RETURNS TO EDUCATION

Returns to education

Returns to Educations

Introduction

There are at least three distinct ways of defining the 'returns to education': (a) the private return, (b) the social return and (c) the labour productivity return. The first of these is made up of the costs and benefits to the individual and is clearly net of any transfers from the state and any taxes paid. The second definition highlights any externalities or spill-over effects and includes transfers and taxes (Dearden, 1999a, pp. 551). The final definition simply relates to the gross increase in labour productivity (or growth). A key component of each of these measures is the impact of education on earnings. This is perhaps the aspect of returns to education measurement where statistical methods have been most developed and most fruitfully deployed and is the central focus of this paper. With extensive data available over time and individuals on schooling and on earnings, the measurement of the education effect on earnings is one area where we might expect agreement (Harmon, 1995, pp. 1278).

However, a casual look through the literature on the impact of education on earnings reveals a wide range of estimates and an equally wide range of empirical approaches that have been adopted to estimate the return. So why do the estimates vary so widely and what is the most appropriate empirical method to adopt? The answer to these two questions provides the central motivation for this paper. It is illustrated using the sample of men from the NCDS Birth Cohort data for the UK (Harmon, 2003, pp. 115).

The Earnings-Education Relationship: Alternative Models

This section highlights two central aspects in the empirical investigation of the earnings return to human capital investments. First among these is the distinction between the homogeneous returns and heterogeneous returns model. In the homogeneous returns model the rate of return to gross earnings of a particular human investment is the same for all individuals. Growing statistical evidence and causal empiricism suggests that the homogeneous returns restriction is unwarranted (Weiss, 1995, pp. 133).

The second aspect is to distinguish between the one factor and multiple factor models of human capital. In the one factor model all schooling can be thought of as an investment in a single homogeneous construct called human capital. Each additional unit has the same return. An example of an empirical model that is both one factor and homogeneous returns is the popular linear regression equation - log earnings regressed on years of schooling (Groot, 2000, pp. 57). The constant parameter on the schooling variable is equivalent to homogeneous returns and the use of years of schooling as a single measure of schooling is equivalent to a single measure of human capital.

Earnings and Education in a Homogenous Returns Framework

For each individual i = 1, ..., n, we let yi represents their earnings or hourly wage opportunities in work. To begin with we will assume that we are measuring earnings at one point in time for a sample of individuals who have completed formal ...