Reducing Entitlements in Sickness Insurance - Evidence from a Danish Natural Experiment

The share of sick-listed workers and workers on (temporary) disability has increased considerably in recent decades, with substantial costs to firms and public finances. Of major concern, of course, is the issue of moral hazard in take-up and, more generally, the role of replacement rates on overall absences from work. Suggestions have been made to shift part of the burden of monitoring and accommodating disabled workers from the general tax-payer to the employers through financial incentives to accommodate workplaces and minimize movements out of the labor force and into the benefit schemes, see Autor & Duggan (2010). Empirical results are sparse, however.

This paper utilizes a major reform in the Danish sickness and temporary disability insurance system implemented in the summer of 2014. Employers receive compensation from the government when workers take-up temporary disability but with the reform, employers faced a lower compensation for workers on temporary disability spells that extended beyond 26 weeks. The drop ranged from 9% of compensation for workers with dependent children to 28% for workers without. As such, this policy change was in line with the suggestions from Autor & Duggan (2010). We evaluate the effects of this natural experiment using the universe of workers on sickness and temporary disability insurance from 2011 until 2016. In practice, we consider both the duration of absences and subsequent employment spells in a multi-state, competing risk duration model with unobserved heterogeneity. We detect an increase in the hazard back to work of around 40% around the time of the reduction in compensation (weeks 26-30), indicative of substantial moral hazard in take-up. We find no evidence of ex ante threat effects of the reform: the reduction did not appear to affect the hazard back to work prior to week 26 of temporary disability spells. Neither do we find evidence of effects on subsequent employment spells. Finally, we document considerable heterogeneity in the magnitude of the response depending on the health reason behind the reduction in work capacity.