This paper studies how Unemployment Insurance (UI) affects labor market outcomes through the incentives of employed workers to exert effort on the job, in addition to the well-documented effects on unemployed workers’ search behavior. Using US data, I find that higher UI benefits significantly increase the likelihood of job separations due to discharge or layoff. I rationalize this finding and quantify its policy implications in a dynamic model that features hidden effort on the job as well as hidden job search effort. In this environment, long-term wage contracts arise endogenously to mitigate moral hazard on the job, prescribing wage cuts and separations when firms receive signals of low effort. In a naive calibration that ignores on-the-job moral hazard, doubling the replacement rate reduces aggregate output by only 0.6% and would appear close to optimal. In the benchmark calibration, disciplined by my empirical estimate of the firing elasticity and observed wage-cut patterns, the generosity of the current US system is already roughly optimal, and doubling the replacement rate would reduce aggregate output by 4.8%. These results highlight the importance of considering the optimal responses of all workers, not just the unemployed, when evaluating UI policy.