This paper reviews the theoretical and empirical literature on job market signalling and on education as a job market signal. Possible economic implications of educational job market signalling to an individual and the society are represented based on existing theories. The paper also reviews central methods in empirical testing of the signalling/screening hypothesis. The empirical section of the paper carries out two alternative methods for testing the signalling/sorting hypothesis. The first method is a so-called natural experiment where the Finnish comprehensive school reform (implemented in the 1970s) is used as an exogenous shock variable in an econometric model explaining educational attainment. Besides qualitative changes to the old comprehensive school system, the reform increased minimum school leaving age from 12 to 15 years. Enforcing the idea of Lang and Kropp (1986), I argue that under the human capital hypothesis the reform should only have affected schooling choices of those individuals whose behaviour was directly constrained by the reform, whereas under the sorting hypothesis it should also have affected those who were not directly constrained. I find no evidence of such an indirect effect on post-comprehensive educational attainment as predicted by the sorting hypothesis. However, my results indicate that the reform may have had an effect on non-constrained individuals tertiary educational attainment. I regard this result as tentative, because it clearly contradicts with the ripple effect observed by Lang and Kropp. The second method studies the importance of relative education as an explanatory variable in a Mincerian-style wage equation. I find the conclusions of this method to be dependent on the reference group used in defining relative education. Consequently also the second method yields somewhat inconclusive results on the importance of education as a job market signal. The sorting hypothesis gains most support when I use jointly the regional distribution of education and the age distribution of education to define an individuals education relative to his reference group. The overall impression from the empirical section suggests that the signalling effect of education on wages is minor compared to the human capital effect. I conclude that even though the comprehensive school reform had a positive effect on average productivity, one should not make any hasty generalizations from this result regarding the whole educational system. The sample used in the empirical analysis consists of over 120 000 men, born in Finland between 1962 and 1966. It is a cross-section of the Finnish Longitudinal Census Data File generated by Statistics Finland.
