On the basis of aggregate data for the early 1990s, we analyse the determinants of unemployment duration for laid-off male workers in Wallonia (Belgium). Our results indicate that if ranking in recruitment occurs, the standard mixed proportional hazard specification can be too restrictive, leading to an overstatement of the extent of true negative duration dependence. We conclude that negative duration dependence is largely spurious. We also decompose the time variation of the hazard in (unobserved) compositional and direct cyclical and seasonal effects. We find counter-cyclical variation in the quality of young workers, but none for the prime aged.
