Partial and general equilibrium effects of unemployment insurance

Abstract

von Buxhoeveden, M. 2019. Partial and General Equilibrium Effects of Unemployment Insurance. Identification, Estimation and Inference. Economic studies 179. 91 pp. Uppsala: Department of Economics, Uppsala University. ISBN 978-91-506-2750-3.

Essay I: Wage setting models typically posit a tight relationship between the generosity of unemployment insurance (UI) and equilibrium wages. This paper estimates the effect of UI on workers’ wages. I build on a unique feature of the unemployment policy in Sweden, where workers can opt to buy supplement UI coverage above a minimum mandated level. In January 2007, the government sharply increased the price of UI, and the share of workers with supplement coverage fell from 90% to 80%. I exploit variation in the price of UI across industries to measure the effect of industry level UI-coverage on wages. My estimates suggest that a 10 percentage point reduction in the share of workers covered by supplement UI reduce wages by
5%. Since I rely on variation in UI-coverage at the industry level, these estimates contain wage adjustments from collective and individual level bargaining. Finally, I use the estimated UI wage effect to derive bounds on worker bargaining power in a simple DMP model and find that it can be at most 0.12. This evidence supports wage setting mechanisms that tie wages to the generosity of UI.

Essay II: This paper estimates the effects of unemployment insurance (UI) benefits on job finding rates and entry level wages for unemployed high school leavers. Up to year 2007, Swedish high school-students who became unemployed shortly after graduation were entitled to UI-benefits once they became 20 years of age. Therefore, the start of an unemployment spell relative to the 20:th birthday creates potentially exogenous variation in time to treatment.
I exploit this to estimate the effect of UI benefits on unemployment duration and entry level wages. The results show that there is a large and statistically significant negative effect of UI benefits on the employment hazard. There are no detectable effects on entry level wages. This would suggest that unemployment benefits induce high school leavers to postpone labor market
entry but does not seem to effect job match quality.

Essay III: Difference-in-Differences (DID) is a quasi-experimental method to evaluate the effect of a treatment. In the basic version, two groups are observed at two dates. The treatment group becomes treated in the second period. The effect of the intervention is estimated by comparing the change in the outcome experienced by the treatment group to the corresponding
change in the control group. However, assessing the impact of an intervention is often complicated by the well-known problem of sample selection. In randomized experiments, one popular method to address this is to implement Lee (2009) bounds. This paper extends Lee (2009) bounds to the DID design. Identification results, estimators and a simple bootstrap
procedure for computing standard errors are presented.

Keywords: Labor Economics, Unemployment Insurance, Wage Level and Structure, Job Search, Duration Analysis, Econometric and Statistical Methods, Selection Models