It is generally accepted that there is a substantial degree of risk associated with investment in human capital. When the decision to invest is made, uncertainty arises for a variety of reasons. An individual faces uncertainty about the length of life, about the market conditions that will prevail during his or her lifetime, and about the actual returns to investment over the life cycle. One consequence of this uncertainty is that human capital investment will be affected not only by expected returns but also by the riskiness associated with the distribution of anticipated returns. Even though risk is deemed important, empirical specifications of earnings functions typically found in cross-section studies ignore the risk implications of the models [7]. This is an important omission since failing to account for risk considerations may lead to biased estimates of the rate of return to education and may lead to inappropriate conclusions concerning wage discrimination. The purpose of this paper is to investigate the impact of human capital investment on both expected earnings and the variance in earnings, one measure of the riskiness of investment, and to demonstrate the importance of risk and risk aversion in determining the rate of return to schooling. The paper begins by showing that the standard stochastic specification of the earnings function is overly restrictive with respect to risk considerations. In particular, the traditional specification implies that the explanatory variables must have the same qualitative affect on both expected returns and risk. Thus, the standard earnings function can not be used to examine the relation between human capital investment and risk in any meaningful way. Next we examine the risk implications of a more general stochastic earnings function. The specification we use is based on the work of Just and Pope [3] and Griffiths and Anderson [2]. It allows a variable to be either risk increasing or risk reducing independent of its impact on expected returns. Section III contains the empirical results. Using data from the National Longitudinal Surveys, we estimate the generalized earnings function. We find that higher levels of general human capital, as proxied by years of schooling, experience, and I.Q., are positively correlated with the variance in earnings; higher levels of specific human capital, as proxied by job tenure, are nega-
Read full abstract