In light of nonstationary search theory (van den Berg, 1990), this paper estimates the effects of benefit entitlement periods and the size of unemployment ben- efits on unemployment durations and post-unemployment earnings in West Germany. For the unemployment duration, we estimate censored Box-Cox quantile regression, which is robust with respect to the specification of the unobserved error distribution and avoids the common proportional hazard assumption. Our results suggest that the length of benefit entitlement is only of minor importance for the duration of search unemployment and for post unemployment wages. A high wage replacement rate in the low wage sector seem to considerably elongate the duration of unemployment and it is associated with higher post unemployment wages.


Box Cox quantile regression, hazard rate, unemployment, wage