Changing the unemployment insurance duration: heterogeneous effects and an unbudging exit spike

I study the effects of the maximum duration of unemployment insurance benefts on unemployment. I find highly heterogeneous effects across two consecutive reforms in Finland. Both reforms cut the maximum duration by 20 weeks. The first reform targeted those with short work histories and did not affect their mean time in unemployment. The second reform applied to everyone and reduced the mean duration by three weeks. The first reform also did not move the spike in job-finding rates: it stayed at the old maximum duration. The second reform moved the spike to the new exhaustion point, also among the short-history group. This difference can be explained by the first reform's unique implementation, which separated the time when insurance ends from when individuals must switch benefits agencies. This switch causes observable frictions, which appear to be an essential driver for the observed spike. I present how the job-finding rates can be adjusted for observed frictions. Adjusting the job-finding rates for observed frictions substantially flattens the spike.