Andrén, Daniela2004-08-312007-02-092007-02-0920041403-2465http://hdl.handle.net/2077/2778Using a sample of 2,789 Swedish residents on working age, this paper analyzes long-term absences from work due to sickness. The database contains all compensated sickness spells in the period January 1986 to December 1991. Earlier studies of work absence due to sickness did not analyze multiple spells of sickness. Moreover, such data requires estimation techniques that were not often used in the previous studies. The analysis is performed using mixed proportional hazard models. The results show that the loss of earnings reduced length of absence, while high regional unemployment increased it. There was more heterogeneity among diagnosis-groups and individual-groups than among regions as groups.31 pages354665 bytesapplication/pdfenlong-term sicknessabsenteeismmultiple spellsunobserved heterogeneity.ReportEconomics