Government Policy and Probabilistic Equilibrium Selection

We study an economy where search frictions create a coordination problem among agents - each agent wants to produce if and only if enough other agents are producing. This environment easily generates multiple Pareto-ranked equilibria. Our interest is in how likely it is that the economy will find its way to each of these equilibria when agents are learning about some fundamental parameters of the economy. Specifically, we study Bayesian learning and show how this process generates a probability distribution over the equilibrium set. We then study a particular type of demand-management policy that the government can use to encourage agents to engage in production. We show that this policy can make it more likely that the economy converges to the Pareto superior equilibrium, but that in the process it reduces the value of being in that equilibrium. Hence a tradeoff arises in this model between the likelihood of attaining a particular equilibrium and the value of being in it. We analyze this tradeoff in the context of a numerical example.