170 likes | 299 Views
Educational Mismatch among Ph.D.s: Determinants and Consequences. Keith A. Bender and John S. Heywood Department of Economics and Graduate Program in Human Resources and Labor Relations, UW-Milwaukee Presentation to SEWP Conference 19 October, 2005. Motivation.
E N D
Educational Mismatch among Ph.D.s: Determinants and Consequences Keith A. Bender and John S. Heywood Department of Economics and Graduate Program in Human Resources and Labor Relations, UW-Milwaukee Presentation to SEWP Conference 19 October, 2005
Motivation • What is an ‘educational mismatch’? • Why should it matter? • Worker costs • Employer costs • Social costs • Why scientists with PhD’s? • Homogeneity of sample • Key for innovation and R&D • Concern that trained scientists are leaving science (Preston 2004)
Some Literature • How many are mismatched? • Approximately 40% in national samples • Why does mismatch persist in equilibrium? • Govt subsidization causes an oversupply of the highly educated (Groot and Maassen van den Brink 2000a) • Educational signals imperfectly correlated with worker productivity, which is difficult/costly to detect (Tsang and Levin 1985) • Internal Labor Markets force maintenance of pay hierarchies (Thurow 1975)
Literature Continued • How is educational mismatch measured? • External job analysis • Worker perceptions of requirements • Effects on earnings? • Falls by approx 14% (Groot and Maassen van den Brink 2000b; Borgans et al. 2000; Allen and van der Velden 2001) • Effects on job satisfaction? • No (Buchel 2002) or slightly negative effect (Solomon et al. 1981, Allen and van der Velden 2001; Moshavi and Terborg 2002) • Effects on turnover? • Generally increased (Solomon et al. 1981; Allen and van der Velden 2001)
Data and Methods • 1997 Survey of Doctorate Recipients • Primary Measure of Mismatch: • Thinking about the relationship between your work and your education, to what extent is your work related to your doctoral degree? • ‘Closely related’ (69.3%) • ‘Somewhat related’ (23.4%) • ‘Not at all related’ (7.3%)
Data and Methods con’d • Outcome Measures • Annual Earnings • Job Satisfaction (four point scale) • Job Change • Other variables • Sector, Demographic, Job characteristics, Current Discipline
Table 2: Educational Mismatch, Earnings, Job Satisfaction and Turnover
Consequences of Mismatch • Log Earnings Regressions • Job Satisfaction
Consequences con’d • Changing Jobs
Consequences con’d • Influence of Age • Possible that since accumulation of and return to human capital takes time, consequences may grow over time • Age-Earnings profiles by Mismatch • Age-Satisfaction profiles by Mismatch
At age 28: 4.5% and 8.6% decrease At age 62: Over 10% decrease At age 28: under 2% decrease At age 62: 12.1% and 20.9% percentage point decrease Age-Earnings Profiles
At age 28: 6.7 and 17.5 percentage point decrease At age 62: 13.3 and 27.2 percentage point decrease At age 28: 11.4 and 16.9 percentage point decrease At age 62: 15.3 and 19.9 percentage point decrease Age-Satisfaction Profiles
Reasons for Mismatch • People who report their job not being closely related to their education are asked why.
Conclusions • Educational Mismatch occurs in the highly educated sector – approx 30%, but more in the nonacademic sector, 43% • Mismatch results in adverse outcomes • 6-12% lower earnings • 10-18% reduction in probability of being satisfied in job • 4-7% more likely to change jobs • Who are the mismatched? • Nonacademics, older workers, those not doing teaching or research, ‘hard’ scientists • No appreciable difference between genders or races