The Effects of Job Evaluation Systems on Wage Effects of Education and Over-education

Unions are held responsible for various wage effects, such as higher wages and w age compression. In this paper we investigate another possible union effect on w ages: are the actions of unions responsible for the observed relation betw een required education and w ages? It is w ell know n that wages are related to education, and there is also strong evidence that working in a job requiring a level of education below one's own level has a negative effect on wages. These effects are usually ascribed to productivity differences, but recent research indicates that at least some of these effects may be nonproductivityrelated. In this paper we examine the extent to w hich relative wage levels of recent Dutch labour market entrants, and of workers in jobs requiring their own level compared to a lower level, are influenced by the institutional arrangements specified in collective bargaining agreements. In particular, we look at differences in the pattern of wages of those working under a bargaining agreement without a job evaluation system compared to those w orking under agreements containing clauses specifying a job evaluation system, the latter being further subdivided into systems excluding and systems allowing w age variability based on individual performance. In line with our expectations, we find that job evaluation systems w ithout flexible components compress w age differences relative to the w ages under collective agreements w ithout job evaluation systems. Contrary to our expectations, flexible components in collective agreements do not decompress w age structures. In fact the opposite appears to be true: the w age differences between education levels are smallest for this group. Finally, in line with our expectations, in highly standardized systems (job evaluation systems without flexible components) wage penalties are lowest, and increase for higher educated w orkers w hen flexibility is introduced. THEORETICAL BACKGROUND The strong relation between education and labour market outcomes is often explained in terms of human capital theory (Becker 1964, Schultz 1963), which claims that people w ith more years of schooling earn more because the competencies they acquired in education have made them more productive. There is also a growing body of research emphasizing the idea that productivity not only depends on characteristics of workers, but also on the nature of the jobs they perform. A poor fit between the educational level of a worker and that required in his/her job is seen as an important determinant of the w age differential observed in the labour market between comparable w orkers (Sicherman 1991, Alba-Ramirez 1993, Cohn and Khan 1995, Kiker et al. 1997; Cohn and Ng 2000, Ng 2001, Bauer 2002). Workers in jobs requiring a low er level of education than their ow n (overeducated w orkers) are penalised in terms of w ages compared to similarly educated w orkers w orking in jobs that match their education (adequately educated workers). These findings have usually been justified by means of the assignment theory (Sattinger 1993), w hich predicts that productivity, and thus w ages, depend on the match between actual and required education. The reasoning is that low er level jobs are thought to give w orkers less opportunity to utilize their competencies than jobs at the appropriate level. Although w age effects of both w orkers’ education and the educational requirements of their jobs is well established, recent analyses have cast doubt on the extent to which these wage effects can be accounted for by individual productivity. For example, scholars such as Spence (1973) and Arrow (1973) have pointed out that pay is often based directly on formal education as a proxy or signal of expected productivity rather than on actual productivity at the level of individual w orkers. Further, recent research by Allen and Van der Velden (2001), Green and McIntosh (2002) and Di Prieto and Urw in (2006) did not find empirical support for the assumption of assignment theory that w age penalties associated w ith over-education are actually due to diminished utilisation of competencies. To the extent that w age effects of education and over-education are at least linked to expected productivity, the implications for workers may be relatively modest. Signalling effects are usually explained in terms of incomplete information and bounded rationality, and as such should diminish over time as employers accumulate more accurate information on the actual performance of individual workers. However, scholars such as Collins (1979) and Bills (2003) have cast doubt on w hether wage effects of education are related to productivity at all, w hether at the level of individuals or of groups. Collins asserts that workers engage in political as well as productive labour, and points to the importance of credentials such as formal education as a kind of entry ticket to the most privileged positions. They refer implicitly to institutional effects on wages. The institutional dimension of w age formation has been extensively explored by scholars of industrial relations. Industrial relations research has found that unions reduce wage inequality within unionized sectors. This effect is attributed to unions’ policy to promote standard rates and thereby create greater uniformity within and across firms (Reynolds and Taft 1956, Freeman 1980, 1982, Lemieux 1998). Unions favour standard rates and seniority based progression in order to prevent subjective judgement on the basis of personal characteristics (Freeman 1980, 1982). This compresses w age structure and w ould favour less skilled w orkers relative to their more skilled counterparts. Thus, unions’ policy to standardize w ages disconnects pay from actual productivity. How ever, the effects found may be due to (self-)selection rather than to policies of unions per se (Kahn 2000, Card et al. 2007.): unions jobs are less attractive for highly skilled workers than non-unions jobs, since their higher productivity due to unobserved skills (i.e. not measured in terms of educational level) is not rewarded in terms of higher w ages. Therefore, highly educated workers with high unobserved skills have more to gain in a competitive non-union labour market, w hereas the highly educated w orkers with low unobserved skills will be more inclined to accept a union job. This selection effect or unobserved heterogeneity-effect may be an alternative explanation for the observed flattening effect in unionized sectors. Research on the effects of institutional factors simply uses union coverage as a predictor for w age fixation. How ever, such an indicator can not distinguish flattening effects result ing from explicit union policies from selection. In order to make such a distinction, w e need to examine the actual clauses of collective agreements, in particular the different levels of w age rate standardization that are specified. If w age levels are influenced by union policies, we should observe strong differences in w age patterns across these different collective agreement clauses. We expect that a w age compression effect of collective agreement clauses that standardize w age rates (proposition I). When standardization of wages in collective agreements is based on education as an objective indicator for productivity, w e expect w age penalties to decrease: equally educated school leavers will be more equally paid (proposition II).