Please use this identifier to cite or link to this item:
Leoni, Thomas
Year of Publication: 
Series/Report no.: 
WIFO Working Papers No. 372
Unemployed workers suffer from poor health conditions, a fact which is documented by a large number of studies covering objective health measures, satisfaction with health status and mortality. This paper contributes to the literature with an empirical analysis of sick leave micro-data from Austrian social insurance agencies. The data represent an interesting source of information because in Austria both employed and unemployed workers are entitled to sickness benefits and both groups are subject to almost identical sick pay regulations. Aggregate statistics show that the unemployed spend close to 9 percent of their time on sick leave, against an average of 3.4 percent for the employed. Further evidence indicates that they report much longer illness spells and a higher number of hospitalisations. Both selection and causation effects can help to understand this large gap in health outcomes. Workers who become unemployed had markedly higher absence rates in employment than fellow workers who stay in employment. This difference, which can be interpreted as an approximation for the selection effect, accounts for roughly half of the observed gap in sick leave rates between the employed and the unemployed. On the other hand there exists a positive albeit non-linear relationship between sick leave and unemployment duration, corroborating the view that unemployment impacts health negatively. In accordance with previous studies I find that the unemployed suffer very often from mental disorders. Although women have a higher incidence of mental disorders than men in both employment and unemployment, it is unemployed men who experience the sharpest increase in mental problems in the wake of unemployment.
sick leave
social insurance data
Document Type: 
Working Paper

Files in This Item:
888.41 kB

Items in EconStor are protected by copyright, with all rights reserved, unless otherwise indicated.