Changes in the return to unobservable skills are often inferred from changes in the variance of wages within groups of workers with common observable characteristics. The crucial assumption for such an approach to be valid is that the variance of unobservable skills within these groups remains constant over time. Motivated by the fact that the expansion of higher education has likely led to a pool of university-educated workers which is more heterogeneous in terms of their unobservable skills, we propose a new identification strategy which relaxes the assumption of constant within-group variance and can be implemented using longitudinal data. We estimate the changes in the return to skills and the changes in the variance of unobserved skills within 12 education-experience groups over the period 1982-2012 using our new identification strategy with data from the Current Population Survey’s Merged Outgoing Rotation Groups. We find that relaxing the assumption of constant within-group variance is crucial. Contrary to the conclusions drawn when this assumption is imposed, we find that the return to skills has fallen over our sample period among college graduates. There is evidence of substantial increases in the dispersion of unobserved skills within this group.
Presented by:
Matias Cortes (University of Manchester)
Date & time:
March 9, 2015 4:00 pm - March 9, 2015 5:30 pm
External seminars home