ISER Working Paper Series 2006-37
Do wages compensate for anticipated working time restrictions? Evidence from seasonal employment in Austria
01 Jul 2006
A substantial body of research in labor economics examines the existence of compensating wage differentials. These arise when apparently identical workers earn different wage rates across sectors, employers and even within the same firms. According to the competitive view, these differentials are due to the existence of positive or negative attributes of the job, which are 'offered' by the employer to their workers and whose price is reflected in the wage rate.
Numerous studies have explored compensating wage differentials arising from characteristics of the job such as employer-based pension schemes or flexible working hours. In this paper we examine the existence of compensating wage differentials due to expected or anticipated wage losses, which may occur because of employer-determined working time restrictions.
Previous attempts to quantify the effect of working time restrictions on wages rely on a worker's industry affiliation or his self-reported contract, and are often based on cross-sectional data. This makes it difficult to distinguish industry and individual effects from the true effect of the working time constraints.
In contrast with this approach, we derive a more flexible definition of working time constraints. Specifically, we use the longitudinal information in the Austrian social security records and exploit the pattern of employment and unemployment spells observed for the same individual over time to identify jobs characterized by regular interruptions. Since the Austrian labor market exhibits significant seasonal variation in employment, and seasonal jobs are by definition subject to time constraints, our data provides a natural way to identify anticipated working time restrictions.
Using this approach we can analyse the correlation between wages and working time restrictions (measured by yearly unemployment rates) controlling for industry as well as individual specific effects. We also use variation in the starting month of a job as an instrument to address the potential endogeneity of individual unemployment experience. The idea is that starting a job in different months of the year is a good predictor of the amount of anticipated unemployment which will be observed.
Our results show that while the effect of unemployment on the wage rate is negative in a pooled regression, controlling for individual and industry fixed effects as well as for the endogeneity of unemployment results in a positive impact. We find that the average wage differential for seasonal jobs is about 11% percent of the wage in a permanent job, and that a similar amount is covered by the unemployment insurance system. This implies that employers and workers who operate on the basis of seasonal demand fluctuations receive an indirect subsidy from other firms and other workers.