I. Introduction, 299. — II. Changes in employment patterns, 300. — III. Negro population changes in the urban areas, 304. — IV. A model of decentralization of employment and its impact on the urban negro's employment problem, 304. — V. The regression results, 307. — VI. Conclusions and implications, 309.

SUBSTANTIAL interest in recent years has centered on the relationship between personal earnings and a myriad of education related variables.' In this paper we present estimates of the impact on earnings of schooling, an index of ability, and a set of other relevant variables for a cohort of recent entrants to the labor market who have had some graduate education in the arts and sciences. Aside from a purely intellectual curiosity, there are several other reasons for investigating the annual earnings for a group of this sort. First, estimates of an earnings function are necessary for calculation of rates of return to various quantities of educational investment. To date, very little work of an economic nature has been done in the growing field of graduate education.2 It is hoped that the estimates presented in this paper may be viewed as an exploratory attempt to come to grips with problems in this important and neglected area. Secondly, we have explicitly attempted the specification of an earnings relationship which allows the differential impact of schooling related variables to depend on the values of other relevant explanatory variables.3 The existence of such interactive effects is interesting in itself and has important implications both for the rate of return analyses already available in the literature and for any further work on graduate education which may be attempted. Finally, there is substantial interest in the specification of an earnings relationship which explicitly attempts to deal with the slippery concept of ability to earn income. We would like to know: (a) What sort of ability index is relevant in the context of highly educated persons, (b) the quantitative importance of an ability index, and (c) how parameter estimates of schooling related variables are changed by the inclusion of an ability variable. We do not hope to provide definitive statements on these issues, but our results should be of some interest to those working on related problems in the economics of education. The plan of the paper is as follows: Section I outlines the nature of the data, variables, and methods used in estimation. Sections II and III present the results of the additive and interactive models. Section IV contains a few concluding remarks.