Projects per year
Using the methodology developed in Alvarez, Borovickova, and Shimer (2014), we decompose the variance of unemployment duration in Spain into three main components: labour market frictions, duration dependence and heterogeneity. Crucial to our analysis are the comparisons by different demographic groups over the business cycle to shed light in the mechanisms behind duration dependence. We offer a general approach for interpreting administrative data that is closer to unemployment than non-employment and includes different types of unemployment spells, particularly short spells without the right to claim benefits. Labour market frictions account for the bulk of the variance of unemployment duration, and become more prominent during recessions. Duration dependence constitutes 25% of the total variance, followed by heterogeneity, which represents 18%. Women and college graduates have larger shares of duration dependence. Finally, the share of duration dependence does not vary over the business cycle for non-college workers. But it is higher during expansions for college workers, which is consistent with a statistical discrimination mechanism due to dynamic sample selection.
- unemployment duration
- administrative social security data
- duration dependence