"Unemployment Benefits and Unemployment in the Great Recession: The Role of Macro Effects" by Marcus Hagedorn, Fatih Karahan, Iourii Manovskii, and Kurt Mitman. October 2013. NBER #19499.
Unemployment following the Great Recession has remained unusually high. One possible reason is the extension of unemployment benefits from their usual 26-week limit to a period as long as 99 weeks. The conventional wisdom is that such extensions have positive effects on the macroeconomy because they have very little effect on labor supply and also increase aggregate demand because unemployed workers have a large marginal propensity to consume any benefits they receive.
This view has been supported by low estimates of the effect of unemployment insurance extensions on labor supply. Economists have used the cross-sectional variation across states in extension initiation and duration to estimate the effect of benefit variation on the search behavior of the subset of the unemployed who receive benefits. For example, in a fall 2011 Brookings Papers on Economic Activity article, Berkeley economics professor Jesse Rothstein compared those unemployed who are eligible for unemployment insurance with those who are not, thus isolating the search behavior of those receiving benefits. He concluded that the increased duration of benefits has a very small effect on the duration of unemployment and concludes that only a small fraction of the increased unemployment in the Great Recession can be attributed to reduced worker job search effort.
The authors of this paper argue that Rothstein's research design underestimates the total effect of unemployment insurance extensions on labor supply because it does not include the indirect effects on labor demand, i.e., the posting of vacancies by firms. The effect of benefit extensions on labor demand arises because the existence of unemployment benefits reduces labor supply, which increases the wage that would-be employers have to offer, which in turn decreases firm expected profits and reduces the posting of vacancies.
To detect this effect on vacancies, one cannot simply regress the increase in benefits on wages, in general, because the wage data would include both the effect of reduced labor supply (less searching) that would increase wages as well as the effect of reduced labor demand (fewer job creations) that would decrease wages. The authors propose instead a regression of difference in wages against difference in benefits across time, but only for those workers who stay on the job. The differences in benefits arise through estimation on contiguous counties on opposite sides of a state border. The authors conclude that a rise in unemployment of 3 percentage points is the predicted result and that "unemployment benefit extensions account for most of the persistently high unemployment after the Great Recession."
PETER VAN DOREN is editor of Regulation and a senior fellow at the Cato Institute.