Demographic Variables in the Economics of Education

Abstract

In this paper I shall try to trace associations between certain demographic, educational, economic and social variables and to examine the form of these associations. The method I shall use is a cross-section analysis of country data which, despite differences in definitions and in reliability, has the merit of enabling a wide range of variation to be studied. My first effort in this direction was set out in a paper (1972) in which I made exclusive use of log-linear relationships. Clearly this form could only be approximate since most of the relationships with which I am concerned must, in principle, be sigmoid: for instance, the proportion of the population under 15 or illiterate must lie between 0 and 1. Accordingly, in this paper, besides repeating and extending the results of (1972), I have experimented with several other forms of relationship and displayed the results in a series of diagrams.

I should like to thank my friend and colleague Angus Deaton for many interesting discussions in connection with the preparation of this paper and for programming and carrying out the calculations relating to the sigmoid relationships. He is not, of course, responsible for any errors or ambiguities that may be found in what I have written.