Principal-agent models take outside options, determining participation and incentive constraints, as given. We construct a general equilibrium model where workers' reservation wages and the maximum punishment acceptable before workers quit are instead determined endogenously. We simultaneously...