Causal effects of Economic Incentives on Absence from Work: A Duration Analysis Using Fixed Effects

The incidence and duration of work-absence spells for a sample of Swedish blue-collar workers in 1990 and 1991 are analyzed using the Kaplan-Meier non-parametric estimator, discrete-time hazard regression as well as stratified Cox regression. We focus on the effect of economic incentives, i.e., costs associated with absence from work and individual income, as well as gender differences in work absence behavior. In this context we analyze the effect of a reform of the compulsory sickness insurance scheme, i.e., a decrease in the replacement level, that took place in Sweden in March 1991. Special attention is devoted to gender differences in work-absence behavior. We find that it is important to consider unobserved heterogeneity and state dependence in the estimations. The results also show that economic incentives seem to matter for the incidence of work-absence. Gender differences in the way economic incentives affect work-absence behavior, rather than differences in the measured variables, appear to be the predominant factor in explaining the higher work-absence rate of females.