EducationTotal=59.2 Million Adults

WorkRelatedTraining

68%

VocationalSchools2%

Post-secondaryEducation24%

GED/BasicSkills 2%

ESL 3%

Lehman Brothers as taken from Jeanne C. Meister, Corporate Universities: Lessons inBuilding a World-Class Work Force (New York, NY: McGraw-Hill, revised and updatededition, 1998).

Figure 6-2

Post-secondary Education by lype of StudentTotal=14.6 Million Adults

TraditionalFull-time

Students32%

Non-traditionalWorking Adults32%

Part-timeStudents

32%

Lehman Brothers as taken from Jeanne C. Meister, Corporate Universities: Lessons inBuilding a World-Class Work Force (New York, NY: McGraw-Hill, revised and updatededition, 1998).

100ri

examples include the individual whois seeking:

training for a specific career(e.g., computer operator);

career advancement (e.g., acomputer operator training to be acomputer network administrator);

movement toward aprofessional career (e.g., a computeroperator seeking a job that normallyrequires a degree);

continuing education tomaintain up-to-date technical skills(e.g., an engineer seeking to learnabout a new technology);

a career shift (e.g., a me-chanical engineer retraining tobecome a software engineer); or

specific product informationor usage skills (e.g., how to use aspecific information technologyproduct).

Companies are also using thiskind of training to impart knowl-edge about the corporate missionand practices, to improve employ-ees' non-technical skills such ascommunication and teamwork, orto give them non-technical knowl-edge of best practices in theirindustry.

What Non-DegreePrograms DoTraditionalColleges andUniversitiesoffer?

Non-degree programs areoffered at every level from two-year colleges to graduate school.Certificate programs and shortcourses, both specializing in someparticular area within informationtechnology (such as networking,electronic commerce technology, orsoftware project management)."

The typical IT graduatecertificate program is an adultcontinuing education program,directed at a person who possessesa bachelor's degree (but not neces-sarily in an IT major). A workingknowledge of information technol-ogy may suffice to perform well inthese programs. At the end of thecourse of study, the student isexpected to have learned the basictheory and concepts of the particu-lar area of IT under study, and tobe familiar with the state of the artand trends. Graduates should haveacquired enough knowledge of thearea that they can keep their knowl-edge current as the field develops.The program will typically includebetween three and six courses at thepost-baccalaureate level, andpossibly an independent project.Completing these programs shouldextend the range of jobs availableto the graduates to include various

76 See IDC IT Training and Educational Services reports on "Worldwide and US ITTraining and Educational Markets and Trends" and "The Business of IT Certification:Market Potential and Trends." The reports can be purchased by contactingctoffel@idcresearch.com.

94 101

occupations that involve creating,extending, or tending technicalinformation systems.

Many two-year colleges havecertificate programs in informationtechnology. These programs trainpeople for occupations that areconsidered highly skilled in com-parison to most occupations butare, in fact, among the less skilledkinds of IT work. They generallyinvolve the tending or extending oftechnical information systems, andcarry titles such as computer pro-grammer, office systems specialist,network technician, computerrepairer, multimedia and Webdesigner, or vendor-specificcertified technicians (such as aMicrosoft or Novell technician).The students entering these pro-grams include persons who are justbeginning their education or theircareers, those who have nowreceived non-IT bachelor's degreesand may have some work experi-ence but who want a better-payingor more fulfilling job, as well asothers.

The offerings of the four-year colleges and teaching-orienteduniversities are somewhat harderto classify. They are targetedsomewhere between the offeringsof the research university and thetwo-year college, and they containelements of each. They offermoderately low-level courses intopics such as Web design orvendor-specific certification,undergraduate-level courses intopics such as Java programming,and post-baccalaureate coursessimilar to those offered in theresearch universities. Often thesecourses are tailored for specificcompanies that are located near theschools. This has become a majoractivity and significant source of

102

revenue for some schools, such asthe American University in theWashington, DC, region andBentley College in the Boston area.

What OtherGroups SupplyNon-DegreePrograms?

In addition to two- and four-year colleges and the universities,many other groups supply non-degree training in informationtechnology. One is the vocationaltraining school, such as DeVry orIIT Training Institute. These schoolsoffer programs typically aimed atindividuals who do not have acollege degree. The nature of theinstruction is primarily vocational,and the training programs typicallyprepare the students for specificjobs in the lower-end occupations inthe IT workforce.

Private educators also supplytraining. Some are individualconsultants, others are well-estab-lished firms such as SoftwareLearning Associates. These privateeducators offer mainly consultingservices or short courses focused onspecific IT skillsranging fromproject management, to computerprogramming, to people manage-ment skills, to the use of specifictools and IT products. Courses areavailable at every level of skill,directed at virtually every occupa-tion within the IT field.

Product suppliers oftenprovide training for people whowill use or maintain their productsat customer sites. Many IT prod-ucts are very complicated, makingtraining essential in these cases.

r

Concerned about the technicalcapabilities of people who are usingand maintaining their products, ITvendors such as Microsoft andNovell have developed more than150 vendor-specific certificationprograms for technicians. Certifica-tion assures customers that they arehiring qualified people to use ormaintain these IT products. Thesecertification programs often arelicensed to other kinds of suppliers,such as independent educators andtwo-year and four-year colleges. Atwo-year college might want, forexample, to offer a certificateprogram in networking, and it mustchoose some particular technologyto use in teaching general principles;to meet this need, it might chooseto license a particular vendor'straining program. This can be veryattractive to college administratorsor state legislators because thevendor may be willing to: guaran-tee employment for all studentswho successfully complete thecertification process; provide acompletely developed curriculumand curricular materials; help trainthe school's faculty to teach theprogram; and perhaps even provideinstructional equipment free ofcharge or at reduced cost. Thesevendor-specific programs may alsohave a downside, however. Someeducators are concerned because ofthe undue influence they are havingon the design of curricula.

What Is theRole ofCorporateUniversities in?Mining andEducating ITWorkers?

Corporate universities haveexisted since the 1950s, but duringthe last decade the number hasmultiplied from 400 to more than1,000 (at the same time that morethan 200 accredited colleges havegone out of business)." Corporateuniversities may now be the fastestgrowing sector of higher education.They have developed because ofthe widely perceived need in thecorporate sector for life-longlearning, and the belief of manyemployers that the existing post-secondary system is not able todeliver what they want their em-ployees to learn. There is a widelyheld view in corporate managementthat most workers in the future willbe "knowledge workers," who willneed to keep abreast of the latestdevelopments in their rapidlychanging field; thus these managersregard continuous education as animportant way to keep the com-pany competitive. Given that manycompanies headquartered in Europeand Asia are spending two to fivetimes as much on training asAmerican companies, this is oftenviewed as a global competitivenessissue.

Ten years ago corporateuniversities usually had physicalcampuses, just like accredited

77 The material for this section is taken primarily from Jeanne C. Meister, CorporateUniversities: Lessons in Building a World-Class Work Force (New York, NY: McGraw-Hill,revised and updated edition, 1998).

103

universities. Some still do.Motorola University, for example,has campuses around the world.Others, such as Dell University orSun University, have no campus atall. The emerging view of,thecorporate university is as a processrather than a placea process forproviding life-long learning to allwho are involved in the company'swell-being (employees at every level,customers, suppliers, dealers,distributors, wholesalers, etc.).Physical locales are increasinglybeing considered not so much asthe principal places of learning, butas a good place to carry out onespecific aspect of learningthesharing of best practices. Thesephysical classrooms are supple-mented by many technical ways ofdelivering courses at a distance.

Corporate universities usemany technologies in the learningprocess, such as satellite broadcasts,video and audiotapes, knowledgedatabases, tutorial CD-ROMs, andInternet/intranet-based courses.The amount of learning carried outthrough the use of technology isincreasing rapidly, and a recentsurvey indicates that half of allcorporate training will occur in thisfashion by the year 2000." This newway of delivering learning workswell for employees, suppliers, andcustomers who are spread through-out the world and who cannotafford the time or expense to cometo a physical site for training. It alsofits well with the rapid pace atwhich the business world, and theknowledge required to be effectivein it, is changing.

More than 4 million peoplealready learn through corporate

universities and their learningpartners (traditional training firms,accredited universities, for-profiteducational firms, etc.). Nearly halfof all corporate universities alreadyhave some kind of alliance with anaccredited educational institution,and 40 percent of corporateuniversities expect to start grantingaccredited degrees in collaborationwith an accredited college oruniversity. The degrees they conferrange from the associate's throughthe master's level. For example,AT&T School of Business partnerswith the University of Phoenix,which is one of the fastest growingfor-profit educational firms (havinggrown from 3,000 to 40,000students in the past decade). AT&Temployees can take courses thatcount toward a degree at more than50 University of Phoenix campusesand learning centers around thecountry. The corporate universitiesregard the traditional colleges anduniversities as suppliers, just like theirsuppliers of raw materials, parts,and services. They bring all suppli-ers into their educational programto instill shared values and ap-proach. Thus Motorola partnerswith various two- and four-yearcolleges not only for the traditionalteaching services these schools canprovide, but also so that the schoolswill learn what skills and knowledgeMotorola wants to impart to itsworkforce.

What corporate universities areteaching is driven by what thecompanies believe they need mostin order to succeed in businesstoday. In this respect, corporateuniversities differ markedly fromtraditional post-secondary educa-tion, which has basic education as its

78 Annual Survey of Corporate Future Directions, as cited in Meister.

104

primary focus. Some technical skillsare taught in the corporate universi-ties, but much of what is taught isnon-technical. The corporateuniversities seek to develop corpo-rate citizenship (learning the cultures,values, traditions, and vision of thecompany), provide a contextualframework to the company (learnthe company's customers, competi-tors, and industry's best practices),and teach core competencies. Thesecompetencies include:"

learning skills

communication andcollaboration

creative thinking andproblem-solving

technological literacy

global business literacy

leadership development

career self-management

The direction in whichcorporate universities are headed isclear from examples of four typesof programs that have been success-fully placed in practice. The ex-amples are not taken from the ITindustry and do not apply specificallyto IT workers, but it is clear thatthese approaches could and prob-ably will be applied in this domain.

1. Customized executiveeducational programs, ranging fromshort courses to full-fledged MBAprograms, allow corporations tocustomize a program that uses itsown corporate culture and com-pany-specific case studies in the

79 Meister, p. 13.

courses. This is what the WhirlpoolCorp. has done in collaborationwith the University of Michigan,Indiana University, and a Frencheducational institution.

2. Some corporations allythemselves with colleges anduniversities to develop an accred-ited, customized training curriculumthat teaches the exact skills thecorporation needs for specific jobcategories. For example, MegatechEngineering has joined with CentralMichigan University to develop abachelor's degree program inautomotive vehicle design toalleviate a perceived shortage ofautomotive designers. MegatechAcademy, which is located on thepremises of Megatech Engineering'svehicle design building complex, hasreceived accreditation from theState of Michigan to award thisbachelor's degree. Megatechemployees can also take thesecourses toward a design certificate.

3. Corporations and universi-ties are beginning to form consortiaso that a group of companieswithin an industry can gain from asingle corporate university, or sothat one or more companies cantake advantage of faculty talentfrom a collection of universities.For example, the Southern Com-pany and eleven other companies inthe Atlanta area have formed aconsortium with Emory Universityfor a three-week training coursespread over a four-month period toteach corporate strategy, globalbusiness environment, and leader-ship skills. United Healthcare andUnited Technologies have enteredinto a consortium with RensselaerLearning Institute, which brokers

105

courses from Boston University,Carnegie-Mellon University,Stanford University, and MITthrough Interactive CompressedVideo technology available to the200,000 United workers at speci-fied work sites.

4. Some corporate universi-ties have decided to becomeaccredited on their own. Anexample is the Arthur D. LittleSchool of Management, formedoriginally to handle the corporatetraining requirements of Arthur D.Little clients around the world, andnow an accredited program ofThe International Association ofManagement Education.

The corporate universities areable to respond to change muchmore quickly than the traditionalcolleges and universities. Corpo-rate university courses are some-times placed online within a weekof their being adopted, whereas itusually takes a college a year toimplement a new course. DellUniversity updates its catalog ofofferings every two weeks, com-pared with every one to two yearsfor the typical university.

What Is theRole ofDistanceLearning inEducating theIT Workforce?

The success of corporateuniversities is likely to be depen-

dent on distance learning. Distancelearning may also become asignificant tool for those whosupply traditional post-secondaryeducation. Correspondence schoolshave existed since the nineteenthcentury, but the introduction ofmany new technologies is raisingnew possibilities and challenges foreducators."

Distance learning has ad-vanced in a number of ways. Inthe self-learning area, there areimproved methods, such asbroadcast television, rule-basedsoftware, CD-ROMs, and video,for delivering distance education. Itis sometimes difficult, however, toknow how much someone haslearned from courses based onthese media. Many believe thatstudents learn more effectivelythrough interactive processes, suchas interactive television. Theinteractive approach has severaladvantages in that an instructor candirectly apply many of his or herskills from the traditional class-room, and the only start-up costsare the capital equipment and theleased line. (There is no major up-front curricular development cost,although it may take some effortand learning, including some newskills, to produce visuals that canbe televised effectively.) Oneshortcoming is that the instructionis restricted to a specific time andplace. To overcome these limita-tions, some people are proponentsof asynchronous distance learning,which can occur at any time andany place, within certain limits. Thechallenge is to build an affordable,user-friendly system that allows

80 This section is based largely on discussions with Michael Teitelbaum and FrankMayadas of the Alfred P Sloan Foundation, an organization that is working to makedistance learning a reality.

106 a CI%15

interactive teacher-student andstudent-student exchanges aboutproblems and concepts.

Most asynchronous distancelearning courses in operation todayare experimental. However, anumber of projects are now underway to develop systems that can beplaced into regular operation, meetrealistic learning objectives, and arenot too expensive for students.Most of these systems involve Web-based learning, perhaps supple-mented by a course textbook orsome other technology such as CD-ROM. Schemes have been devel-oped, for example, that allow aninstructor to ask questions, requireanswers to be turned in, gradeassignments and give individualstudents feedback, and post modelanswers. There are also ways for allthe students to see answers given toan individual student's questions, andways for students to work with oneanother. It appears that experimen-tation will continue about how tobuild such systems, and continuedprogress is likely.

Some of the barriers toimplementing distance learningsystems have been overcome in thepast five years. Communicationexpensesfor dedicated data linesor national toll-free numbersprovided by the supplier, forexamplewere extremely high forearly systems. Today, a student cansign up for $20 or less per monthwith a local Internet service pro-vider and have communicationsaccess. The costs of the computersused by students to obtain theireducation have also fallen, frommore than $3,000 to less than$1,000 for an adequate machine. Asmore people become familiar withthe Internet and the World WideWeb, it is likely that prospective

students and employees will increas-ingly consider distance learning aconvenient alternative to moretraditional educational methods.

While progress has been madein distance education, some barrierspersist. Video delivery systems needto improve; the pedagogy of dis-tance learning warrants furtherstudy; and economic factors are stilla consideration. Institutional barrierscan also be strongin fact, ITcompanies themselves have notbeen leaders in the adoption ofasynchronous distance education.

It is too early to tell whetherteachers will like this format. Onepotential advantage is that once aquestion has been answered for onestudent, it can be recorded perma-nently and made available ondemand to all the other students.The asynchronous mode also seemswell suited to teaching technicalsubjects because a student who getsstuck on a problem can put it asideto think over or ask for a hint fromthe instructor or other class mem-bers. Instructors have also foundthat the format works for groupdiscussions, and that technical designproblems can be solved by theclass as a whole and result insatisfactory designs. Many of theinstructors who have tried asyn-chronous distance learning haveliked it, but they are the earlyadopters that one encounters withany new technology; and some-times the reactions of the largercommunity are not as positive asthose of the early adopters.

The Alfred P. Sloan Founda-tion is working with thirty-fivedegree-granting institutions, includ-ing Stanford, Pennsylvania State,Drexel, Johns Hopkins, and theUniversity of Illinois, to develop

130'107

distance learning programs. Hereare three promising examples:

Drexel University is offeringa master's degree in informationsystems entirely by asynchronousdistance learning. The courses,exams, and degree awarded areexactly the same as Drexel offers ina traditional classroom format oncampusand the success rate in thetwo formats is comparable.Currently the distance learningformat is marketed only to majorcorporations in the Philadelphia-New York area, such as Metropoli-tan Life, Vanguard, Cigna, Unisys,and Merck. These companies selectprogrammers and middle-levelmanagers from their informationsystems departments to enroll in theprogram. The companies find thisarrangement are extremely attractivebecause the employees can continueto work full-time and take theircourses from home or the office.Employees do not have to take aleave of absence from work for ayear or two to complete the degree,and they typically finish the courseof study in about the same amountof time as a part-time student whocomes to campus for classroominstruction.

The University of Califor-nia-Berkeley Extension Programoffers more than 200 asynchronouslearning courses at reasonable pricesto individuals. These courses, whichare distributed by America Online,range across many disciplines. Afew of them concern informationtechnology. However, BerkeleyExtension is not authorized by theUniversity of California to issuedegrees or certificates, whether thecourses are taught online or in theclassroom.

108

The National AdvisoryCoalition for TelecommunicationsEducation and Learning(NACTEL) has recently reached anagreement to support an asynchro-nous online Associate of AppliedScience degree in telecommunica-tions. NACTEL is a partnership oflarge telecommunications compa-nies and unions (Bell Atlantic,Communications Workers ofAmerica, GTE, InternationalBrotherhood of Electrical Workers,SBC, and US West) that representmore than 500,000 telecommunica-tions workers. NACTEL is con-cerned about a serious shortage oftrained telecommunications techni-cians able to handle the rapidlychanging telecommunicationstechnologies of advanced tele-phones, caller ID, voice mailboxes,Internet services, fax machines, andmodems. The degree program willbegin with 100 students in 1999,and it is expected to grow to 3,000students by the third year. PaceUniversity, with guidance fromNACTEL, will prepare courses intechnical mathematics, telecommu-nications, electronics, and networks.

IssuesA number of qualitative

observations and issues about non-degree training arose in the courseof the study:

There is a lack of standards,consistency, and quality control inthe non-degree training market.There are essentially no standards oraccreditation processes. Individualsand employers do not have a goodunderstanding of what they arebuying, and employers have adifficult time evaluating the knowl-edge and skill sets employees haveacquired from such training. Stan-dards alone do not guarantee qualityprograms, but it is an important

first step. What exists today isbasically a caveat emptor market.

There are low barriers toentry in providing certain kinds oftraining, such as short courses. Insome cases all that is needed is ameeting room and a single, knowl-edgeable instructor. Companieswho want training for their employ-ees will often supply the meetingroom, and it is not costly for thetrainer to rent a classroom at a hotelor community center on a dailybasis, if need be. Ease of entry intothe training domain, however, alsomakes it more difficult to enforceminimal standards on the kind oftraining offered.

Non-degree programsprovide a good opportunity forexperienced practitioners to pass ontheir knowledge to others. Suchprograms are popular with earlyretirees because of the flexibilitythey offerworking out of thehome, flexible schedules, and settingones own pace.

While it is expensive to run afull-scale corporate university, manyof the associated technologies arerapidly becoming less expensive.Tools for course preparation (e.g.,for Web-based courses) are beingdeveloped, and a body of knowl-edge is accumulating about how touse these technologies effectively.Thus it is anticipated that the costs,effort, and knowledge to runcorporate universities will lessen to apoint that many more workers willhave access to them.

Is RetrainingOccurring, andif so, How LongDoes it Take ToRetrain for anIT Job?

Little information is availableabout the nature and extent ofindustry's commitment to retrain ITworkers, although some is availableon corporate retraining in general.It appears that companies areincreasingly prepared to devoteresources to retraining in order tokeep their existing workforce up todate and productive. There is littleevidence, however, that companiesare willing to spend resources onentry training that would allow aperson with good general skills tochange occupations at companyexpense. Many companies stillbelieve it is the individual's responsi-bility to learn the basic technologicalskills the occupation requires beforebeing hired. Even with the basicoccupational skills in place, ittypically takes a new employee sixmonths (or more, in the case ofpositions that involve advancedskills or extensive education) to learnenough about the organization andthe particular job to be fully pro-ductive. Some companies, espe-cially small ones, have troubleaffording the costs of carrying aless-than-fully productive employeefor six months or a year, much lesshaving the responsibility as well forproviding the basic occupationalskill training.

The news is more positivewhen considering job seekers.There seems to be a large pool ofpeople willing to retrain for ITwork. Presumably this is because

102 109

of the widely perceived view aboutthe availability of jobs, good wages,the professional nature of the work,the opportunities to do it withoutrelocating, the increasing number ofpeople who have some familiaritywith computers and networks, andthe intrinsic interest of the work.An examination of three institutions(Northern Virginia CommunityCollege, the Applied ManagementInstitute in Omaha, and ClaytonCollege and State University inGeorgia) confirms that significantnumbers of persons twenty-fiveyears or older will elect to pursue ITpreparation if it is available at times,in places, and in ways that accom-modate the constraints of theirlifestyles. These schools have seenpeople with both baccalaureate andadvanced degrees seeking bothcredit and non-credit IT coursesand training.81

It does not take all that longfor someone with good academicskills to retrain for many IToccupations. The length of thesetraining programs varies. Forexample, about six months of full-time training can prepare anindividual with some scientificprogramming experience tobecome an entry-level Unixsystems administrator. A two-yearassociate's degree program canenable a high school graduate towork as a computer programmeror a computer maintenancetechnician. The most advanced IToccupations are not generallyaccessible to people with this kindof retraining, but instead require anadvanced degree in an IT-relateddiscipline that may take many yearsto complete.

81 The Winter 1999 catalog for the Loudoun Campus of Northern Virginia CommunityCollege (only one of several of their campuses) gives a sense of the popularity of theseprograms. The catalog of IT courses ran for more than 35 pages (about 300 courses).They included courses associated with certificate programs in Cisco networking, client/server computing, C/C++, Oracle database, Microsoft Solution Developer and Program-mer, Powerbuilder, Networking, Virtual Reality, Web and Internet Programming, ComputerAided Design, Multimedia, PC Engineering and Computer Repair and Support, DigitalPhotography, Configuration Management Unix, and GIS; plus noncertificate programcourses in application development and design, databases and spreadsheets, Internettopics, office applications, publishing and graphics, ISO 9000, and general computing.

1 I') 0u110

Chapter 7. Women, Minorities, and Older Workers

Several groups of Americansare represented in the informationtechnology (IT) workforce inpercentages that are far lowerthan their percentages in thepopulation as a whole. Theseinclude African Americans,Hispanics, Native Americans, andwomen generally. There are alsoopen questions about the repre-sentation of older workers in theIT workforce.

How DoWomen Relateto the WorkerShortage?

If the number of women inthe IT workforce were increasedto equal the number of men, eventhe tremendous shortages of ITworkers noted in the ITAAstudies could be filled. However,according to the Department ofCommerce, only 1.1 percent ofundergraduate women choose IT-related disciplines as compared to3.3 percent of male undergradu-ates.

Tables 7-1 and 7-2 providestatistics about the percentage ofwomen being educated in ITfields. Table 7-1 shows thenumber of women in formaldegree programs in computer andinformation science at all U.S.colleges and universities, whereastable 7-2 shows the number ofwomen in formal degree pro-grams in computer science andcomputer engineering at only thePh.D.-granting institutions. Table7-2 works from a smaller sample,but it provides more currentinformation. The percentages ofwomen in bachelor's and master'sprograms are much lower in table7-2, which is attributed not to anymethodological problem witheither data set, but rather to thefact that table 7-1 includes infor-mation systems degrees and table7-2 does not. There is anecdotalevidence that women are enteringinformation systems programs ingreater percentages than computerscience and computer engineeringprograms. The reason given is thatinformation systems is perceivedas more people-oriented andmore attuned to the uses ofinformation technology, whereascomputer science and computer

1 V, 111

Table 7-1

Number of Degrees Awarded in Computer and Information Sciences by Level and Gender

Academic Ph.D.s MS BA/BSYear Awarded % Women Awarded % Women Awarded % Women

1984-1985 240 10.0 6,942 28.9 38,589 36.81986-1987 374 13.9 8,481 29.4 39,590 34.71989-1989 551 15.4 9,414 28.0 30,454 30.81989-1990 627 14.8 9,677 28.1 27,257 29.91990-1991 676 13.6 9,324 29.6 25,083 29.31991-1992 776 13.8 9,534 27.8 24,578 28.71992-1993 808 14.7 10,171 27.1 24,241 28.11993-1994 810 15.4 10,416 25.8 24,200 28.4

Source: National Center for Education Statistics, Digest of Education Statistics.

Table 7-2

Academic Ph.D. MS BA/BSYear Awarded % Women Awarded % Women AwarckW % Women

1984-1985 326 11.0 -1985-1986 412 12.1 -1986-1987 559 9.7 -1987-1988 744 9.0 5,159 12,6871988-1989 807 13.3 5,457 10,6061989-1990 907 12.6 5,116 9,6811990-1991 1,074 12.1 4,993 9,3531991-1992 1,113 11.3 5,121 9,8131992-1993 997 13.3 4,523 8,2181993-1994 1,005 15.6 5,179 19.1 8,216 17.91994-1995 1.006 16.2 4,425 19.7 7,561 18.11995-1996 915 11.7 4,260 20.0 8,411 15.91996-1997 894 14.4 4,430 22.3 8,063 15.7

Source: Computing Research Association, Taulbee Survey. 1984-86 Ph.D.for CS&CE departments, all other years CS departments only.

numbers

engineering are more focused onthe technology itself.82

One of the obvious patternsin these two exhibits is that thepercentage of women entering thecomputer science pipeline andearning the bachelor's degree inthese IT fields has been dropping

steadily since 1984. While thenumber of computer and informa-tion science degrees awardeddecreased every year between 1986and 1994, the decrease is occurringat a faster rate proportionately forwomen. This is in contrast togeneral trends in the graduationfigures of U.S. colleges and universi-

82 For a more general discussion of information systems versus computer scienceprograms and the demand for them, see http://www.ne-dev.com/ned-01-1998/ned-01-enterprise.t.html

112

0 5

ties during these same years, whenthe percentage of bachelor's degreerecipients who were womenincreased from 50.8 percent to 54.6percent. It is also in contrast to thetrends in scientific and engineeringdisciplines generally. The decrease inbachelor's degrees awarded towomen has also affected thenumber of women in the graduatedegree pipeline, contributing to thedecrease in women completing amaster's degree in the computer andinformation sciences area. Thepercentages at the doctoral levelhave stayed somewhat flat, with areduction in the number of U.S.women apparently offset by anincrease in the number of femaleforeign students entering the systemat the graduate level. There are noreliable data on the number ofwomen in the IT workforce.

The decline in womenengaging in formal IT training since1984 is in sharp contrast to thepattern of the late 1970s and early1980s. In that period, concertedefforts were made to recruitwomen to the field, and theseefforts resulted in a rapid increasein the number of women students.Thus the subsequent decline in thepercentage of women entering thefield is especially disheartening.However, the experience of theearly 1980s shows that program-matic efforts can make a significantdifference. There has been muchspeculation about the causes for thedecline in women entering the ITtraining pipeline." Some of thereasons that have been suggestedinclude:

The lack of equipment in thehigh schools, so that women do notgain early experience with thetechnology.

The way in which computersare presented to many high schoolstudentsoften in terms ofaggressive or violent games. Thesegames tend to involve a small set ofskills, which boys seem to enjoymastering by playing the gamesagain and again, while girls seem toget bored with the repetitiveness.

The lack of K-12 teachersand guidance counselors who areknowledgeable about the widevariety of career paths and oppor-tunities in IT.

The image of computing asinvolving a lifestyle that is not wellrounded or conducive to family life.

A perceived image of ITwork as being carried out in anenvironment in which one has todeal regularly with more competi-tion than collaboration.

Differences in socializationbetween men and women aboutwhether they are performing wellacademically, which may encouragemen and discourage women fromthe study of information technol-ogy in collegeeven when the maleand female students are performingequally well academically.

A perception of computingas a solitary occupation, not wellintegrated into social discourse orsocial institutions.

83 Some methodologically rigorous research on this issue is under way. See AllanFisher and Jane Margolis, Computer Science Department, Carnegie Mellon University, on"Women in Computer Science: Closing the Gender Gap in Higher Education" (http://www.cs.cmu.eduk-gendergap).

1 6 113

Table 7-3

Ph.D. Degrees Awarded in Computer Science and Engineering By Minority Ethnicity

AcademicPh.D.

AwardedAfrican-

American HispanicNative Asian or

AmericanPacific

islander Other

Year # % # % # % # % # % ,

19841985 326 3 1.0 7 2.1 92 28.2

1985-1986 412 6 1.5 6 1.5 151 36.7

1986-1987 559 3 0.5 9 1.6 197 35.2

1987-1988 744 6 0.8 8 1.0 281 37.8

1988-1989 807 0 0.0 12 1.5 - 299 37.0

1989-1990 907 4 0.4 11 1.2 281 31.0 148 16.3

1990-1991 1,074 8 0,7 26 2.4 349 32.5 151 14.0

1991-1992 1,113 11 1.0 17 1.5 412 37.0 131 11.8

1992-1993 997 7 0.7 13 1.3 319 32.0 118 11.8

1993-1994 1,005 14 1.4 9 0.9 0 0 154 15.3 76 7.6

1994-1995 1,006 9 0.9 28 2.8 1 0 149 14.8 92 9.1

1995-1996 915 11 1.2 27 3.0 5 0 143 15.6 59 6.4

1996-1997 894 6 0.6 3 0.3 0 0 107 12.0 60 6.7

Source: Computing Research Association, Taulbee Survey.

A perception that softwarejobs are not family-friendly (e.g.,long hours, lack of awareness ofopportunities for telecommutingand other flexible schedules).

Courses in mathematics andscience that are requirements fordegree programs in computerscience and computer engineering,which women have not beenencouraged to pursue based onoutdated stereotypes of aptitudeand interest.

The lack of women role models.

The large percentage offoreign-born teaching assistantsand faculty, some of whom havecultural values that are perceivedas not being supportive ofwomen being educated or joiningthe workforce.

Concerns about safety andsecurity felt by women and theirfriends and families about womenworking alone at night and onweekends in computer laboratories.

114 4,4

How DoMinoritiesRelate to theWorkerShortage?

The number of persons frommost minority groups either trainingor working in information technol-ogy occupations is very low. Oneprobable reason is the small numberof minority students movingthrough the educational pipeline.Considering only those studentswho graduate from college, thepercentages of Native Americans,African Americans, and Hispanicsreceiving a degree in computer orinformation science is actually higherthan the percentage among non-Hispanic white males. However,this promising statistic is more thanoffset by the fact that minoritiesattend college in much lowerpercentages than whites do. Table7-3 shows the low percentages ofAfrican Americans, Hispanics, and

Native Americans training in IT-related disciplines.

Many of the reasons thatdiscourage women from IT careersalso apply to minorities. There arevery few minority role models ininformation technology. Minoritystudents are less likely to havecomputers at home or at school onwhich to gain early exposure toinformation technology." Studentswho attend historically blackcolleges and universities facelimited computing facilities,compared with the average U.S.college or university. But there areother reasons as well. For ex-ample, minority students who wantto devote their lives to helping theircommunities do not regardinformation technology as a social-conscience field. Students with thatgoal are much more likely to trainfor careers in law, medicine, orpolitics.

One intriguing piece ofanecdotal evidence is that AfricanAmericans who have receiveddoctoral degrees in computerscience over the past decade haveoverwhelmingly chosen (more than90 percent) industrial rather thanacademic careers. Speculation isthat companies have done a betterjob than universities at makingdiversity an integral part of theirorganizational values and salariesare better. Also, minority employ-ees are more likely to be able to geton with their work in companieswithout being burdened by heavycommittee workloads that theywould experience in universities

that are reaching out for diverserepresentation.

How Do OlderWorkersRelate to theWorkerShortage?

There is a common percep-tion that information technology isan occupation for younger work-ers. We all have the image of theyoung programmer staying awakeon massive doses of caffeine whileundertaking a thirty-six-hourprogramming session. There issome truth to this image. Pro-gramming is increasingly becominga skilled entry-level position, andmany of the programming posi-tionsbut by no means all ofthemare populated by recentcollege graduates.

There is a less savory side tothis image, howeverthat of theprogrammer being washed up andput out of a job by the uncaringemployer by the age of 40. Howare older information technologyworkers treated? Unfortunately,there is very little credible evidenceon which to judge. Given the highdemand for IT workers and thepremium that companies claimthey place on teamwork, commu-nication skills, and knowledge ofthe business and industry, onewould expect companies towelcome older workers. It hasbeen reported to us that less thantwo percent of programmers over

84 Minority students who do have computers in their schools are more likely to usethem for repetitive math skills, instead of simulations and real-life applications ofmathematics concepts. See Educational Research Service, "Does it Compute? TheRelationship between Educational Technology and Student Achievement in Mathematics."

ri 57; 115

age 40 are unemployed, but wehave been unable to track downthe hard data to support thisstatement.

If one considers olderunemployed workers in otherprofessions who want to becomeIT workers, the scant and anecdotalevidence presents a much less rosypicture. The downsizing of thedefense industry and of corporateAmerica more generally during the1980s created unemployment for asignificant number of electricalengineers. The IEEE-USA hasrecently completed a survey of itsunemployed members, although thesample size is small (335 of 1,288 =26 percent). Their study indicatesthatholding constant highestdegree, Internet access, dependenceof job on government funding,whether respondent is retired/voluntarily unemployed, and typeof job search techniqueeach yearof age above 45 adds three weeksto the duration of unemployment."The IEEE has been appropriatelycautious in reading evidence of agediscrimination into these statistics:

Overall, the survey indicatesthat age has a persistent effect onthe duration of unemployment, but itcannot be determined whether thatis attributable to productivitydifferences, price differences, agediscrimination or some other factor.For example, the longest duration ofunemployment occurs in thedefense industry and the secondlongest unemployment periodoccurs in the aerospace industry.Older engineers are more likely tobe working in aerospace anddefense and these industries aremore likely than others to rely ongovernment funding; thereforeindustry, and not age alone, canaccount for unemployment.86

Given the lack of statisticaldata or other close examination ofthis issue, the study group looksforward to the results of therecently inaugurated NationalResearch Council study that willinclude an examination of olderworkers. Until then, all discussionof this issue is likely to bespeculative.

85 Chris Currie, IEEE-USA External Communications Supervisor, personal communication,December 11, 1998.

86 Annette Codispotti, Assistant Editor, "Survey Gives IEEE-USA a Picture of UnemployedEngineers," The Institute, IEEE, February 1999.

1161 ;I Q

1..)

Chapter 8. Seed-Corn Issues

Is the StrongIndustrialDemand for ITWorkers Harm-ing the Educa-tional System?

Many educators, industriallaboratory leaders, and govern-ment science officials are con-cerned that the high industrialdemand for information technol-ogy (IT) workers will siphon outof the educational systems manystudents who would otherwisepursue an advanced degree. Thisdiminishes pool of people whowill join the university facultiesthat perform basic research andteach the next generation ofstudents. This problem is com-pounded when industry alsosuccessfully recruits current facultymembers, including junior facultywho would become the academicleaders of the profession in thecoming decades. This is known asthe "seed-corn" problemananalogy to those who consumetoo much of this year's crop,reserving too little for next year's

planting. A similar situation oc-curred in 1980.

In 1980 the Bureau of LaborStatistics reported 1.4 millionemployees in "computer occupa-tions" in the United States, andestimated that the number wouldgrow to 2.1 million by 1990. Bythis date, after 15 years of continu-ing expansion in academic comput-ing programs, about 25,000bachelor's degrees, 5,000 master'sdegrees, and 250 doctoral degreesin computer science were beingawarded annually in the UnitedStates. The estimated annualdemand for these categories was50,000, 30,000 and 1000, respec-tively, and a significant concernarose about whether this demandcould be met in the coming years.

The supply shortfall was beingcovered by people educated inrelated disciplines who acquired, invarious ways, the skills necessary foran IT career. However, as techno-logical change continued to acceler-ate, it became evident that a higherpercentage of workers in the fieldwould need an education thatfocused on computing and infor-mation technology. A closer

I t) 117

examination of academic pro-grams in 1980 showed that,although the numbers of bachelor'sand master's students and degreeseach year were expanding rapidly,the number of doctorates incomputer science had leveled offat 250. Significant numbers ofgraduate students were leaving thedoctoral programs to go directlyinto industry. Less than half of thenew doctorates were acceptingfaculty positions, which appearedinsufficient to sustain the academicprograms needed to meet thedemands of undergraduate andmaster's degree programs. Thechairs of essentially all of thePh.D.-granting computer sciencedepartments in the United Statesand Canada gathered together andidentified three immediate needs:

a computational infrastruc-ture for faculty and graduatestudents comparable to that foundin the best industrial research labs;

funding to support Ph.D.students throughout their studyperiod; and

new faculty positions toaccommodate the burgeoningundergraduate enrollments, whilecontinuing to provide the facultyadequate time (comparable to thatprovided in other scientific andengineering disciplines) to supervisegraduate students and conductresearch.

In large part because of theunanimity of the agreement on theneeded steps, they were takenswiftly. The National ScienceFoundation (NSF) took the lead inproviding computational infra-structure support, which wassignificantly supplemented bygrants from companies. NSF and

118

companies established new pro-grams of doctoral student support(with several companies making thesupport conditional upon thestudent's commitment to pursue anacademic career). Universitiesadded significant numbers offaculty positions to cover thegrowing enrollments. By 1990 theannual Ph.D. production hadquadrupled to about 1,000.

However, the problemspersisted. By the mid-to-late1980s, it was becoming apparentthat the growth in faculty was notbeing accompanied by a commen-surate growth in funding forfaculty research and graduateeducation activities. Also, the callfor full funding for Ph.D. studentshad not been completely answered.As the number of students in-creased, this lack of fundingbecame increasingly apparent.Although the new steps taken wereon track at this time to enable theproduction of 1,000 Ph.D.s peryear, they would prove inadequatefor building an academic enterprisesufficiently large to both continuethe production of Ph.D.s andeducate the growing number ofundergraduate and master'sstudents. For example, a 1983study showed that the computerscience bachelor-degree-to-facultyratio was twice that in electricalengineering and four times ormore the ratio in any other relateddiscipline. In addition, the com-puter science full-time-graduate-student-to-faculty ratio was 50percent higher than that of electri-cal engineering and of all engineer-ing, and three times higher than theratio of any other related discipline.Such workloads made alternatecareers for computer science facultyvery attractive indeed, and discour-

aged students from choosing such afaculty career.

With this history providingperspective, there are early signs ofanother cycle of "eating our seedcorn." The conditions are similar:aggressive recruiting by industry thatis luring high-quality undergraduatesaway from considering graduateschool; doctoral-caliber graduatestudents leaving graduate programsafter completing only a master'sdegree; faculty shying away fromhigh-pressure teaching positions;and burgeoning undergraduateenrollments that are creating largeclass sizes, an inflated faculty-to-student ratio, and overcommittedfaculty." Not surprisingly, there is adownward trend in the number ofcomputer science doctoratesawarded annually during the 1990s(1,074 awarded in 1990-91, 894 in1996-97)." The number of newdoctoral graduates enteringacademia is slightly more than 40percent if postdoctoral andacademic research positions, as wellas faculty positions, are included.This percentage has not beenincreasing, which means that thetotal number entering the teachingfield is lower. Meanwhile, thenumber of faculty positions beingadvertised has skyrocketed. Ad-vertisements in Computing ResearchNews, for example, have doubledover the past two years.

Other signs of a seed-cornproblem are appearing. Universi-ties have already experienced severeshortages in several faculty areas,

including networking, databases,and software engineering, andfaculty recruiting is becoming muchmore difficult.. There are fewerqualified applicants, positions aretaking longer to fill, and some aregoing unfilled. The general attitudeof the computing research com-munity at the moment is to moni-tor the situation closely, until thedata and qualitative evidence makeit more apparent that a seriousseed-corn problem does exist. Ifthis is determined, then actionssimilar to those taken in the 1980sby government, industry, andacademia working together may bewarranted.

The situation today is, how-ever, different in some respectsfrom 1980." Today, IT facilities inuniversities are more like those inindustry than they were in 1980;and a healthy research program inexperimental computer sciencenow exists in the universities.However, the focus of universityresearch has become much moreshort-term than it used to be,making it less different fromindustrial research; this change hasremoved one incentive for facultyand graduate students to remain inthe universities. High-level ITprofessionals today employedacross a much larger number ofemployers, including many outsidethe IT sector. This makes it moredifficult for industry to worktogether, as companies did in the1980s, to restrain the raiding offaculty and graduate students fromuniversities.

87 See the anecdotal account of this situation in Bronwyn Fryer, "College ComputerScience Enrollment Skyrockets," Cornputerworld, October 22, 1998.

88 CRA Taulbee Surveys.

89 The President's Information Technology Advisory Committee has addressed some ofthese issues in its report to the President. See http://www.ccic.gov/ac/report/.

1 0 119

Chapter 9. Data Issues

What Are theSources ofData on ITWorkers?

Data are available from threeprimary sources: the federalgovernment, professional societies,and the private sector. In a sepa-rate project, the National ScienceFoundation (NSF) funded theCommission on Professionals inScience and Technology (CPST) todevelop an Internet-based Guide toData on Scientists and Engineers."

Federal data. The mostrelevant federal data on informa-tion technology (IT) workers areprovided by NSF's ScienceResources Studies (SRS) division,the National Center for EducationStatistics (NCES; Department ofEducation), and the Bureau ofLabor Statistics (BLS; Departmentof Labor).

Data on the employment ofthose working in the U.S. ontemporary visas are very limited andwere not particularly helpful to thisstudy. The relevant agencies includethe U.S. Department of Justice,Immigration and NaturalizationServices (issuance of Hl-B visas toindividuals) and the U.S. Depart-ment of Labor (employer applica-tions for Hl-B visas)."

NSF has a wealth of data onscientists and engineers." Thelargest amount of worker-relatedNSF data is on doctoral scientists,since this has long been consideredthe principal professional degree forindependent researchers. Data alsoare collected on those who hold abaccalaureate or a master's in scienceor engineering. The data aremaintained at the direction ofCongress to provide an inventoryof human capital for nationalprosperity and security, and a sourceof information for policy formula-tion by other agencies.

9° For further details and hyperlinks to the data sources described in this section, thisguide can be accessed at www.cpst.org.

91 See: www.oig.dol.gov/public/reports/oa/1996/foreign_labor_cert.pdf for a concisereport subtitled, "The System is Broken and Needs to be Fixed."

92 www.nsf.gov/sbe/srs

IL 0 121

NSF's Science ResourcesStudies division oversees the Surveyof Doctorate Recipients (SDR), abiennial survey of a representativesample of those who have receiveddoctorates in the United States in ascience or engineering field.

In addition, NSF oversees theSurvey of Earned Doctorates(SED), which is an annual popula-tion survey of individuals receivingdoctorates that year in science orengineering. SED data are stored ina Doctoral Records File (DRF),from which the sample for theSDR can be pulled for longitudinalfollow-ups and for replenishmentsand retirements every two years.

NSF also collects workforce-related data on individuals withbachelor's and master's degrees inscience and engineering fields, aspart of the biennial NationalSurvey of Recent College Gradu-ates (NSRCG) and the NationalSurvey of College Graduates(NSCG)." Data at all degree levelshave been merged recently into adatabase called SESTAT, which isaccessible from the Web site.

In addition to the surveys ofindividuals described thus far, NSFannually collects institutional-leveldata on enrollments and degreesfrom the baccalaureate and abovein science and engineering at U.S.universities. These data are avail-able in a database called CASPAR,which is accessible from theWeb site.

The National Center forEducation Statistics focuses onenrollments, degrees, facultycounts, and numerous otherstatistics on the status of theeducational enterprise in the UnitedStates." The annual Digest ofEducation Statistics, available on theirWeb site, contains voluminous data.NCES also sponsors periodiclongitudinal surveys that follow upon the work experiences ofsamples of selected cohorts ofhigh school and college students.

The Bureau of LaborStatistics provides employmentdata and outlooks based primarilyon data collected from employersand extrapolated into the futureBLS data are not usually sorted bydegree level or degree field, sincethe reporting burden on employerswould be somewhat onerous."BLS is planning a national jobvacancy survey in 1999 for the firsttime in this country.

Professional society data.The Computing Research Associa-tion (CRA) conducts an annualsurvey of Ph.D.-granting depart-ments in computer science andcomputer engineering to ascertainenrollments, degrees, faculty sizeand salaries, as well as the place-ment of doctoral graduates asreported by departments.96

The Engineering WorkforceCommission of the AmericanAssociation of EngineeringSocieties collects data on enroll-

93 The National Survey of College Graduates is conducted by the Census Bureau for theNational Science Foundation as a baseline survey every ten years; the participants arethen followed up biennially during that ten-year period.

94 http://nces.ed.gov

http://stats.bls.gov

www.cra.org/statistics

122 1 q

ments, degrees, and salaries ofengineers at all degree levels."

The American Society forEngineering Education surveyedthe employment experiences ofrecent engineering doctorates aspart of an NSF-funded CPSTproject on outcomes of doctor-ates." CPST also coordinated acomparable survey of computerscience doctorates." The NationalAssociation of Colleges andEmployers (formerly the CollegePlacement Council) collects datafrom college and university place-ment/career centers covering joband salary offers made to newgraduates, primarily at the bacca-laureate level.'"

The Council of GraduateSchools conducts an annual Surveyof Graduate Enrollments, whichincludes data on applications tograduate schools, as well as enroll-ments and degrees granted.10'

The International Associationof Managerial Education collectsdata on information technologygraduates from business schools.

Private sector data. Privatedata sources include Abbott, Langer& Associates,'" Computerworld,'"

and Datamation.m4 The first is aprofessional survey researchorganization that specializes in salarysurveys of employers; the lattertwo organizations publish IT-related magazines and periodicallysurvey their individual readers.These salary surveys are particularlyinteresting for a project such as thisbecause they reflect the diversity,inconsistencies, and change amongjob position titles in organizationsthat employ IT workers.

Other data sources. TheInformation Technology Associa-tion of America (ITAA) is a tradeassociation that conducts an annualcompensation survey with the helpof William M. Mercer, Inc.'" Aswith the private sector surveys, theITAA survey includes paragraph-length descriptions of an extensiveinventory of more than 80 IT andrelated job descriptions.

As part of the CooperativeInstitutional Research Program ofthe Higher Education ResearchInstitute (HERI) at the University ofCalifornia, Los Angeles, data havebeen collected annually since 1966from freshmen in the U.S. on theirintended majors and career plans 1"The American Council on Educa-tion is a sponsor. Periodic, longitu-

97 www.aaes.org/ewc

www.asee.org/data

www.nextwave.org/survey. The survey results are posted, with related articles, in theCommunications of the ACM (September 1997 and November 1998).

100 www.jobweb.org

101 www.cgsnet.org/vcr

182 www.abbott-langer.com

183 www.computerworld.com

104 www.datamation.com

188 Results are available at www.itaa.org

106 www.gseis.ucla.edu/heri/heri.html

5 123

dinal follow-ups of selected cohortshave been performed as researchfunding has allowed.

What Are theLimitations ofExisting Dataon the ITWorkforce?

Numerous and serious prob-lems with the supply and demanddata make it difficult to establish asound basis for making policydecisions. Much of the datacoming from non-federal sourcesexhibit the following problems:

The domain of workers studiedwas too narrow. A number of datastudiessome of which are welldoneare available on specificregions (IT workers in Georgia,'"software workers in WashingtonState,'" etc.) or specific categoriesof workers (National SoftwareAlliance study,'" U.S. InternationalTrade Commission study of soft-ware and service workers, etc.).Unfortunately, it is not clear that thegeographical regions or categories ofworkers covered in these studiesprovide results that are representativeof the national IT worker situation.

There were methodologicalproblems with the gathering or analysisof data. A number of studies areflawed because the sample size orthe response rate was too small,

questions were framed in ambigu-ous or misleading ways, or finalresults were calculated from rawdata in a questionable fashion. TheU.S. General Accounting Office hasrightly criticized the low responserates of the ITAA studies, forexample. This does not mean,however, that the conclusions ITAAreached are false; it means only thatthey cannot be assumed to be true.The statistical results do not have thereliability or weight of evidence thatone could draw from a survey withmuch higher response rates.

Considering both domain andmethodology, data collected andprocessed by federal organizationssuch as BLS, NSF, and the Depart-ment of Education typically havethe highest reliability. However,even these data sets have seriousproblems from the standpoint ofbeing useful to decision-makersinterested in IT worker issues.

The data are not timely. It takestime to collect high-quality surveydata, to achieve high response rates,and to clean and analyze dataproperly. Also, because of theexpense involved, some surveys areonly conducted every other year. Asa result, in the case of NSFworkforce data, the most recentdata available in late 1998 werefrom 1995. In a field that ischanging as rapidly as informationtechnology, data that are three yearsold have limited value. Data that areat most one year old are needed in

"7 Ephraim R. McLean and Scott L. Schneberger, ICAPP Strategic Industry NeedsAssessment: Information Technology, Board of Regents of the University System ofGeorgia, December 1997.

109 See: http://www.wsa1.org/gotwork/wasoft.htm

109 National Software Alliance, "Software Workers for the New Millennium: GlobalCompetitiveness Hangs in the Balance," Arlington, VA, January 1998.

124 1 6

order to understand the currentstatus of this rapidly paced ITindustry. While the delays are in partunderstandable, ways need to befound to speed up data collectionand analysis.

Worker categories have not beenupdated sufficiently to reflect the currentstate of IT work. Survey researchersand policymakers must make atrade-off between updating surveysto be more responsive to today'smarketplace versus maintainingconsistency from survey to surveyso that results can be comparedhistorically. The Standard Occupa-tional Classification (SOC), which isused by all federal agencies, hadnot been changed from 1980 until1998. Older occupational catego-ries have been preserved over time.

Even apparently good occupa-tional categories, such as computerprogrammers, are too broad and illdefined to allow poligmakers to under-stand kg issues well. Different kindsof programmers, for example, arenot necessarily interchangeablebecause their work involvesdifferent skill sets. Companies havehad only modest success at retrain-ing COBOL programmers to beeffective Java programmersbecause of the wide differences inthe programming methodology ofthe two languages. It is not clearthat there is much chance ofconvincing the federal data collec-tors to break their categories intosmaller ones to resolve this issue;but if the professional communitycollects data, it should keep thisissue in mind.

The levels of data aggregation,in either what data were collected or whatdata were reported publicly, are oftenproblematic. For example, computerscience is sometimes combined

with mathematics, other timescombined with the physicalsciences, and still other times withengineering. In most cases, basedon the purposes for which the datawere originally collected, this isunderstandable. But it posed achallenge for researching this study.

Data have generally beencollected hy the federal government only inrelation to earlier issues of polig concern.These data do not provide an adequatebasis for making current polio; decisions.For example, non-degree trainingsuch as certificate programs, shortcourses, and corporate universitiesare an increasingly important partof the supply system for ITworkers. Because these trainingvenues have not been important tonational policy in the past, federalstatistics about them do not exist.

Even in cases where there hadbeen previous polig issues, data collectionis sometimes inadequate. One exampleconcerns H-1B visas. There are nostatistics available on how manytemporary workers are actually inthe United States on H-1B visas,only on the number of approvedlabor certificates and the numberof visas that were awarded. Norare data available on how many ofthe visa-holders are doing IT work.

Data on the demand for ITworkers are especially poor. Compa-nies are reluctant to report dataabout their need for workers ordisclose the effects of a shortageon their business. This is consid-ered proprietary information.Companies worry that if it weremade public, it could harm theirpublic reputation or be exploitedby their competitors.

The mismatch between supplyand demand data, as collected by different

7 125

federal agencies, is problematic. Onemajor frustration in researchingthis report was trying to matchdata broken out by educationaldegree fields (computer science,information systems, softwareengineering, etc.) and levels(bachelor's, master's, etc.), withdata based on occupationalclassifications or job positiondescriptions (programmer,systems analyst, computer scientist,for example). The study group'sattempts to cross-tabulate byeducation and occupation hadlimited success. NSF was the onlysource that had useful occupa-tional data broken out by educa-tional background. BLS data areoccupationally oriented, while

126

NCES data are educationallyoriented.

Using onlin4 publicly availabledatabases is too challenging. TheInternet can provide access tomicrodata or other "raw" data.Researchers can analyze datacollected to answer one set ofquestions in an attempt to answeranother new set of questions.However, most current systems arenot yet very user-friendly in termsof search engines, time required toprocess requests, variable namingconventions, table or reportgeneration, etc. In most cases,users have to take considerabletime to learn the systems andcannot easily retrieve data.

Chapter 10: Recommendations

As noted in the ExecutiveSummary, these recommendationsare offered on the basis of thestudy group's collected experienceand study of the supply of infor-mation technology (IT) workers. Itis hoped that at a minimum theywill spark productive discussion.They are organized by intendedaudience. A summary of therecommendations can be found inbox 10-1 following this chapter.

1. FEDERALAND STATEGOVERNMENTS

1.1. Data-collection prac-tices must be improved.

The federal government is byfar the most reliable and compre-hensive source of data on thesupply of and demand for infor-mation technology (IT) workers,and it continues to be in the bestposition to collect these data.However, current data are not asuseful as they could be because theyare not timely, they are not suffi-ciently comprehensive, they do notemploy adequate categorizations in

many cases, and there is insufficientcomparability among data collectedby different agencies. There needs tobe better coordination amongfederal agencies when collectingdata about supply and demand, anddata should be collected in a waythat permits meaningful comparisonacross data setsno matter whichfederal agency collects the data.

1.2. A new system fortracking the demand for andsupply of IT workers should becreated.

The Bureau of Labor Statistics(BLS), the National Science Founda-tion (NSF), and the National Centerfor Education Statistics (NCES)have done a good job using theirtraditional approaches, but thosemeasures are not sufficient to fullyinform decision-makers. It is thefederal government's responsibilityto bring together representativesfrom industry and academia tocreate a new system that will bettermeet the needs of policymakers forsupply and demand data.

1.3. Data collected must becomprehensive to be useful forpolicy deliberations.

127

The data need to cover allparts of the United States. This hasnot generally been a problem infederal data, but is sometimes aproblem with data from othersources. The regional studiesreviewed by the study group appearto be useful in their geographicaldomain, but they do not adequatelyaccount for interstate movementsthat could be extensive. On thesupply side, data need to be col-lected not only about formal degreeprograms in IT-related disciplinessuch as computer science andinformation systems, but also aboutother disciplines that train many ITworkers, such as business, engineer-ing, and science. Data should alsobe collected about non-degreeprograms, including those offeredby for-profit and other non-traditional organizations. The dataneed to cover all U.S. demand forIT workers, not only that ofcompanies in the IT sector. Com-panies outside the IT sector and allsmall companies may be under-represented in existing demandstatistics. The demand data shouldbe broken down into categories thatare useful to suppliers for settingtheir curricula and enrollment sizes.

1.4. Tbe Standard Occupa-tional Classification (SOC)categories for informationtechnology occupations need to bereviewed and refreshed on aregular basis.

These categories currently areof marginal use in connection withmany IT policy issues. Although thetask is onerous and there arepotential risks to the longitudinalcontinuity of data, classifications ofIT workers should be reviewedevery couple of years and refreshedas needed to reflect the current stateof the field. Classifications must be

128

based not on job titles, but on thetasks performed and the skill setsrequired to do these tasks. Someimportant sources of supply arecurrently not tracked well (e.g.,information systems bachelor'sdegrees and continuing educationenrollments).

1.5. Federal and stategovernments, with industryinvolvemen4 should improve IT-related training mecbanisms attbe K-12 educational levels andkeep them current. Counseling,teacher training, curricula, andcomputing facilities all needimprovement relative to infor-mation technology.

Students are being uncon-sciously eliminated from thecandidate pool of IT workers bythe knowledge and attitudes theyacquire in their K-12 years. Manystudents do not learn the basic skillsof reasoning, mathematics, andcommunication that provide thefoundation for higher education orentry-level jobs in IT work. Anumber of students who havesome or all of these basic skills donot consider IT careers because theybelieve (often incorrectly) theycannot succeed in this area, orbecause they have misperceptionsabout the opportunities or thenature of the work. It is in thenation's interest to have as large apotential pool of IT workers aspossible. In any case, many of theskills learned in preparing for apossible career in informationtechnology are skills that will servepeople well in other careers, andmore generally as citizens in thetwenty-first century.

This is not a task solely forgovernmentat any levelto doalone. There is a large role that

r\4:-U

industry can and should play in localcommunities in providing new andused equipment, enriching curricula,supplying adjunct instructors, andtraining teachers. In addition todirect financial investment in thepublic schools, governments cansponsor programs to teach K-12students about IT opportunities,develop model programs, dissemi-nate information about bestpractices, stimulate academic-industry partnerships, and make thepublic aware of the importance ofimproved education. A great dealhas already been accomplished inthis area, particularly in providingadequate computing facilities to theschools. It appears that the schoolshave made more progress inobtaining facilities than they have inthe other three areas: adequatecounseling, teacher training, andcurriculum development.

1.6. Tbe federal govern-men4 and especially state gov-ernments, should help tostrengthen traditional highereducational programs in IT-related areas.

The higher educational systemis a tremendous national resourceand remains the foundation forproviding IT-workers. It is basicallyhealthy, but needs some augmenta-tion in the IT areas. The governmentcan encourage colleges and universi-ties to expand and create degreeand non-degree options at all levelsthat better prepare students to meetindustry needs. These programs cansimultaneously provide up-to-datebasic skills training, while beingmore attentive to the workforceobjectives of industry.

The need seems to be greatestat the associate's and master'sdegree levels, which are more

career-oriented than the baccalaure-ate and the doctorate. Fewer thanone-third of the communitycolleges are offering associatedegree programs in IT-related areas.Many universities could offerprofessionally oriented master'sprograms that better meet theneeds of local industry for mid- tohigh-level IT workers. Govern-ments can have a profound influ-ence on the programs offered bythe formal educational system,using the power of leadership byelected officials, incentives added tocurrent funding programs tocolleges and universities, newfunding programs, statistics aboutthe need of programs, and devel-opment and communication forbest practices.

1.7. Government can helpfaculty and educational staffadapt to the new demand for IT-trained students.

There is a long-standingtension, not only in IT-relatedcurricula, between basic educationaland career goals. U.S. colleges anduniversities are the envy of theworld, which indicates that manyaspects of the curriculum arealready balanced appropriately, butthere is room for improvement.For example, in doctoral programsmany faculty believe that their goalis to produce clones of them-selvesgraduates who have thesame skills and aspirations as theydo. These programs tend toproduce too many people who aretrained for careers in researchuniversities, and not enough whoare trained to enter industry or toteach in the teaching-orientedinstitutions that train the majority ofundergraduates who enter the ITfield. At the associate degree level,the faculty need to review whether

1 1 129

vendor-specific curricula aremeeting their basic educationalneeds.

1.8. Government shouldhelp to attract more students intograduate programs in IT-relateddisciplines.

Although enrollments haveincreased slightly in the past severalyears, universities are still havingdifficulty recruiting studentsespecially those who are U.S.citizensto attend graduate schoolin IT-related programs. They alsoare experiencing increasing diffi-culty in retaining their graduatestudents long enough to completetheir degrees, especially the Ph.D.degree. The percentage of U.S.students in these graduate pro-grams has recently fallen to lessthan fifty percent. Governmentshould provide new fellowshipsand traineeships to encouragegraduate study in these disciplines,especially for U.S. citizens. It hasbeen the general practice to basefellowship support in computingdisciplines (as in other science andengineering disciplines) primarily onintellectual and scholarly merit.There are sound reasons for thispractice, but other factors, includingthe intention to enter critical areas(such as teaching) and geographicallocation, should also be considered.The National Research ServiceAwards offered by the NationalInstitutes of Health might be onemodel to consider for informationtechnology.

1.9. Government shouldhelp faculty and staff cope withthe greatly increased demand inthe IT area.

Many university faculty in IT-related disciplines are feeling heavily

130 9

overworked, which is affecting theirability to carry out their principalduties of teaching and research, andalso causing some faculty to leavethe profession. Contributing factorsinclude heavy undergraduateteaching loads (compared withother science and engineeringdisciplines) due to lack of faculty,large classes, heavy committee workassignments to help out the univer-sity central administration and otheruniversity departments with theirexpanding IT needs, lack ofadequate support staff to help withteaching laboratories, increasedpressure from university centraladministrations to bring in externalfunding, and the need to be con-stantly writing funding proposalsbecause of the small, short-termgrants given by federal agencies.Federal and state governmentscould ease this situation by provid-ing funding for support staff, bothto the faculty themselves and totheir departments, and by havinglarger individual grants that coverlonger periods of time.

1.10. The federal govern-men4 and NSF in particular,must be vigilant and prevent aseed-corn problem in IT-relateddisciplines.

Many members of thecomputing community believe thatIT graduate students and facultyare being attracted to jobs inindustry in increasing numbers, andthere is worry that there will not beadequate teacher-scholars left totrain the next generation of ITworkers (the so-called "seed-corn"problem). There are preliminarysigns of a seed-corn problem.This is similar to the one computerscience experienced around 1980,when there was a flight fromacademic to industrial careers,

leaving the community wonderingwho would teach the next genera-tion of students. If it becomesclearer that there is a seed-cornproblem today, it will be up togovernment, industry, andacademia to work together to fixthe problem. NSF is in the bestposition to assess the situation andto lead solution efforts, if needed.The key to preventing the problemis to make graduate study and afaculty career attractive to well-qualified people. This meansmaking the work interesting andthe work environment favorable.In 1980 this involved bettercomputing facilities for research,more federal funding for bothresearch and graduate studentsupport (especially at the schools atgreatest riskthose with mid-levelrankings), an emphasis on experi-mental research, and voluntaryrestraint against raiding by themajor industrial employers ofcomputer science doctorates.

The situation today is similar,but not identical to the situationtwenty years ago. There is not asgreat a need today for bettercomputing facilities for research orfor more experimental research,but there is a need for moreadequate funding for long-termresearch and the environment tosupport it. A voluntary restraintprogram in industry may be harderto achieve this time, however,because the number of companieshiring doctorates in IT-relatedfields is much greater today thantwenty years ago and the hiringcompanies are scattered acrossmany industries, rather than being

heavily concentrated in the com-puter manufacturing sector.

1.11. Federal and stategovernments must enhance theresearch climate in the universi-ties.

A strong research milieu isessential to producing the nextgeneration of creators of new andinnovative ideas in informationtechnology. By making universitiesa more attractive place for bothfaculty and graduate students, thelikelihood of a seed-corn problemis diminished. Healthy researchprograms also have a strong impacton undergraduate education, both inthe classroom and through involve-ment in the research. The report ofthe President's Information Tech-nology Advisory Committee (thePITAC Report) provides a set ofrecommendations for enhancingresearch in the nation's universities."°This study is in full agreement withthose recommendations, which arenot repeated here.

1.12 Federal and stategovernments should activelyencourage universities andindustry to form a variety ofpartnerships to train the ITworkforce.

Much of the work that needsto be done to train the IT work-force in the future should occurthrough partnerships between theacademic and industrial sectors:industrial advice to universities ontheir curriculum; industrial employ-ees teaching part-time; industryproviding equipment and support

"Information Technology Research: Investing in Our Future," President's InformationTechnology Advisory Committee Report to the President, February 24, 1999 (http://www.hpcc.gov/ac/).

I 131

assistance; university programsteaching additional career-orientedmaterial; operating specialized,company-specific programs; andassisting companies in establishinginternal educational programs.Government organizations canfacilitate these kinds of collabora-tion.

1.13. Tbe governmentshould encourage the develop-ment of programs in academiaand industry that attractunderrepresented groups to ITcareers.

If underrepresentedgroupsincluding women, NativeAmericans, Hispanics, AfricanAmericans, and perhaps olderworkerswere entering IToccupations in proportion to theirnumbers in the American popula-tion, the supply of IT workers inthe United States would be muchmore adequate. This study doesnot take a position either way onaffirmative action programs, but itis clear (from the programs in the1980s to encourage women toenter the computing field) that suchprograms can be successful inincreasing the supply of qualifiedIT workers.

1.14. Special efforts shouldbe made to utilize the skills ofolder workers.

Older workers are a potentialsource of supply of IT workersthat could and should be tapped toa greater extent than it currently is,although no one seems to havegood estimates of the number ofsuch workers. (The study grouplooks forward to the results of thestudy just now being undertaken bythe NRC with support from NSF,which will include an examination

132 1 2

of this issue.) No doubt manyolder workers have well-developedorganizational and communicationskills, which means that retrainingcould be completed quickly forthese older workersat least thosewith some technical backgroundto enter IT occupations. Govern-ment retraining programs toprovide these skills to olderworkers may be cost effective.Such programs may have a particu-larly salutary effect in lessening theburden on individuals and commu-nities created by lost jobs in thedownsizing of the defense industry.These programs might counterbal-ance a perception that an IT careerhas a short duration and is only forthe young.

2. HIGHEREDUCATION

2.1. Colleges and universi-ties should keep their focus onproviding strong basic educa-tion.

In the face of the rapidlychanging world and even morerapidly changing IT field, it mayseem impossible or irrelevant toprovide a basic education that willendure. There is no question that itis difficult and that the nature of abasic education must evolve. Thecollected experience of the studygroup and observations of others,however, strongly indicate that thebest preparation is still a basiceducation that provides its graduateswith a solid foundation for life-longlearning in multiple disciplines andskills.

2.2. Universities mustrecognize that there is a funda-mental IT-related shift occurring

in the economy and in mostprofessions, and tbat tbey mustreallocate resources for betterand more extensive training intbis area.

It is not surprising thatuniversity central administrations arechary to support departmentalrequests for additional resources.Resources are scarce in highereducation, and reallocations tie upresources for long periods of timein bricks-and-mortar or capitalequipment that need to be main-tained, or commitments to thesalaries of tenured faculty that maylast thirty years. Nonetheless,universities must recognize thatinterest in IT-related disciplines isnot a fad, but will continue to growwell into the next millennium asinformation technology continuesto grow in prominence in oursociety. Some of that growth isoccurring right now, as the doublingin newly declared computer sciencemajors over the past two yearsattests. Not all the growth ininstitutional programs needs to beplaced in the traditional IT disci-plines, such as computer science andengineering. Some growth mightoccur in new interdisciplinaryprograms and departments, or indistributing information technologythroughout the university, intovarious existing faculties anddisciplines. The specific approachwill need to be made on an indi-vidual basis, depending on localcircumstances. But wherever in theacademy this instruction is located,the colleges and universities shouldplan on significant, continuingallocation and reallocation of

resources into IT-related facultyand curricula.

2.3. Higher educationshould provide faculty supportto revise tbeir curricula toprovide more and better patbsin tbe training of IT workers, aswell as to provide better ITeducation for all students.

Most faculty have been doingtheir best to keep the curriculacurrent, but the rapid rate ofchange and the press of otherresponsibilities make this taskincreasingly difficult. Someamount of information technologyshould be made a part of all basiceducation programs. Knowledgeof the basic concepts of informa-tion technology is now almost asimportant as knowledge of thefundamentals of mathematics. Thisissue is not addressed in this report,but readers are referred to therecently released National ResearchCouncil report, Being Fluent WithInformation Technology." The highereducational system needs toprovide a variety of training pathsto IT careers. The system will besuccessful to the extent that it canaccommodate students who havedifferent interests, educational andvocational objectives, levels oftechnical ability and preparedness,and levels of self-confidence in theirpath to an IT career. By making thefield more open to students withdifferent objectives, backgrounds,and confidence levels, the field ismore likely to attract students of allkinds.

2.4. Faculty in IT-relateddisciplines need to retbink tbeir

National Research Council, Being Fluent With Information Technology, Washington, DC:National Academy Press, 1999 (www2.nas.edu/cstbweb).

133

introductory undergraduatecourses.

One barrier to attractingstudents to major in IT-relateddisciplines is the way in which someIT-related departments, especiallycomputer science and computerengineering departments, introducethe subject to students. Facultyunderstandably do not want towater down their programs inorder to attract students, but theysometimes mistakenly equate thispositive desire for high standardswith a need to make their entry-levelcourses overly rigorousforexample, with substantial mathemat-ics and science requirements. Theseact as unnecessary barriers to entryfor some students who could besuccessful in the IT field. Certainjobs, such as those in scientificprogramming, do require substantialtraining in calculus and science, butother kinds of IT jobs wouldbenefit more from having theirworkers receive other kinds oftraining in problem-solving. Someof the requirements in IT-relatedundergraduate programs may be aholdover from earlier days, when ascientific programming career wasthe norm for college graduates incomputer science. In any event,computer science and computerengineering entry-level coursessometimes have unusually highattrition levels, which is not in thelong-term interest of students,employers, or the departmentsthemselves.

Another problem of theintroductory course in most com-puter science departments is that it isgenerally devoted entirely tolearning the principles of com-puter programming. Such a courseis narrowly focused and technical.It does not give students a good

134 1 '

introduction to the nature of thecomputing field. For example, itdoes not show the engineering,mathematical, or theoretical aspectsof the subject, how computersrelate to business and managementor to psychology, or how infor-mation technology can be appliedto a wide range of problems inmany fields of endeavor. Many ofthe scientific and engineeringdisciplines have developed intro-ductory courses that survey thebasic concepts and methods oftheir field, and the study groupbelieves this would be appropriatefor the IT-related disciplines aswell. Students need to get, at anearlier stage in their education, asense of what this subject is allabout in order to make better-informed decisions about theirstudies and their career. The studygroup heard anecdotes aboutstudents who had indicated thatthey did not want to major incomputer science because theyperceived that it led to a careerworking with machines instead ofwith people. Some did not want tospend all their time involved in thetechnical problem-solving arcanaof programming, but insteadwanted to be working on real-world applications. Some studentsdid not like the uncollaborativeattitude they had seen amongprogrammers. Courses that bettershow the human dimension ofcomputing and its applicabilitywould help to address theseconcerns.

2.5. IT-related departmentsshould increase ratber tbanrestrict access to their coursesand programs.

Because of the strong needfor substantive education in com-puter science for other curricula (to

prepare for IT or IT-enabled jobs),this increased demand may requireboth new funding from the univer-sity and reallocation of departmen-tal funds. There is substantialenrollment pressure on faculty in IT-related departments today. At manyschools the size of the undergradu-ate enrollment is increasing muchfaster than the size of the faculty orthe graduate student population.This leads to large lecture classes,excessive demand on computerfacilities, and student-teachercontacts that are stretched thin.Departments are limited in thenumber of students they can handlewithout additional resources, andwhen they become stressed byenrollments, they understandablytake action to relieve the stress.

All of these strategies limitstudent opportunities to train for ITcareers. It also means that thosestudents who are admitted into thecourses have a certain intellectualprofile, which may be good forsome IT occupations and lessappropriate for others.

2.6. IT-related departmentsshould develop graduate-levelprograms.

People with strong back-grounds in mathematics, science,engineering, or business haveknowledge and skills that willallow them to retrain quickly forIT occupations, such as managersor creators of new technology.There have been surpluses ofmathematicians, physicists, biolo-gists, and engineers in recent years.Many of these scientists andengineers are of very high intellec-tual and technical quality, but are infields that are unable to providegood careers. Technically orientedgraduate programs in computer

science, computer engineering, orinformation science could, withadditional resources, accommodatethe retraining of people with theseskills by repackaging existingcourses into a certificate program.Certificate programs that arefocused rather than general seemto be particularly attractive to bothstudents and employers. Examplesof certificate programs of thiskind are human-computer inter-faces, bioinformatics, and high-performance computing.

2.7. University practicesshould be adjusted in order to bemore supportive of the educationof IT workers.

Better counseling is needed tolet students know about opportuni-ties in the IT field, the nature of thework, and the skills required to besuccessful in the field. This isparticularly wanting in two-yearcolleges and high schools, but maybe deficient in many four-yearcolleges and universities as well.

Existing IT-related depart-ments (typically in computer science,computer engineering, or informa-tion science) can do more toprovide additional paths to ITcareers. This can be achieved, forexample, through more flexibleentry requirements into majors andby adding minors, cross-disciplinaryprograms, and certificates. Theuniversity administration should besupportive of these efforts in theexisting departments.

However, there is a limit tohow much any given departmentcan do and still remain manageablein size and focused in mission. Withgood planning, there can be value inhaving multiple IT-related depart-ments, each with its own mission, in

?.7135

the same university. This is alreadycommon at large universities. Someuniversities, such as PennsylvaniaState, Berkeley, Rensselaer, andMichigan, have kept their existingdepartments intact and focused ontheir traditional missions, while atthe same time forming newdepartments or schools to addresssome other, often interdisciplinary,aspects of information technology.

2.8. New ways are neededto improve the articulationbetween different levels ofeducational institutions.

This is a long-standingproblem in many fields of highereducation, but the IT situationhighlights it and argues for newapproaches. We have heard manycomments about the difficultiesthat students experience in trying totransfer credits from one programto another as they move throughthe higher educational system inpreparation for an IT career. Thisoccurs at all levelsas a studentmoves from the vocational to thecollege preparatory associatedegree, from the associate to thefinal two years of the baccalaureatedegree, from one IT-related majorat the baccalaureate level to another(management information scienceto computer science, for example),or from the professional master'sdegree to the doctorate. Thesekinds of problems are bestaddressed by having institutions ata variety of levels work together toset standards and procedures.Because procedures already existfor this kind of system-widediscussion and coordination in statehigher educational systems, theyneed to take a leadership positionon these issues and provideexamples of best practices that can

136

be followed by the entire U.S.higher educational system.

3. INDUSTRY

3.1. Industry should makedata available regarding tbedemand for IT workers.

In the course of this study, itwas almost impossible to obtaincurrent data about industry demandfor IT workers. Federal data wereout of date and had problems ofclassification, and most industry datawere both firm-specific andproprietary. Without cooperationfrom industry in supplying informa-tion about the kind of personnelthey need, it is impossible forsuppliers to plan accordingly. Itwould be useful to have data thatmeet all of the requirements set outabove in the section on recommen-dations to federal and state govern-ments about data-collecting prac-tices.

3.2. Companies shouldinvest more in entry-level train-ing and the retraining of existingpersonnel.

Some medium and large-sizecompanies already have extensivetraining programs for their ITworkers. However, some small andmid-size companies do not investsubstantially in training their work-ers, believing that they cannot affordthe cost or time of training, or outof fear that they will be trainingemployees who will eventually go towork for their competitors. Somecompanies that do invest do notinvest optimally by planning fortraining, rather than responding toindividual cases in an ad hoc way. Inthis fast-paced field, where product

lines turn over in a matter of twoor three years and where knowledgebecomes obsolete in about the samelength of time, it is critical that theworkforce retain outstandingtechnical and communication skills,as well as industry and businessknowledge. Companies shouldvalue the training provided bycolleges and universities, but shouldnot expect these schools to haveproduced the perfect employee.Even new college graduates fromthe strongest universities will nothave all the knowledge they need toexcel in the workplace. Given themany options in terms of subject,supplier, cost, and length of trainingavailable in the marketplace, there islikely to be a training programavailable to suit any employer need.Companies could broaden thecandidate pool by their willingnessto carry out more entry-leveltraining, rather than expecting newemployees to enter the companywith all the requisite skills in place.Industry should also encourage itscurrent employees to increase theirskills by providing flexible workschedules, tuition assistance, andopportunities for distance or otherkinds of learning experiences.

One of the impediments tomore training for workers is thatcompanies fear their training fundswill be counterproductive if theytrain an employee who leaves thecompany to work for a competitor.A few attempts have been made bya collection of employers in ageographical region interested in thesame labor pool to band togetherand jointly offer training programsfor their collective group ofemployees. Sematech is an examplein the semiconductor area. Compa-nies participating in these arrange-ments have less fear of this migra-tion problem because they believe

they will profit as often as they loseunder this system, and that all theemployers will have access to amore highly qualified labor pool.

3.3. Companies outside ofthe IT sector need to recognize thatinformation technology maybecome a core competency forthem.

Some members of the studygroup believe that the Y2K prob-lem is as serious as it is todaybecause many companies outside ofthe IT sector reduced their ITworkforce a few years ago, at thetime when many corporationsdownsized. Companies need torecognize that information technol-ogy is essential to their livelihood,even if their main business is not ITproducts or services. These compa-nies need to keep the quantity andquality of their computer operationsat a high level in order to remaincompetitive. Given the competitionfor good IT workers and thestriking variations in capabilities andproductivity among these workers,managers will have to work tomaketheir company a place that isperceived as an attractive place towork by IT workers.

3.4. Industry should workclosely with the higher educationsystem to improve education forIT workers.

The higher education system isone of this nation's great strengths,and it needs industry's support toremain vital. In recent years, somecompanies have de-emphasizedtheir reliance on the higher educa-tion system and formed new kindsof training programs, either in-house or purchased from for-profitvendors. Companies gain fromhaving programs that customize the

1 9 137

curriculum to their particular needs,but the higher education system alsoplays an invaluable role in theproduction of a trained workforce.Individual companies, as well asindustry groups, should workclosely with colleges and universitiesto develop a more comprehensivecurriculumone that teaches thefoundational skills that are thehallmark of the formal educationalsystem as well as offering current,practical knowledge. Industryshould support the formal univer-sity program by sponsoring scholar-ships and internships, providingaccess to facilities not available at theuniversity, helping to outfit universitylaboratories, helping the schoolswith their staff development, havingcompany professionals give lecturesand teach courses at the university,and entering into various kinds ofprogrammatic partnerships with theschools.

3.5. Industry should nottake actions that in the long runharm the supply system.

Companies should restrainthemselves in hiring students beforethey have completed their degrees,or at least support them to finish thedegree as part-time students if theydo hire them. This is a way inwhich companies can both be goodcitizens and help themselves in thelong run. Students who are able tocomplete their education havelearned a set of foundational skillsthat will make them more effectiveemployees. They will be less likelyto become obsolete quickly, and willbe better able to learn new skills andknowledge as needed.

Companies should alsorestrain themselves from hiringaway faculty. Unless there is anadequate number of high-quality

138 I r-, 01 0

faculty and enough students complet-ing their doctorates to renew thefaculty ranks, the IT community willagain have a seed-corn problem andthe universities will become progres-sively less able to train students whowill meet industry needs. Industryneeds to sacrifice some of the short-term payoffs of hiring faculty andstudents, seeing them instead as long-term investments in their own futureand in the strength of the national ITworkforce.

3.6. Companies should hirefor diversity and tap aggressivelyinto groups that are under-represented in the IT profession.

This strategy will increase thecandidate pool. It will also adddiversity to the workforce that can bebeneficial to the company in under-standing customers, understandingcomputer applications, and obtaininga diversity of views in the workforce.

4. PROFESSIONALSOCIETIES

We believe that there is a greatopportunity for the professionalsocieties to provide additional, muchneeded services to the professionalcommunity. This may serve not onlytheir members, but the societiesthemselves by building up theirmembership rosters.

4.1. The professional societ-ies should provide greater assis-tance in the retraining andcontinuing education of ITprofessionals.

The formal education an ITworker receives in college (andgraduate school) provides adequatebackground preparation for the job

for perhaps only two to three years.It is increasingly apparent that ITworkers need to be engaged in acontinuous process of education,and that it may be necessary toundertake major retraining moreoften. The professional societieshave given some attention to theseissues in their continuing educationprograms. They should take aneven more active role in assuringthat their members have goodopportunities for continuousretraining.

4.2. Tbe professionalsocieties should take a moreproactive role in tbe certifica-tion of IT professionals.

As jobs and the requisite skillsets change rapidly in the IT field, itis difficult for any worker to knowwhich of his or her skills is ad-equate and which need to bebrushed up. As the training oppor-tunities multiply, workers are at aloss to know which ones areappropriate for them. If theprofessional societies can providecertification standards, workers canassess their own skills and seek outappropriate training, and employerscan determine if applicants andemployees are well qualified tocarry out specific IT jobs.

4.3. Professional societiesshould continue to play a strongrole in curriculum development.

For more than thirty years, thecomputing professional societieshave played an important role indeveloping and publicizing modelcurricula in IT-related fields. This isone of the most important tasksthey can carry out in the future. Indeveloping these curricula, theprofessional societies should listento the needs and expectations of

employers as well as of suppliers.Given the rapid pace at whichinformation technology is beingdeveloped, and at which knowledgebecomes obsolete, it is important todevelop model curricula that aredesigned for change. What isneeded are curricula that canaccommodate small modificationson a regular basis, rather thancurricula that are expected to standwithout revision for many years.One place where the professionalsocieties can be especially helpful isin developing model curricula forassociate degree programs, and inparticular in setting minimal stan-dards and procedures for a two-year college that wants to adopt acurriculum developed by one ofthe technology vendors. Attentionshould also be given to curricula forprofessional master's programs andto post-baccalaureate certificateprograms in specific technical areassuch as bioinformatics or networkadministration.

4.4. Tbe professionalsocieties should take considerablygreater interest in non-degreeprograms tbat train IT profes-sionals.

Most model curricula andaccreditation of IT training offeredby the professional societies aredirected at formal degree programsoffered by traditional colleges anduniversities. The fastest growingsegment of IT training, however, isthat of the non-degree programsoffered by two- and four-yearcolleges and universities, as well asby for-profit training companies,individual consultants, and thecompanies themselves. Theseinclude short courses, certificateprograms, distance learning pro-grams, and other kinds of training.Because there are currently no

'131 139

guidelines, these are caveat emptorpurchases, both for the studentsthemselves (who are often mem-bers of these professional societ-ies) and for the companies thatoften pay for their workers toattend. The professional societiescould take a major role in settingstandards and accrediting pro-grams. Formal continuing profes-sional education programs of thetype used in K-12 teacher trainingand in the medical and legalprofessions may also be helpful.

4.5. The various IT profes-sional societies should communi-cate, cooperate, and collaboratemore with one another on issuesof worker supply and demand.

As the IT field broadens, thenumber of professional societiesdedicated to information technol-ogy grows. There are at leasttwenty IT-related disciplines taughtin U.S. universities today. Many ofthem have at least one professionalsociety representing them. Manyof these societies are focused on aparticular aspect of informationtechnology. Several societies,seemingly independently of oneanother, were already working onissues related to IT workers andtraining programs when this studybegan. There is no singleoverarching society for the entireIT field, nor it does appear thatone will be formed anytime soon.Absent such a society, it is impor-tant that the existing societiescommunicate with one another andcollaborate, or at least cooperate,on study groups, model curricula,and accreditation. It may even bedesirable to form a loose federa-tion of IT-related societies for thispurpose. The Computing Re-search Association has taken first,

140

preliminary steps in this directionthrough its annual computingleadership summit.

5. INDIVIDUALS

5.1. Workers should recog-nize that tbey must take responsi-bility for remaining individuallycompetitive.

The United States is a free andmostly unregulated market in whichcompanies do not hesitate to lay offeven good workers if they do nothave the currently needed skills. Thismeans that employees must plan tobe flexible and not view as a failurea need to change jobs or return forfurther education in the middle oftheir career. Workers should becareful to choose jobs so that theybroaden and update their skills, andthey should be open to learning.

5.2. Individuals mustcommit themselves to life-longlearning in order to remaintechnically current and com-petitive.

There are many ways inwhich an individual IT worker canobtain this training, such as collegeand university courses, lectures,seminars, and self-study. Theindividual worker should takeadvantage of the opportunitiesoffered by the employer andshould also personally seek outother opportunities. IT workerswho become complacent abouttheir knowledge and skills canbecome obsolescent in as little astwo years. The professionalsocieties can be a good source ofinformation about what knowl-edge and skills are in demand, and

they may even be able to providethem or point to reliable providers.

5.3. Individuals should dotbeir part to see tbat peoplewith appropriate skills enter tbeIT workforce.

People choose an occupationin large part because of personalcontacts they have hadthe strongteacher, the knowledgeable and fairsupervisor, the caring colleague.Individuals can attract people withgood skills to IT careers by encour-aging IT as a career, by serving as amentor to students or less experi-enced employees, by advisingpeople on their training and careerdecisions, and by serving as a goodrole model.

5.4. Individuals shouldhelp to build up tbe IT profes-sion through its professionalorganizations.

The IT profession is muchstronger because of its professionalsocieties, its university trainingprograms, and the public andprivate organizations that look outfor its well-being. Members of theprofession should be good citizensby offering to serve on nationalcommittees, review proposals,volunteer to participate in profes-sional society activities, lecture orteach and the local college, and doother volunteer and paid effortsthat strengthen these professionalorganizations.

133 141

Box 10-1: Recommendations

Federal and State Governments

1.1. Data-collection practices must be improved.

1.2. A new system for tracking the demand for and supply of IT workers shouldbe created.

1.3. Data collected must be comprehensive to be useful for policy deliberations.

1.4. The Standard Occupational Classification (SOC) categories for informationtechnolo6/ occupations need to be reviewed and refreshed on a regular basis.

1.5. Federal and state governments, with industry involvement, should improveIT-related mechanisms at the K-12 educational levels and keep themcurrent. Counseling, teacher training, curricula, and computing facilities allneed improvement relative to information technology.

1.6. The federal government, and especially state governments, should help tostrengthen traditional higher educational programs in IT-related areas.

1.7. Government can help faculty and educational staff adapt to the new demandfor IT-trained students.

1.8. Government should help to attract more students into graduate programs inIT-related disciplines..

1.9. Government should help faculty and staff cope with the greatly increaseddemand in the IT area.

1.10. The federal government, and NSF in particular, must be vigilant and preventa seed-corn problem in IT-related disciplines.

1.11. Federal and state governments must enhance the research climate in theuniversities.

1.12. Federal and state governments should actively encourage universities andindustry to form a variety of partnerships to train the IT workforce.

1.13. The government should encourage the development of programs inacademia and industry that attract underrepresented groups to IT careers.

1.14. Special efforts should be made to utilize the skills of older workers.

Higher Education

2.1. Colleges and universities should keep their focus on providing strong basiceducation.

2.2. Universities must recognize that there is a fundamental IT-related shiftoccurring in the economy and in most professions, and that they mustreallocate resources for better and more extensive training in this area.

2.3. Higher education should provide faculty support to revise their curricula toprovide more and better paths in the training of IT workers, as well as toprovide better IT education for all students.

2.4. Faculty in IT-related disciplines need to rethink their introductory undergraduate courses.

2.5 IT-related departments should increase rather than restrict access to theircourses and programs.

2.6. IT-related departments should develop graduate-level programs.

142 13 -41

2.7. University practices should be adjusted in order to be more supportive ofthe education of IT workers.

2.8. New ways are needed to improve the articulation between different levels ofeducational institutions.

Industry

3.1. Industry should make data available regarding the demand for IT workers.

3.2. Companies should invest more in entry-level training and the retraining ofexisting personnel.

3.3. Companies outside of the IT sector need to recognize that informationtechnology may become a core competency for them.

3.4. Industry should work closely with the higher education system to improveeducation for IT workers.

3.5 Industry should not take actions that in the long run harm the supply system.

3.6 Companies should hire for diversity and tap aggressively into groups thatare underrepresented in the IT profession.

Professional Societies

4.1 The professional societies should provide greater assistance in theretraining and continuing education of IT professionals.

4.2. The professional societies should take a more proactive role in thecertification of IT professionals.

4.3 Professional societies should continue to play a strong role in curriculumdevelopment.

4.4. The professional societies should take considerably greater interest in non-degree programs that train IT professionals.

4.5. The various IT professional societies should communicate, cooperate, andcollaborate more with one another on issues of worker supply and demand.

Individuals

5.1. Workers should recognize that they must take responsibility for remainingindividually competitive.

5.2. Individuals must commit themselves to life-long learning in order to remaintechnically current and competitive.

5.3. Individuals should do their part to see that people with appropriate skillsenter the IT workforce.

5.4. Individuals should help to build up the IT profession through its professionalorganizations.

Source: Computing Research Association, Intersociety Study Group on InformationTechnology Workers, April 1999.

135 143

Appendix A. Methodology

In the course of this study, the study group has tried to be as objectiveas possible in its approach. The Computing Research Association (CRA) andthe other professional societies organizing this study did not take a positionon the reports prepared by the Information Technology Association ofAmerica (ITAA) and the Department of Commerce, or on the H-1Blegislation."' Prior to the study, some of our academic members reportedrapidly increasing undergraduate enrollments in computer science and similarrises in information technology (IT) recruitment on campus; and some ofour industrial members had mentioned their difficulties in hiring workers.But we did not know whether these were isolated or general phenomena;and we certainly had not reached any conclusions or recommendations inadvance of the study. We endeavored to hear as many viewpoints as timewould permit. We were well positioned to do so because the societiesparticipating in the studyCRA and its five affiliatesrepresent approxi-mately 100,000 U.S. computing professionals; and because CRA is one of thefew computing organizations that includes both academic and industrialorganizations as members. We tried to broaden the perspective further by:

1. Consulting with government officials and staff (from the NationalScience Foundation, DARPA, the White House Office of Science andTechnology Policy, the U.S. General Accounting Office, and Congress) atthe beginning of the study to learn what issues related to the worker issuewere of most concern to them.

112 Four of the five affiliated societies had no programmatic contact with these issues.The situation of the remaining collaborator, the IEEE Computer Society, is more compli-cated. The parent organization of the Computer Society, the Institute of Electrical andElectronics Engineers, is a worldwide organization that operates a national U.S. profes-sional organization in Washington, DC, known as the IEEE-USA. The IEEE-USA was vocallyopposed to the H-1B legislation. However, the Computer Society did not take a position oran active role in the lobbying efforts of the IEEE-USA and had no control over them. In itsmajor publication, Computer, the Computer Society ran guest editorials and acceptedletters to the editor that were fairly evenly split between pro and con on the H-1Blegislation. CRA regards the Computer Society as a neutral organization with respect tothe IT worker legislation, although the IEEE-USA is clearly not neutral on this issue.

1 3 6147

2. Forming a study group of twenty-three professionals, each ofwhom brought some particular expertise such as:

a. knowledge of supply issues (e.g., specialists on K-12, two-yearcollege, four-year college, graduate, and continuing education);

b. knowledge of demand issues (e.g., college recruiters, managers ofresearch laboratories, workers in non-IT industries that needed IT work-ers); and

c. knowledge of contextual factors (e.g., people familiar with past ITworker shortages, demographers, and social scientists familiar with soft-ware production and use).

3. Engaging consultants who had knowledge of data collection andanalysis in science and technology, and labor economists familiar with ITwork.

4. Collecting information from many organizations, such as theIEEE-USA and other professional societies that represent IT occupations,such as the Association of Information Science and the Society for Infor-mation Management.

5. Having the draft report reviewed by more than fifty people frommany different sectors of U.S. society, including people who took variouspositions on the ITAA and Commerce reports and on the Hl-B visa debate.

As CRA is an organization representing research scientists, one way inwhich we tried to be objective was to take a "scientific" approach to thisstudy by letting the data drive the analysis wherever possible. Neither time(8 months) nor budget allowed us to collect new data. However, wesearched extensively to locate and gain access to existing data sets that webelieved might have a bearing on this study. Our list of data sources isdescribed in chapter 9. NSF generously ran some new cross-tabulations forus on some of their existing data sets. For each data study, we reviewedthe organizations and individuals surveyed, the categories in which dataresponses were reported, the sample size, and other factors to determinethe general reliability of the study. We compared and contrasted studiesand paid particular attention where studies gave conflicting reports.

We were only modestly successful in this quantitative approach. Wewere frequently frustrated by how old the data were, how poorly theoccupational categories used in surveys fit with the current situation, andhow many questions were simply not addressed by any existing data. Tofully address the worker shortage issues, we often had to go beyondquantitative evidence and rely on qualitative evidence and expert judgment.Where we did this, we were particularly careful to obtain extensive review.We did not employ any formal method to reduce the many opinions weheard to a general consensus. Instead, we simply weighed all the evidenceand used our long experience in the field to make our best judgments.

148 1 r)

Appendix B. Related Studies

In the course of this study, we had access to the two reports pre-pared by the Information Technology Association of America (ITAA), theDepartment of Commerce report, the U.S. General Accounting Office(GEO) report, the GAO's House testimony, a National Research Council(NRC) report on computer professionals, a variety of statistical informa-tion collected by the federal government and private organizations, andmany other materials produced by the private sector. These sources arecited in footnotes.113 It was our intention in this study to take a fresh lookat the evidence and conclusions presented in these studies. We have consid-ered them and the issues they discuss in terms of general questions aboutinformation technology in the national economy.

The ITAA, Commerce, and GAO studies are basic to understandingthe current debate, and they are discussed in some detail in chapter 1. Inaddition, there are three other reports that have not received the publicattention they deserve. The Jerome Levy Economics Institute of BardCollege convened a symposium in June 1998 entitled, "Is There a Shortageof Information Technology Workers?" The proceedings of this sympo-sium present measured, succinct statements by ITAA, Commerce, andGAO, as well as by other leading labor economists and policy analysts."4The second report is a 1993 study on computing professionals preparedunder the auspices of the National Research Council (NRC)."5 Some of

113 The study group did not have an opportunity to consider three recent reportssponsored by the Sloan Foundation and the United Engineering Foundation as part of anIT Workforce Data Project. See Richard Ellis and B. Lindsay Lowell, "Core Occupations ofthe U.S. Information Technology Workforce," January 1999; "The Production of U.S.Degrees in Information Technology Disciplines," January 1999; and "Foreign-OriginPersons in the U.S. Information Technology Workforce," March 1999. A final report willassess indicators of demand for people with IT skills. [e-mail: ellisOcvns.net]

114 "Is There a Shortage of Information Technology Workers?" Symposium Proceedings,The Jerome Levy Economics Institute of Bard College, June 12, 1998.

115 Computing ProfessionalsChanging Needs for the 1990's, National Research Council,National Academy Press, 1993.

1 Q Q 149

the conclusions seem wrong in hindsight: a slowing demand for comput-ing professionals in general and high-level researchers, and the adequacy ofsupply. However, the NRC report shares many conclusions with this study:the importance of information technology to the national economy, theopportunity losses created by worker shortages, the serious lack of dataneeded to reach conclusions, the rapid change occurring in desired skill setsof IT workers, the need for increased efforts by the universities, and theneed to improve and increase continuing education.

Third, we point the interested reader to a paper on "Skill Mismatchesand Worker Shortages" prepared by Burt Barnow, John Trutko, andRobert Lerman on behalf of the U.S. Department of Labor."' The studygroup has relied heavily on their analysis of federal data, although in thisreport it has, in some cases, been presented in a different form. We havealso relied on their economic analysis of indicators and responses ofworkers and employers in the face of labor shortages, although we havetried to go beyond their analysis by giving detailed information about howtheir mechanisms work in the IT labor market.

116 Burt S. Barnow, John Trutko, and Robert Lerman, "Skill Mismatches and WorkerShortages: The Problem and Appropriate Responses," submitted by The Urban Instituteto the Office of the Assistant Secretary for Policy, U.S. Department of Labor, draft finalreport, February 25, 1998.

150 1 3

Appendix C. About the Authors and the Sponsoring Organization

The study was conceived and directed by the Computing ResearchAssociation (CRA), a nonprofit educational organization located in Wash-ington, DC, whose mission is to promote research and advanced educationin computing. Its members are university departments of computerscience and computer engineering, as well as laboratories and centers inindustry and government that conduct research in computing. The studywas carried out in collaboration with five other major computing profes-sional societies that are affiliates of CRA: the American Association forArtificial Intelligence, the Association for Computing Machinery, the IEEEComputer Society, the Society for Industrial and Applied Mathematics, andthe Usenix Association.

The principal authors of the study are Peter Freeman and WilliamAspray. Peter Freeman (Ph.D., computer science, Carnegie Mellon Univer-sity) is Dean and Professor in the College of Computing at GeorgiaInstitute of Technology. He is a long-standing member of CRA's Boardof Directors and former chair of its Government Affairs Committee.William Aspray (M.A. mathematics, Wesleyan University; Ph.D., history ofscience, University of Wisconsin) is the Executive Director of the Comput-ing Research Association.

140151

Appendix D. Study Group Members

Peter Freeman (study Co-Chair)*

William Aspray (study Co-Chair)*

Kenneth Anderson

Avron Barr

Douglas Busch

Robert D. Campbell

Morgan Cole

Paul Davis*

Timothy Finin*

Thomas D. Fleming

Dennis Frailey

Jimmie Haines

Georgia Institute of Technology

Computing Research Association

Pennsylvania Institute ofTechnology

Stanford University

Intel

Rock Valley College

Microsoft Corp.

Worcester Polytechnic Institute

University of Maryland, Baltimore Co.

IBM

Raytheon Systems Co.

Boeing

141 153

Mary Jane Irwin* Pennsylvania State University

Richard H. Jaross Sun Microsystems, Inc.

Stephen Johnson* Transmeta Corp.

Rob Kling Indiana University

Doris K. Lidtke Towson University

Robert Ritchie (formerly Hewlett-Packard)

Richard Skinner Clayton College and State University

Michael Teitelbaum Alfred P. Sloan Foundation

Shirley Tessler Stanford University

Stephen You* Arizona State University

Stuart Zweben* Ohio State University

[* Steering Committee Member]

1 4 7154

Appendix E. Computing Research Association Board of Directors

Frances E Allen IBM Tj. Watson Research Center

Sandra Johnson IBM T.J. Watson Research CenterBaylor

Anita Borg Xerox PARC

Robert Cartwright Rice University

Timothy Finin AAAI University of Maryland, BaltimoreRepresentative County

Peter Freeman Georgia Institute ofTechnology

James D. Foley Treasurer Mitsubishi Electric IT CenterAmerica

John Gannon University of Maryland

C. William Gear NEC Research Institute Inc.

Mary Jane Irwin Pennsylvania State University

Leah Jamieson Purdue University

David S. Johnson AT&T LabsResearch

Stephen C. Johnson USENIX Transmeta Corp.Representative

Sidney Karin University of California, San Diego

143 155

Edward Lazowska Chair

Nancy G. Leveson

Jill P. Mesirov SIAMRepresentative

James H. Morris

David Patterson

Guylaine M.Pollock

Daniel A. Reed

Barbara Ryder

Kenneth C. Sevcik

Larry Snyder

Mary Lou Soffa

Eugene Spafford

John A. Stankovic

Jeffery Ullman

Mary K. Vernon

Stephen S. Yau

Stuart Zweben

156

IEEE-CSRepresentative

Secretary

Vice Chair

ACMRepresentative

IEEE/CSRepresentative

ACMRepresentative

University of Washington

Massachusetts Institute ofTechnology

Whitehead/MIT Center forGenome Research

Carnegie Mellon University

University of California,Berkeley

Sandia NationalLaboratories

University of Illinois

Rutgers, The StateUniversity of NJ

University of Toronto

University of Washington

University of Pittsburgh

Purdue University

University of Virginia

Stanford University

University of Wisconsin

Arizona State University

Ohio State University

Appendix F. External Reviewers

Early drafts of the report were reviewed by the 23 members of thestudy group and many of the 31 members of the CRA Board of Direc-tors. In addition, the following external reviewers provided comments:

Eleanor Babco Commission on Professionals inScience and Technology

Marjory Blumenthal Computer Science and Telecommu-nications Board, NRC

Toni Carbo University of Pittsburgh

Eileen Collins National Science Foundation

Bill Curtis Teraquest

Peter Denning George Mason University

Jorge L. Diaz-Herrera Southern Polytechnic State University

Larry Finkelstein Northeastern University

Brian L. Hawkins EDUCAUSE

John Keaton IEEE Computer Society

Charlotte Kuh National Research Council

157

Richard LeBlanc Georgia Institute of Technology

Ephraim McLean Georgia State University

Fred Niederman University of Baltimore

Mark Regets National Science Foundation

Ruth Carlson Robertson University System of Maryland

Stephen Seidman Colorado State University

Joseph Turner Clemson University

Margaret Wright Lucent Technologies, BellLaboratories

158 146

Appendix G. Acknowledgments

Catherine Gaddy, who during most of the project was ExecutiveDirector of the Commission on Professionals in Science and Technology,provided expert assistance with identification and analysis of data sources.Near the end of the project, Dr. Gaddy personally ameliorated the laborshortage by becoming an IT worker. Mark Regets of the National ScienceFoundation kindly answered numerous queries and ran new analyses for thestudy group on existing data. Bruce Barnow of Johns Hopkins Universityprovided expert advice on labor economics at several points during thestudy. Robert Lerman of the Urban Institute also shared with the studygroup some of his materials on the labor economics of IT.

Special thanks are due to the three members of the study group whochaired the committees on supply, demand, and contextStuart Zweben,Stephen Johnson, and Paul Davis. They worked extraordinarily hard undertight deadlines, and it is their insights that shape much of the report.

Janice Weber of Georgia Tech provided able support with meetingplanning, note-taking, assisting with text preparation, and general adminis-trative support. Jean Smith of CRA edited the report. Stacy Cholewinski,also of CRA, was responsible for report production and design.

The project would not have been possible without support from theNational Science Foundation, under grant EIA-9812240. John Cherniavskyand Harry Hedges from the NSF staff provided helpful guidance atvarious stages of the project.

159

IMO

Vala g.v-v-my;Gtroliai CaG

1--00ild&MIni

Ilitco:4oe--efr'iotoiltilate 0 A(Ca Pattonmeticooi)

rca-tqiitiaopy iVbxrte_47,

AI-9-0), acornki-fif.Stuart Zweben

Name'CtimM00

711mothyMutaVg, Fralkgy

110ffiffil@ OI@Onw3

Doff%

(1:2204.1-7',-611. &947770Tilitt(g,

kkaitc"AWGVH

Et@ WW1

StephenBarr

(c,0 ovco,v4, (40-7q,/

Pad Otsolt0Mew Jaw; OnvOn

VIRMO©NrKgbffeci SkOnnw

Wili&hae0 T@Inglballuiroadelgf Tessler

Robert

ComputingSeventeenth

C4);();PatiertilZ74;ResearehStreet

Washingtonelephone

Assoeiationegrib 507

20006-463C:202-224.2111,

Son500:Otfa://www. era org

info@era.org

cof-Afteif

CRA

U.S. Department of EducationOffice of Educational Research and Improvement (OERI)

National Library of Education (NLE)

Educational Resources Information Center (ERIC)

NOTICE

Reproduction Basis

ERIC

This document is covered by a signed "Reproduction Release(Blanket)" form (on file within the ERIC system), encompassing allor classes of documents from its source organization and, therefore,does not require a "Specific Document" Release form.

This document is Federally-fmded, or carries its own permission toreproduce, or is otherwise in the public domain and, therefore, maybe reproduced by ERIC without a signed Reproduction Release form(either "Specific Document" or "Blanket").

EFF-089 (3/2000)