The Welfare Effects of Conditioning the Potential Duration of UI Benefits on Prior Work History

Most unemployment insurance (UI) systems condition the potential duration of benefits on claimants' work history, yet little evidence exists on whether doing so is socially optimal. We study this question using 25 regression discontinuities in Canada's UI system, where benefit duration increases discretely when prior hours worked crosses various thresholds. Using administrative records covering the universe of UI claims from 1997 to 2018, we estimate the fiscal effects of extending potential duration across the work-history distribution. On average, an additional week lengthens unemployment by 0.30 weeks and costs the government $96. However, this average effect masks significant heterogeneity: the mechanical cost falls steeply with prior hours, while the behavioral cost increases modestly. Consequently, the fiscal externality of extending UI rises with labor market attachment. Embedding these estimates in a sufficient statistics framework, we find that the fiscal externality of a uniform extension is 76% larger than that of an extension targeted at claimants with short histories. Moreover, aggregate welfare can be raised at zero fiscal cost by redistributing benefits from high to low hour claimants.