This paper implements a simple Bayesian approach to quantitatively study the Hansen and Imrohoroglu (1992) economy, a calibrated GE model with uninsurable employment risk, designed to assess the optimal replacement rate for a public Unemployment Insurance scheme. The results of this sensitivity analysis are consistent with the original findings, but with several caveats. One novel result in particular is that the distribution of the optimal UI is bimodal. Depending on the calibrated parameters, the optimal UI is in one of two regimes: a very generous scheme with high replacement ratios, where insurance is mainly provided by the UI scheme, or one with low replacement ratios, where insurance is mainly achieved through self-insurance. Even in the absence of moral hazard, it is never optimal to provide full insurance. Moreover, for many plausible parameters' configurations, the welfare maximizing replacement rate does not decrease with the level of MH. The qualitative patterns and quantitative findings are not altered substantially when considering an enlarged labor force, which includes the marginally attached workers. Finally, the parameters representing the hours worked, the leisure share in the households' consumption bundle, and the intertemporal elasticity of substitution have a first order impact on the average welfare. The determination of the optimal UI scheme depends heavily on them. This finding suggests that extra caution should be paid when calibrating these parameters in similar environments.
Calibration methods Unemployment Risk Optimal Unemployment Insurance Heterogeneous Agents Incomplete Markets Computable General Equilibrium Monte Carlo