The Determinants of Teacher Supply

Abstract

In the economic literature, occupational choice has been interpreted as a decision which will be determined by the expected lifetime earnings associated with each of the available alternatives. Despite its apparent simplicity, it has been notoriously difficult to incorporate this idea into the analysis of occupational choice, the main reason being the inadequacy of empirical data on earnings. Given this difficulty, some models of occupational choice have from the outset formulated the problem as a one-period decision. All the work carried out in the US to investigate the factors determining the choice of a military career belongs to this class of models (cf. Fisher (1969), Altman (1969) and Altman and Barro (1971)). Other researchers have taken into account the effect of lifetime earnings in their theoretical analysis, but have focused their empirical work more on evaluating rates of return than on estimating supply elasticities (e.g. Weiss (1972)). Probably the most complete attempt to deal with this problem has been that of Freeman (1971). The bulk of his empirical analysis considers either starting or average salaries as the relevant variables of the occupational supply functions, and in this sense it is not substantially different from the studies on military manpower reported above. In his analysis of career decisions of doctorate manpower, however, he explicitly takes into account lifetime earnings as the explanatory variable of occupational choices. For each of the specialities considered, Freeman estimates discounted lifetime earnings on the basis of geometric mean incomes for graduates in several previous classes. However, he is faced with serious problems to arrive at a similar measure for alternative occupations. Although still consistent with theoretical expectations, his results are not as neat and convincing as those obtained with only starting or average salaries, and in subsequent work (Freeman (1975a) and (1975b)) he has abandoned this line of analysis, to come back to the simpler specifications. In our view, the analysis of occupational choice developed in this paper goes some way towards the solution of these persistent difficulties. The effect of lifetime earnings on occupational decisions is explicitly considered in a theoretical model from which we derive empirically testable hypotheses. The basic idea is to reduce the earnings profile to two dimensions: its level and the rate at which earnings progress over time. In this manner we achieve two results. First, by measuring the profile with two variables rather than one (as is in fact done when only average, or starting salaries, or even present values, are used), we are able to disentangle two different supply responses: the response to starting earnings, and the response to future earnings prospects. This procedure gives considerable insight into the behaviour of groups of individuals with different characteristics (e.g. men and women). Secondly, our model suggests an empirically convenient way of taking into account lifetime earnings which relies only on starting and average salaries. Given the inadequate data with which most occupational studies have to work, this procedure may prove a useful methodological advance in this area. The rest of the paper is as follows. In Section 2 we develop a simple model of occupational choice, which will help in organizing the discussion of the factors that determine the supply of teachers. Then, in Section 3, the model is applied to data on entry and wastage of teachers. Finally, Section 4 summarizes the main results obtained.