NBER WORKING PAPER SERIESSKILL GAPS, SKILL SHORTAGES AND SKILL MISMATCHES:EVIDENCE FOR THE USPeter CappelliWorking Paper 20382 BUREAU OF ECONOMIC RESEARCH1050 Massachusetts AvenueCambridge, MA 02138August 2014Thanks to colleagues at the World Economic Forum’s Global Agenda Council on Employment forhelping me understand developments outside the US, to David Coats for help with the UK literature,Pascaline Descy and colleagues at CEDEFOP(European Center for the Development of VocationalTraining), Jerry Jacobs, Larry Kahn, Paul Harrington, and Paul Osterman for helpful comments.Theviews expressed herein are those of the author and do not necessarily reflect the views of the NationalBureau of Economic Research.NBER working papers are circulated for discussion and comment purposes. They have not been peerreviewed or been subject to the review by the NBER Board of Directors that accompanies officialNBER publications. 2014 by Peter Cappelli. All rights reserved. Short sections of text, not to exceed two paragraphs,may be quoted without explicit permission provided that full credit, including notice, is given tothe source.

Skill Gaps, Skill Shortages and Skill Mismatches: Evidence for the USPeter CappelliNBER Working Paper No. 20382August 2014JEL No. J08,J23,J24ABSTRACTConcerns that there are problems with the supply of skills, especially education-related skills, in theUS labor force have exploded in recent years with a series of reports from employer-associated organizationsbut also from independent and even government sources making similar claims. These complaintsabout skills are driving much of the debate around labor force and education policy, yet they havenot been examined carefully. The discussion below examines the range of these charges as well asother evidence about skills in the labor force. There is very little evidence consistent with the complaintsabout skills and a wide range of evidence suggesting that they are not true. Indeed, a reasonable conclusionis that over-education remains the persistent and even growing situation of the US labor force withrespect to skills. I consider three possible explanations for the employer complaints as well as the implicationsassociated with those changes.Peter CappelliThe Wharton SchoolCenter for Human ResourcesUniversity of PennsylvaniaPhiladelphia, PA 19104-6358and [email protected]

Skill Gaps, Skill Shortages and Skill Mismatches:Evidence and Arguments for the USPeter Cappelli1Prepared for ILR ReviewAbstract:Concerns that there are problems with the supply of skills, especially education-related skills, in the USlabor force have exploded in recent years with a series of reports from employer-associated organizationsbut also from independent and even government sources making similar claims. These complaints aboutskills are driving much of the debate around labor force and education policy, yet they have not beenexamined carefully. The discussion below examines the range of these charges as well as other evidenceabout skills in the labor force. There is very little evidence consistent with the complaints about skills anda wide range of evidence suggesting that they are not true. Indeed, a reasonable conclusion is that overeducation remains the persistent and even growing situation of the US labor force with respect to skills. Iconsider three possible explanations for the employer complaints as well as the implications associatedwith those changes.Introduction:The assertions that there are wide-spread problems with the supply of skill in the U.S. have beencommon in recent years. Stories in the popular press of individual employers who report thatthey cannot fill vacancies are frequent,2 but there have also been a large number of detailedreports by business associations, individual companies, as well as independent organizationsarguing that skill problems are widespread. The rise of these stories is especially surprising as1Professor of Management at the Wharton School and Director of the Center for Human Resources and NBER.This paper was prepared for the ILR Review. Thanks to colleagues at the World Economic Forum’s Global AgendaCouncil on Employment for helping me understand developments outside the US, to David Coats for help with theUK literature, Pascaline Descy and colleagues at CEDEFOP(European Center for the Development of VocationalTraining), Jerry Jacobs, Larry Kahn, Paul Harrington, and Paul Osterman for helpful comments.2A Google search for the phrase “Skill gap” shows over 330,000 references just in 2013.1

they appear to have increased since the 2008 Great Recession when the flood of unemployed jobseekers – most of them recently employed - far exceeded available job opportunities (BLS 2014).The evidence driving the complaints about skills does not necessarily appear where labor marketexperts might expect to see it, such as in rising wages. Instead, it comes directly from employers- typically from surveys – who report difficulties hiring the kind of workers they need. Theassertions explaining their reported difficulties center on the idea that the academic achievementof high school leavers is inadequate or that there are not enough college graduates in practicalfields like computer science and engineering. The recommendations from these reports includeincreased immigration and use of foreign workers as well as efforts to shape the majors thatcollege students choose.These reports have had a powerful influence on public debate about the state of the labor marketas well as the performance of high schools and colleges. Virtually all of them are framed interms of concerns about the economy as a whole, but it is difficult to escape the fact that theproducers of many of these reports have a material interest in the outcomes of the policies theyare attempting to influence.The arguments below examine the claims associated with complaints about the supply of skills inthe US. The evidence they present as well as evidence from other sources suggests little merit totheir claims. An important question raised by these reports, however, is who is responsible forthe skill level of job candidates and, ultimately, of employees. Traditional employer humanresource practices, such as assessing the abilities of job applicants, training employees forcurrent jobs, and developing them internally for future roles saw employers as responsible forsecuring the skills they need. The thrust of these reports suggests that the education sector –2

especially public-funded education – and job candidates themselves should be responsible forproducing the skills that employers want. Such a change would have profound implications forsociety and is worth considering seriously.Framing the Problem: The arguments that there are problems with the supply of skills availablein the labor market take various forms. The most extreme complaint is the idea that there arewidespread shortfalls in the basic skills of future employees. The cause is usually attributed tothe failure of the education system, especially K-12 public education, to provide students withthese basic skills. We refer to that position as a “skills gap,” following its use in policydiscussions.A second complaint focuses more on job-related skills of the kind associated with particularoccupations, such as the common assertion that the U.S. is short on engineers or informationtechnology specialists. We refer to this assertion as a “skills shortage.”The final concern, which is much more commonly articulated outside the U.S., is the moregeneral idea that at any given time, the supply of skills and the demand for skills could be out ofsynch in either direction - oversupply or undersupply. This situation could occur in specificlabor markets, although with respect to educational credentials it is usually considered at thecountry level. We refer to it as a “skills mismatch.” A skill shortage is obviously a particulartype of skill mismatch, and a skills gap could be a general form of mismatch. All thesecomplaints collectively can be referred to as “skill problems.”The first challenge in assessing the assertions about skills problems is to have a conceptualframework for understanding the relationship between workers and their skills against employer3

needs. One approach, traditionally associated with the topic of internal labor markets and theacademic field of human resources and before that personnel, suggests that matching skills to jobrequirements is an employer problem. Over time, employers have internalized the supply oflabor, selecting for general abilities at entry-level positions and then training and developingemployees over a working lifetime to meet their specific skill needs (Jacoby 1983). Thatapproach appears to have eroded substantially in recent years (e.g., Cappelli 1999), an issue wereturn to below.The other approach focuses on the labor market as the mechanism for meeting employer skillrequirements. The idea of job matching between employers and job seekers implies job search,which typically assumes that employers have job requirements that are generally determinedexogenously, and employers then go searching for job applicants who have those skills. Thesearch process is realistically described as “two-sided” if both employers and employees arelooking for a match, and a good match is one where the skills of the applicants and therequirements of the job fit closely: neither a shortfall nor an oversupply of skill relative to thoserequirements.In typical economics models of job search, the process is reasonably passive: Employers makeoffers to job applicants who accept the job when the offer matches or beats their reservationprice. Employers raise wages to attract better applicants whose attributes are closer fits for thosejob requirements, and they fall if there is an excess supply of such applicants (see Mortenson1986 for the framework of job search models). The notion of a “shortage” is foreign to thismodel and to most all economics-based models. Indeed, shortages in general are typically seen4

as occurring only in the context of market failure, such as wartime wage freezes or restrictionson mobility, and temporary until candidates and employer adjust.3In practice, of course, employers can search more extensively through recruiting activities andmore careful selection, and applicants can search actively by securing better information aboutvacancies. We also know that job requirements are not exogenous from the supply of applicants:A shortfall of applicants that leads to higher wages in turn causes employers to substitute capitalfor labor in order to create new jobs with lower skill requirements (see below). Empiricalevidence indicates that employers also lower the skill requirements for given jobs when labor isrelatively scarce and raise them when higher quality applicants are plentiful (Walsh 1977;Brencic 2010).What is less clear in typical models is how the supply of skills affects employer decisions onproduction systems and, ultimately, productivity. It certainly is true that “better” workers whoare absent less, who shirk less, and who work harder will improve organizational productivityand performance even if nothing changes about their jobs (e.g., Cascio 2008 for a review). Butwhether more skilled applicants per se cause employers to innovate, to adopt more effectivepractices, or to change the way that jobs are performed is an open question that is often part ofthe skill gap arguments.The recent assertions about skill problems have quite a different underlying model in mind,although it is typically unstated, and it does not include a role either for internal labor markets orfor the labor market. Instead, the arguments are akin to input-output models associated with3See Mortenson 1986 for a basic framework, Rogerson, Shimer, & Wright (2005) for an overview of jobsearch models, and Borjas (2010) Chapter 6 for an survey of general labor supply questions.5

operations research optimization exercises. Perhaps the closest analogy is with supply chainmodels where suppliers are trying to produce just the right amount of output to meet the needs oftheir clients at the previously agreed price (see, e.g., Cachon and Terwisch 2007). Skills are seenas coming with the applicant to the job, and job requirements are absolute such that candidateseither have the necessary skills to do a job or not and if not, they cannot do the job. Finally, animportant goal for public education in these arguments, including public colleges anduniversities, is seen as providing graduates that employers would like to hire.Reports of Skill Gap and Skill Shortages:The Skill Gap Idea: The broadest and perhaps most general complaint about skills has been theskill gap idea, that there is some systematic shortfall in skills, broadly defined, across entire agecohorts of the population. Typically the argument is that the decline is associated with the poorskills of school leavers, and the explanation for that shortfall is usually that schools have failedso that academic performance of students has declined.Concerns like this in the U.S. trace their contemporary roots to the Cold War and the 1958National Defense Education Act, which increased funding for science and engineering educationin an effort to compete with the Soviet Union. The A Nation at Risk (1983) report highlighteddeclines in student achievement of all kinds in the 1970s and helped cement in the mind of thepublic for decades after that US schools were failing.An equally attention-getting report by the Carnegie-funded National Center on Education and theEconomy, America’s Choice: High Skills or Low Wages? (1990), argued explicitly thatproductivity growth depended on increasing the skill level of the US workforce. Although thereport estimated than only five percent of employers thought they face any skill problems, that6

low number was attributed to the fact that employers were not introducing new, highperformance work practices that would increase productivity. Among its recommendations wasto establish national educational standards for students based on international standards, theestablishment of skill certifications, and national training boards to organize smooth transitionsbetween school and work.A separate report for the US Secretary of Labor represented an extension of the America’sChoice study. The Secretary’s Commission on Achieving Necessary Skills (SCANS 1991) alsoenvisioned a future where employee empowerment had increased, where workplaces had movedtoward high performance work systems that required greater skills, and where employers trainedconstantly and saw human resources as investments rather than costs. The recommendations,which became known as SCANS Skills, were central to the policy debates of the 1990s. Theycalled for a generic set of skills from high schools that included basic skills (reading, writing,math, etc.), thinking skills, such as decision making, and personal attributes such asresponsibility. Developed in the Republican Bush administration, the SCANS ideas werenevertheless embraced by the Democratic Clinton administration and dominated the discussionof skills throughout the 1990s.The most important action on skills in the 1990s was arguably the school-to-work movement,based in part on SCANS arguments, which asserted that the way to improve student skills and toincrease employability was to bring school and employers closer together in an effort to smooththe transition from school into work. In practice that meant apprenticeships, coop programs,internships, and other arrangements that would help students see the practical value of classroomlessons first by using more business and workplace examples in the classroom and second byseeing how those examples could be applied in at work. The School-to-Work Opportunities Act7

of 1994 (STWO) provided administrative and financial support to help build those connections.(See Stull and Sanders 2003 for an overview of the school-to-work movement and publicpolicy.)Joyce and Neumark (2001) find that 64 percent of schools had at least one school-to-workprogram with employers (the most popular of which was “job shadowing”), and data from theNational Longitudinal Survey of Youth suggests that 38 percent of students participated inschool-to-work programs. Census data from establishments finds 71 percent of for-profitestablishments reporting that they were involved in some school-to-work program with theirlocal schools (Cappelli 2001). Because establishments represent the location where businesstakes place rather than the firm per se, this suggests quite extensive business involvement.The high water mark of the SCANS effort was the National Skills Standards Act of 1994, whichwas designed to create a voluntary system of national standards for job skills, overseen by aNational Skills Standards Board. The initial efforts were organized around industries, butconsensus on standards proved elusive, and interest waned. The Board’s last publication,seeking comments on an approach for creating standards, was in 2002.The Labor Shortage Argument: With the sunset of the STWO Act in 1999, a newAdministration less interested in a role for government in affairs related to business, and therecession of 2001 that cooled off the tight labor market, the debate about skills changed sharply.Funding for SCANS-related efforts and school-to-work programs ended as did arguments aboutthe need to push employers toward high performance work systems and to create skill standards.Most of the skills-related arguments following the 2001 recession came initially from consultantswho asserted that in the near future there would simply not be enough people to meet labor8

demand. Those arguments were kicked off by McKinsey & Co’s “The War for Talent” study(Chambers, et al. 1998), which observed the smaller “baby bust” age cohort born in the 1970sand asserted that it would soon cause a shortfall in middle-aged employee talent, although whythere would be a special need for middle-age talent was not clear. Similar reports followed, suchas The U.S. Chamber of Commerce (2006:13), which assumed that the impending retirement ofthe baby boom cohort would lead to an absolute decline in the size of the labor force and a“severe worker shortage.” (See Public Policy and Aging Report 2004 for a collection of articlesarguing that a labor shortage was coming and Challenger 2003 and Carnevale 2005 for similararguments.) 4These arguments are puzzling in part because the basic claim that the population and potentiallabor force is or will be shrinking seems to be a simple misreading of the facts: Only the rate ofincrease in the labor force was expected to slow, assuming baby boomers do not delay retirement(see CBO 2013 for a review of forecasts).5Several researchers pointed out other problems with these labor shortage arguments, such as theanticipated slowdown in the rate of increase in the labor force is trivial compared to changes inthe demand for labor associated with the business cycle (Cappelli 2003, Freeman 2006,4Concerns that the demographics of the US will cause some shortfall in the supply of labor, affecting business andthe economy, continued to be voiced by The Aspen Institute (2003), Dychtwald et al. (2006), and others. The GreatRecession and unemployment rates at double-digit levels have not stopped these arguments. Venneberg andEversole (2010), Bluestone and Melnik (2010), and others assert that the retirement of the baby boomers will lead toapocalyptic shortages, with as many as 30-40 percent of jobs in sectors like healthcare being vacant.5The fact that the labor force participation rate in the US has declined since the Great Recession in 2008 may saylittle about the longer-term availability of workers as much of the decline appears to be driven by the lack of jobopportunities. In March of 2014, for example, there were 2.2 million individuals marginally attached to the laborforce who were available for work but not counted as unemployed because they were not actively searching for jobs.In March of 2008, just before the economic downturn began, the figure was 1.4 million, according to the Bureau ofLabor Statistics Employment Situation News Release for those months. . See MLR 2013 for information on thelabor force participation rates.9

Neumark, Johnson, and Mejia 2011, and Harrington and Sum forthcoming). Despite the obviousproblems with these labor shortage arguments, the Society of Human Resource Management(SHRM 2003) reported that large numbers of employers in the early 2000’s were preparing for alabor shortage predicted to occur by 2010. None of these projections proved to be right.Skill Gaps and STEM Skills Reports: If the SCANS approach was aspirational (“here’s whatthe workforce should have”), the more recent arguments have been grounded in the present,asserting that there is currently a shortfall in the skills of the workforce. The American Societyof Training and Development (ASTD), whose members hold jobs as trainers, in employeedevelopment, etc., asserted that view in a series of annual reports beginning in 2003. The mostinteresting part of their most recent report (ASTD 2012) is that the most important explanationsmembers saw as causes of the shortfall in skills were management actions internal to theorganization.6 Skills problem were seen as self-inflicted by management.Other reports argue that there will be a shortage of skills associated with college education. ThePresident’s Council on Jobs and Competitiveness, a business-led council (20 of its 24 membersare from business), claim that the US would have a shortfall of 1.5 million college graduates by2020, citing McKinsey & Co. as the source (Jobs Council 2012). In perhaps the most alarmistreport, Carnevale, Smith, and Strohl (2010) conclude that the demand for college graduates inthe US will fall short of supply by 3 million individuals by 2018. Harrington and Sum (2010)pointed out one fundamental problem with projections like these is the assumption that every jobcurrently held by a college graduate requires the skills associated with that degree. As LevineThe most common response was that their organization had changed “strategy, goals, markets,or business models” such that the current workforce did not have the skills to meet those newapproaches; in second place was the lack of “bench strength” or possible replacements for thecurrent leadership; third place was that cuts in training investments and lack of commitment todevelopment caused a shortfall. A shortfall in the skills of applicants came in fourth place.610

(2013) points out, 60 percent of parking lot attendants in Wisconsin have at least some collegeeducation yet those jobs surely do not require college-level skills.The focus of skill gap arguments in the 2000s shifted from broad SCANS-type skills to academicskills associated with Science, Technology, Engineering, and Math (STEM) education,particularly four-year college degrees. There have been complaints from the businesscommunity about shortfalls in the supply of such skills before (e.g., Atkinson 1990 predicted ashortfall of 400,000 scientists), but the intensity of the arguments increased sharply especiallywhen they were joined to the debate about immigration.A report from the US Department of Commerce (1997) argued that there was a severe shortageof IT workers in the US that required increased immigration as well as expanding education.That claim and the report itself was criticized by a Government Accounting Office report (Joyner1998) that noted methodological problems with it such as the assumption that only graduateswith IT degrees can do IT-related jobs and reliance on evidence from a tech employer group(e.g., with a survey response rate of only 14 percent). The National Research Council’s (2001)report on the IT workforce avoided taking a clear position on whether there was an IT laborshortage, but its recommendations relied heavily on traditional actions that employers shouldtake to address their perceived shortages, such as more extensive training.The attention on skill problems shifted to STEM skills with a joint publication of the NationalAcademies of Science, Engineering, and the Institute of Medicine called “Rising Above theGathering Storm: Energizing and Employing America for a Brighter Economic Future”(2007).The charge given to the Committee was how best to support science and technology in the US,both in research and in business application. A central argument of the report is that the cost andavailability of skilled labor in science and technology drives multinational companies in their11

decisions as to where to operate, and increasing the supply of STEM graduates and allowingmore foreign STEM workers and foreign STEM students to enter and stay in the US wouldbenefit the industry. This view represented a sharp break from the America’s Choice andSCANS idea that an important goal of policy was to raise wages. Here the notion was the higherwages were actually a problem for the technology industry and therefore for the advancement oftechnology.Since then, arguments about skill shortages and STEM skills in particular have beencommonplace. The National Academy of Sciences produced six separate reports related toSTEM skill issues just in 2012, many about expanding the supply of skill. The President’sCouncil of Advisors on Science and Technology was one of many studies asserting a shortfall ofSTEM grads, arguing that the US needed an additional one million such graduates to meetdemand (President’s Council 2012). These arguments about a coming labor shortage are verysimilar to those made a decade earlier, the difference being the focus on particular skills.The Institute of Electrical and Electronics Engineers, which represents engineers, disputed theclaims of a shortfall of engineers in particular, noting a range of evidence about the difficultythat many engineers have in getting jobs and suggesting that there is actually a surplus of STEMgraduates relative to demand and that a large percentage of “STEM” jobs are not performed byindividuals with degrees in that field: Few computer programmers, for example, have a B.S. inComputer Science (Charette 2012). Salzman, et al. 2013 also argue against the STEM shortageidea in a report for the Economic Policy Institute (EPI). The strongest evidence in their studyarguably is that roughly half of recent engineering graduates do not take jobs as engineers and of12

those who do not, roughly 30 percent say there was no job available for them; another 30 percentsay that the terms and conditions for those jobs were below market levels.7The Computer and Technology Industry Association (2012) produced an unusually detailedreport, again based on a survey of employers, where 93 percent of employers responding saidthat they had a skills gap. Yet 90 percent also responded that they are at least “moderately close”to “where they want to be” with respect to skills, only 15 percent said that a factor in their skillproblem was insufficient focus on STEM education, and only 20 percent reported that theproblem was a limited pool of skilled IT workers, the essence of the STEM skill shortageargument. A large part of their perceived skill gaps had to do with “soft skills”- work ethic andmotivation (almost 20 percent said that their concern was only soft skills, and about half reportedthat it was equally divided between “hard” and “soft skills”).Individual employers have also produced reports claiming that there are nationwide skillsproblems. For example, the health science company Bayer issued a report (Bayer 2013) based ona survey of recruiters at Fortune 1000 companies reporting that their companies were creatingmore jobs for STEM graduates than for those with any other credentials, a surprising findinggiven that the Bureau of Labor Statistic’s Occupational Outlook forecast projects that only two7For graduates in IT, perhaps the hottest of the true STEM job markets, only about one-third fail to take IT jobs.But about 30 percent of those say that the reason was that there was no job for them, while a much largerpercentage, 53 percent, say the terms and conditions for such jobs were below the prospects elsewhere. Theextremes were represented by health care graduates, who are not always counted as STEM graduates, wherealmost three-quarters got healthcare jobs, and all other STEM grads (e.g., science and math majors) where onlyabout 22 percent got jobs in their field. Atkinson and Stewart (2013) published a rebuttal to the Salzman et for The Information Technology and Innovation Foundation (ITIF). Perhaps the most interesting point intheir report is the observation that the sponsors of the various reports matter: EPI’s goal, they argue, is higherwages for tech workers, they asserted, while their goal was to promotion economic growth, presumably by makingtech labor easier and cheaper for IT employers to engage. (EPI is sponsored in part by labor unions; ITIF issponsored by tech employers.)13

(registered and licensed nurses) of the top twenty occupations with the most new openingsthrough 2020 require any post-secondary education, let alone STEM degrees (BLS 2014).8A related complaint one often hears in STEM arguments is that US college students do not wantto take the hard majors in college like those associated with STEM degrees and instead arecoasting in easier majors and ones that have poor job prospects like liberal arts. In fact, the mostpopular majors in four-year colleges seem to be the most vocational: business is by far the mostpopular, one-in five students major in it (365,000 per year) and job skills are the focus; thesecond most popular major is education (104,000), another vocational field with a very close tieto jobs. The NCES defines liberal ar

The Wharton School Center for Human Resources University of Pennsylvania Philadelphia, PA 19104-6358 . looking for a match, and a good match is one where the skills of the applicants and the . of course, employers can