Method used

The point of departure is the standard Mincer-type wage equation with log individual earnings regressed on years of schooling and years of work experience and its square. The individual earnings are defined as gross hourly wages (for a few countries, only net hourly wages are available). Years of schooling are transformed from information on the single highest education completed by assigning to each degree level the standard years of schooling required for acquiring a degree at each of the educational levels identified. Years of work experience are chosen to refer to the individuals’ potential years in working life, defined as age minus years of schooling minus age of school start (generally 6 or 7).

Because of the different labour market status of women as compared to men in most PuRE countries, the wage equations have been run separately for men and women. The estimations for men have been restricted to men in full-time employment only. The estimations for women, in contrast, have been undertaken for all women whereby the difference in labour market attachment between women in full-time and part-time employment has been accounted for by adding a part-time dummy indicator to the female wage equation.
 
 

Main findings

The empirical evidence obtained from estimating basic Mincer-type wage equations from individual-level data for the 15 PuRE countries points to highly varying average rates of return to staying an extra year in school. The estimated rates of return for the year closest to 1995 are displayed in the following figure, where the countries are sorted according to the size of the female return to years of schooling.

Several conclusions may be drawn based on the figure. First, there are countries with very low returns to years of schooling, like the Scandinavian ones, and there are countries with comparatively high returns to years of schooling, like Finland, Portugal, Switzerland and the United Kingdom. Second, in some countries the return to years of schooling is significantly higher for women than for men, while in others men are better rewarded for their educational investments than are women. Third, despite of these gender differences in returns the general pattern seems to be that the high-return countries are characterised by a comparatively high return for both men and women, and vice versa. In other words, the country rankings according to, respectively, male and female returns to years of schooling are quite similar.

The sensitivity of these rate-of-return estimates has been tested in several important dimensions. The general outcome from this exercise is that the estimates are highly robust. For instance, the inclusion of actual instead of potential years of work experience change the rate of return estimate only marginally for a majority of PuRE countries. Accounting for additional wage-relevant personal and job-related background characteristics also leaves the estimate roughly unchanged. The only exception occurs when extending the wage equation with socio-economic variables, which causes the rate of return estimate to drop by about one half. This outcome, however, is only to be expected since the socio-economic classifications are largely based on the individuals’ education. Furthermore, the rate of return estimate is found to be unaffected by the potential presence of a sample selection bias arising from the individuals’ decision to work or not to work.

The evidence further indicates that there is no clear-cut common pattern in the trends of rates of return to years of schooling among the 15 PuRE countries. For some countries the average return reveals an increasing trend, as in Denmark, Italy and Portugal. Other countries are characterised by decreasing returns, such as Austria and Sweden, while still others display no trend whatsoever. For some countries the trend, moreover, differs between men and women. At least two conclusions may be drawn: first, decreasing rates of return to education do not seem to be the dominant trend in Europe, and second, no general convergence in returns in Europe is discernible.

When replacing the years-of-schooling variable with dummy indicators capturing the wage effects of educational degree levels, the results suggest that the relationship between wages and schooling is not strictly linear (as is assumed in the standard Mincerian wage equation). The extent of this "non-linearity" in returns varies substantially across PuRE countries, though, and for at least some countries the linearity assumption seems to be quite feasible. An important outcome of this analysis is that most PuRE countries are characterised by weak incentives to continue in education. These weak incentives, however, occur at quite different stages at the educational scale. In some countries the incentives to continue are quite weak at the lowest end of the educational scale, in others in the middle or in the upper part of the educational scale.