A Cohort Analysis of Wage Trends

Abstract

Many commentators view the wage structure in West Germany as particularly stable or even as becoming more compressed over time. The descriptive evidence in the previous chapter shows that such a simple view is not warranted in light of the fairly complex trends during the time period from 1975 to 1990. However, mere descriptive evidence on trends in wage dispersion provides only a limited basis for an evaluation of the two-sides-of-the-same-coin hypothesis that the lack of wage flexibility in West Germany has resulted in a disproportionate increase of unemployment among low-skilled workers. This is true because of prevalent composition and selection effects which are likely to influence aggregate wage measures. In this context, a typical composition effect could be that the share of low-skilled workers at older age increases over time, since younger cohorts tend to obtain higher education levels. Assuming an age-earnings profile with positive slope reflecting increased productivity at higher age, average wages for lows-killed workers might exhibit a disproportionate increase relative to other skill groups even though wage ratios between equally productive workers (after controlling for age) have not changed. This example can also be used for an illustration of two possible selection effects. On the one hand, if it is the case that abilities are equally distributed across generations and education as an investment in human capital is positively correlated with unobserved ability, a trend towards increasing education levels among the younger cohorts due, for instance, to an exogenous reduction in the cost of education is likely to imply that the average ability level among the low-skilled workers (if a low-skill level is measured by educational attainment) is decreasing for younger cohorts. On the other hand, a second selection effect emerges, if a lack of wage flexibility or an ongoing wage compression fosters a rise in unemployment among low-skilled workers such that the ability distribution among the low-skilled workers improves. This is the case if the likelihood of becoming unemployed is negatively correlated with the unobserved ability level.