Jack Robles

Does Evolution Solve the Hold-up Problem?

The paper examines the theoretical foundations of the hold-up problem. At a first stage, one agent decides on the level of a relationship-specific investment. There is no contract, so at a second stage the agent must bargain with a trading partner over the surplus that the investment has generated. We show that the conventional underinvestment result hinges crucially both on the assumed bargaining game and on the choice of equilibrium concept. In particular, we prove the following two results. (i) If bargaining proceeds according to the Nash demand game, any investment level is subgame perfect, but only efficient outcomes are stochastically stable. (ii) If bargaining proceeds according to the ultimatum game (with the trading partner as proposer), only the minimal investment level is subgame perfect, but any investment level is stochastically stable.

Evolution in Finitely Repeated Coordination Games

I apply a version of Kandori et al. (1993, Econometrica, 61, 29–56) and Youngs (1993, Econometrica, 61, 57–84) evolutionary dynamic to finitely repeated coordination games. The dynamic is modified by allowing mutations to affect only off path beliefs. I find that repetition within a match leads agents to sacrifice current payoffs in order to increase payoffs in later stages. As a consequence, evolution leads to (almost) efficiency.

Stochastic Stability in Finitely Repeated Two Player Games

We apply stochastic stability to undiscounted finitely repeated two player games without common interests. We prove an Evolutionary Feasibility Theorem as an analog to the Folk Theorem (Benoit and Krishna, 1985 and 1987). Specifically, we demonstrate that as repetitions go to infinity, the set of stochastically stable equilibrium payoffs converges to the set of individually rational and feasible payoffs. This derivation requires stronger assumptions than the Nash Folk Theorem (Benoit and Krishna, 1987). It is demonstrated that the stochastically stable equilibria are stable as a set, but unstable as individual equilibria. Consequently, the Evolutionary Feasibility Theorem makes no prediction more specific than the entire individually rational and feasible set.