Associate Professor Kurt Mitman

In designing UI, policymakers face the classic insurance-incentive tradeoff. This tradeoff is complicated because unemployed worker’s job search behavior changes with both current and future UI benefits (and economic conditions).

The research provides a dynamic generalization of Baily-Chetty, whereby optimal policy balances the consumption smoothing benefit of UI against its moral hazard cost. The analysis yields three broad insights:

1) Moral hazard cost depends most directly on search efficiency, not on the unemployment rate. Thus, if possible, UI would be indexed to the primitive shocks, not the unemployment rate.

2) The dynamic nature of the environment complicates the standard Baily-Chetty formula because future UI distort current search efforts, hence have a moral hazard cost today. The optimal policy under commitment accounts for this. The optimal policy under discretion does not: it effectively follows a sequence of static Baily-Chetty formulas.

3) The level of unemployment doesn't matter per se, but its level over time does. For example: imagine that unemployment is high today but expected to be low in the future. Under commitment, promising to cut UI when unemployment is low is a cheap way to provide incentives when unemployment is high.

How does this apply to a COVID-like shock? With commitment (Ramsey), a large but short-lived increase in benefits is optimal. Without commitment, the government can't commit to lowering future UI, thus it cannot provide as much insurance today, so implements a smaller more persistent increase.

How can the Ramsey policy be implemented? A simple rule that depends on the change in the moving average of unemployment does a good job at replicating the Ramsey policy. A rule that depends on the level, keeps benefits too high for too long, generating persistently high unemployment.

The research also considers what happens if there's heterogeneity in search efficiency and low-efficiency workers get hit harder by COVID. Benefits come down more slowly to account for dynamic selection, but the main insights follow from before.

The model and shocks are quite stylized, but believed to provide useful insights for thinking about optimal policy design and practical implementation going forward. 

Contact Associate Professor Kurt Mitman.

To read more about Kurt's research, visit his personal web.