Optimal Unemployment Insurance with Limited Government Commitment and Heterogeneous Agents

In this paper, we consider a model where each one of a continuum of workers can influence the stochastic evolution of his employment/unemployment state, i.e., the number of periods he has been employed or unemployed, respectively, through the choice of an effort variable. We analyze the dynamic game between the workers and a long-lived government who attempts to control social welfare through a combination of income taxation and unemployment insurance, and characterize the sequential equilibrium payoff correspondence when the game is characterized by an initial distribution over employment/unemployment states. In contrast to existing principal-agent models of optimal unemployment insurance, we study equilibrium policies when: (i) the government faces a heterogeneous population of workers; (ii) the government cannot commit to its policy plan; (iii) the policy can be conditioned on the evolution of the distribution of worker states; and (iv) the government faces a budget constraint. These features allow us to characterize the feasible set of equilibrium payoffs for workers as a function of the initial distribution over states, without needing to resort to exogenously fixed promised utilities. We also extend existing algorithms used for the computation of equilibrium value correspondences V : S ⇒ R, where S ⊂ R and n ≤ 3, to the computation of value correspondences V : P ⇒ R, where P ⊂ R, m > 1 and n > 3. << PLEASE DO NOT DISTRIBUTE ONLINE OR OUTSIDE OF WORKSHOP >>