Hidden unemployment in disability insurance in the Netherlands; an empirical analysis based on employer data

In this paper, we construct and estimate a (semi-) structural model, so as to uncover the fraction of hidden unemployment in the Disability Insurance (DI) enrolment rate. For this purpose, we use longitudinal administrative data of Dutch employers for 1994-2003. We find the (average) fraction of hidden unemployment in DI enrolment to amount to about 11%. This corresponds to 2.6% of the ‘true’ unemployment insurance (UI) enrolment rate of employers. Over the years, we observe a strong decrease in this fraction, from 5.4% in 1995, to 0.7% in 2003. In addition, our estimates suggest that most of correlation that is observed between the UI and DI enrolment rates can be explained by substitution effects, and not by ‘true’ correlation between the schemes that is exogenous to the firm. In the model, the fraction of hidden unemployment in the DI scheme is (over-)identified from various restrictions imposed by the data. First, identification follows from exclusion restrictions obtained from the coefficient estimates of variables that are assumed to influence the UI enrolment rate only. For this purpose, we use information on the wage distribution of workers employed at the firms in our sample, and sectoral growth rates. Second, identification of substitution effects follows from the observed correlation between both enrolment rates.