Abstract
The effect of economic incentives on absenteeism in Sweden is examined. Based on a utility function, frequently used in labour supply studies, absenteeism is modelled as an individual day-to-day decision. Implications of compensating wage differentials and efficiency wage hypotheses are discussed. Two data sets are used: survey data from 1981 and an annual panel 1981-1984. The dependent variable is the number of days absent per year for each individual, i.e. the outcomes of the day-to-day decision are aggregated over the time period of one year. Theory consistent models allowing for state dependence or unobserved heterogeneity are derived. Binomial, overdispersed binomial, mixture distribution and Markov chain models are estimated using maximum likelihood and least squares techniques. The results suggest that changes in the compensation level of the sickness insurance and changes in the unemployment rate could explain about 72 per cent of changes in work absence rate 10 year later.