The Effect of Changing Mental Health on Unemployment Duration and Destination States after Unemployment

We study how purchase of pharmaceutical products for severe mental illnesses during unemployment affects re-employment and labour market exit probabilities. Within the framework of a multivariate duration model we apply the 'timing-of-events' method, which explicitly makes use of the information that pharmaceutical treatment can begin at different points of time during an unemployment spell. In the absence of instrumental variables this method allows for causal inference in presence of unobserved heterogeneity, but at the cost of strong assumptions. The basis for our analysis is state-of-the-art register-based data, which gives insight on the timing, type, and volume of drug purchase as well as labour market histories for a random sample of the Danish population. We find a significant and strong negative effect of periods with drug treatment on the employment chances. During the treatment with drugs, the job-finding rate is reduced substantially relative to what it would have been in absence of a drug treated mental illness. Importantly, our results not only show that drug treated mental illness prolongs the unemployment duration, but it also increases the labour market exit rate.