In the context of unemployment insurance, it is well known that one should reduce the duration of benefits or the amount of those benefits to elicit the right amount of search effort by unemployed workers. Indeed, being too generous on either dimension allows them to shirk and take advantage from the insurance pool others are paying for. These results are well established in theory, and there empirical evidence that they present. But what works best to get the unemployed back to work remains to be established.
Peter Haan and Victoria Prowse estimate a very rich structural life-cycle model and then use the estimated model for a few interesting experiments. One interesting conclusion they find is that shortening the eligibility period for unemployment insurance benefits works much better than cutting the benefits.
Clearly, such an analysis is far superior to other exercises that perform diff-in-diff cross-sectional reduced-form regressions because the response of workers to policy changes can be highly non-linear. Unfortunately, the model does not allow for savings, and I disagree with the authors that this is trivial. Indeed, when workers have to rely completely on current income for consumption, it is obvious that shortening their eligibility for benefits, even if there is social assistance as a fallback, is going to discipline them more than a revenue neutral reduction in the replacement ratio. It is about standard precautionary savings in the face of idiosyncratic employment risk and risk-averse utility.