While the German wage structure has long been considered relatively stable at lower percentiles, the past two decades have seen a clear tendency towards more earnings inequality at the bottom end of the earnings distribution as well as declining wage mobility. As a consequence, the low-wage sector has increasingly grown in importance.

Two competing explanations may be relevant for the observed phenomenon. On the one hand, a larger degree of persistence in terms of “true" or “genuine" state dependence of low-wage employment may exist, which occurs if low-wage employment today causes low-wage employment in the future for reasons of stigmatization or human capital depreciation. Alternatively, a larger degree of persistence may be the result of a more unfavorable composition of the low-wage sector. The overall aim of our analysis is therefore to explore to what extent the observed decline in (West) German low-wage mobility reflects a rise in “true" state dependence by distinguishing the evolution of genuine state dependence from pure composition effects. We do so for the years 1984 to 2004 by using the administrative data set of the IAB Employment Subsample (IABS) which is a two per cent subsample of the German Employment Statistics Register.

The descriptive statistics confirm the common trend of an increasing low-wage sector in Germany since the mid-1990s. With respect to the degree of low-wage persistence, we find a distinct increase in the share of individuals staying in the low-wage sector. We see, however, that part of this increase is due to a large shift of compositional factors. The proportion of young male low paid workers, e.g., who face a much smaller risk of sticking to a low wage than older employees, fell from 52% in 1984 to 24% in 2002. This illustrates the necessity to econometrically filter out the compositional factors by taking into account all observable characteristics that affect the extent of true state dependence.

To obtain a measure of genuine state dependence, we make use of a trivariate probit model that accounts for the selection into low-wage employment and earnings retention. Our analysis shows that the extent of genuine state dependence among low-paid workers has increased most notably between 1987 and 1995. The results of a decomposition analysis illustrate that up to 54 per cent of the increase in genuine state dependence during the 1990s can be attributed to compositional shifts towards more unfavorable observable characteristics among the low paid. Our results therefore lend strong support to the notion that appropriate policy interventions should aim at working against such compositional shifts by, e.g., improving low-wage earners' skills and intensifying older low paid employees' vocational training opportunities.


Wage Mobility, Trivariate Probit, Administrative Data