ESSAYS ON THE ECONOMICS OF EDUCATION

Abstract

In the first chapter I study racial differences in the impact of education on labor income volatility. Using panel data on black and white males from the Panel Study of Income Dynamics I find that education reduces labor income volatility more for blacks than for whites. The central specifications indicate that college graduation reduces transitory labor income volatility by more than 65 percent relative to high school dropouts for blacks, whereas whites receive no statistically significant reduction. I also find that more risk averse blacks obtain more education while more risk averse whites do not. I argue that these results imply: (1) that precautionary demand for human capital is quantitatively important; and (2) the black differential investment puzzle can be explained by accounting for racial differences in the impact of education on exposure to labor income volatility. The results can be explained by the precision of employer's beliefs about a worker's productivity increasing more with education for blacks, so that more-skilled blacks face less labor income volatility.

Participants, like econometricians, may have difficulty in constructing the counterfactual outcome required to estimate the impact of a program. In this chapter, this question is directly assessed by examining the extent to which program participants are able to estimate their individual program impacts ex-post. Utilizing experimental data from the National Job Training Partnership Act (JTPA) Study (NJS) experimentally estimated program impacts to individual self-reports of program effectiveness after the completion of the program are compared. Two methods are implemented to estimate the individual experimental impacts based on: (1) subgroup variation; (2) the assumption of perfect rank correlation in impacts. Little evidence of a relationship between the experimentally estimated program impacts and self-reported program effectiveness is found. There is evidence found that cognitively inexpensive potential proxies for program impacts such as before-after differences in earnings, the type of training received, and labor market outcomes are correlated with self-reported program effectiveness.