A Theory of Partnership Dynamics: Learning, Specific Investment, and Dissolution

Abstract

This paper explores the benefits and drawbacks of potential partnership dissolution through an infinite-period, dynamic game-theoretic model of learning and endogenous dissolution. As partners learn about the quality of their partnership relative to their outside opportunities, the rents associated with the partnership change, effecting a related change in the strangth of incentives to provide effort. The paper develops an incentive-constrained dynamic programming algorithm for the computation of optimal symmetric equilibria of dynamic games with known worst punishments (such as dissolution here). The scheme is much simpler than the more general set-valued approach pioneered by Abreu, Pearce, and Stacchetti in that it only requires the computation of one value function at each iteration. The algorithm is then used to show that rather mild supermodularity conditions lead to effort levels in the optimal equilibria which rise in the expected quality of the partnership.