Adverse weather conditions that hamper outdoor construction work have been the prime justification for promoting all-season employment in the construction sector in many European countries. The aim is to reduce seasonal layoffs and thereby reduce spending on unemployment compensation by reducing some of the risk that a weather-induced shortfall of work means to both employers and workers. In Germany, recent years have seen a major shift from a system based on mainly publicly funding a weather allowance to a system that promotes the use of working hours accounts to compensate for seasonal fluctuations in the workload. For two of the main employment promotion schemes that can be found across Europe, the regime shifts in Germany thus constitute a prime opportunity for comparing the effectiveness of such measures in preventing seasonal layoffs. In particular, we would expect measures that lower a firm's layoff costs relative to the costs of maintaining an employment relationship during a seasonal labour shortfall to reduce transition rates to unemployment. This paper therefore empirically investigates the determinants of individual layoff probabilities in the German construction sector. Based on daily data of more than twenty subsequent years, our analysis explores the impact of the institutional changes that occurred during the 1990s while taking account of other factors such as weather conditions, the business cycle, and individual level characteristics. We are thus able to disentangle the relevance of each of these factors as a determinant of seasonal unemployment. Our results generally confirm the effects suggested by economic theory, i. e. a higher inflow into unemployment during periods of adverse weather conditions, unfavourable business conditions as well as reduced layo® costs. In particular, the longstanding bad weather allowance that is financed mainly by both the unemployment insurance fund and by employers turns out to significantly increase layoffs compared to the introduction of a winter allowance in addition to the use of working hours accounts. Our results also suggest, however, that actual weather conditions and the legal setup are less relevant than thought by the public as most of the layoffs simply take place at fixed calender times.


panel data, temporary layoffs, employment stability