Learning to Play Limited Forecast Equilibria

This paper provides a learning justification for limited forecast equilibria, i.e., strategy profiles such that (1) players choose their actions in order to maximize the discounted average payoff over their horizon of foresight as given by their forecasts and (2) forecasts are correct on and off the equilibrium path. The limited forecast equilibria appear to be the stochastically stable outcomes of a simple learning process involving (vanishing) trembles.Journal of Economic LiteratureClassification Numbers: C72, D83.     

The evolution of conventions under incomplete information

Summary. We formulate an evolutionary learning process with trembles for static games of incomplete information. For many games, if the amount of trembling is small, play will be in accordance with the games (strict) Bayesian equilibria most of the time. This supports the notion of Bayesian equilibrium. Often the process will select a specific equilibrium. We study an extension to incomplete information of the prototype conflict known as Chicken and find that the equilibrium selection by evolutionary learning may well be in favor of inefficient Bayesian equilibria where some types of players fail to coordinate.Received: 17 March 2003, Revised: 3 December 2003, JEL Classification Numbers: C72.     

Symmetric Separating Equilibria in English Auctions

We characterize the set of perfect Bayesian equilibria in symmetric separating strategies in the model of English auctions given by P. R. Milgrom and R. J. Weber (1982, Econometrica50, 1089–1122). There is a continuum of such equilibria. The equilibrium derived by Milgrom and Weber is that in which bids are maximal. Only in the case of pure private values does a restriction to weakly undominated strategies select a unique equilibrium. This has important implications for empirical studies of English auctions, particularly outside the pure private values paradigm. Journal of Economic Literature Classification Numbers: D44, D82.     

Learning to collude: An experiment in convergence and equilibrium selection in oligopoly

The paper considers a simple oligopoly model where firms know their own and the average pay-off in the industry. Firms choose decision rules for trading. The theory predicts that there are three types of Nash equilibria in this game (collusive, Cournot and Stackelberg). Our experiments test the selection process. We find that there is clear evidence of convergence to an equilibrium, and whilst both Cournot and collusive outcomes were selected, the collusive equilibrium is more common. The experimental results also give insights into the process of individual learning, confirming that subjects follow aspiration rules rather than reinforcement rules.     

Starting Small and Commitment

I study a model of a long-term partnership with two-sided incomplete information. The partners jointly determine the stakes of their relationship and individually decide whether to cooperate with or betray each other over time. I characterize the extremal—interim incentive efficient—equilibria. In these equilibria, the partners generally “start small,” and the level of interaction grows over time. The types of players separate quickly. Further, cooperation between “good” types is viable regardless of how pessimistic the players are about each other initially. The quick nature of separation in an extremal equilibrium contrasts with the outcome selected by a strong renegotiation criterion (as studied in Watson (1999, J. Econ. Theory85, 52–90). Journal of Economic Literature Classification Numbers: C72, C73, D74.     

Markov rational expectations equilibria in an overlapping generations model

In this paper, we analyze rational expectations equilibrium paths in a stochastic overlapping generations model. The work presented here builds on results of S. E. Spear (J. Econ. Theory 35 (1985), 251–275), where is is shown that in a model with multiple goods and time non-separable preferences, a stochastic steady state equilibrium will generically fail to exist. A stochastic steady state is defined as an equilibrium in which the stochastic process of endogenously determined variables is measure isomorphic to the exogenous process driving the model. In this paper, we establish the existence of non-steady state equilibria and provide a characterization of their stochastic properties.     

How Communication Improves Efficiency in Bargaining Games

We study a double auction with two-sided private information and preplay communication, for which Myerson and Satterthwaite (1983, J. Econ. Theory28, 265–281) showed that all equilibria are inefficient and the Chatterjee–Samuleson linear equilibrium is most efficient. Like several others, we find that players use communication to surpass equilibrium levels of efficiency, especially when the communication is face-to-face. Our main contribution is an analysis of how communication helps the parties achieve such high levels of efficiency. We find that when preplay communication is allowed, efficiency above equilibrium levels is a result of what we call “dyadic” strategies that allow the parties to coordinate on a single price that reflects both parties' valuations. Journal of Economic Literature Classification Numbers: C78, D82.     

Sequential Equilibrium and Competition in a Diamond–Dybvig Banking Model

Within the framework of a Diamond–Dybvig model [J. Polit. Econ.91(1983), 401–419], but with explicitly modelling the autarky choice during the planning period, we demonstrate that a mixed strategy bank run equilibrium that does not rely on sunspots may coexist with the sunspot run equilibrium previously studied in the literature. In a version of the model with multiple banks, there exist sequential equilibria that imply positive profits. However, the zero-profit contract in which runs never occur can be supported as the unique equilibrium outcome if the agents play pure strategies only and their beliefs are restricted to be consistennt with a forward induction argument.Journal of Economic LiteratureClassification Numbers: C72, G21     

Government policy and the probability of coordination failures

This paper introduces an approach to the study of optimal government policy in economies characterized by a coordination problem and multiple equilibria. Such models are often criticized as not being useful for policy analysis because they fail to assign a unique prediction to each possible policy choice. We employ a selection mechanism that assigns, ex ante, a probability to each equilibrium indicating how likely it is to obtain. We show how such a mechanism can be derived as the natural result of an adaptive learning process. This approach leads to a well-defined optimal policy problem, and has important implications for the conduct of government policy. We illustrate these implications using a simple model of technology adoption under network externalities.     

Implementation,Elimination of Weakly Dominated Strategies and Evolutionary Dynamics

This paper studies convergence and stability properties of T. Sjöström's (1994, Games Econom. Behav.6, 502–511) mechanism, under the assumption that boundedly rational players find their way to equilibrium using monotonic evolutionary dynamics and best-reply dynamics. This mechanism implements most social choice functions in economic environments using as a solution concept one round of deletion of weakly dominated strategies and one round of deletion of strictly dominated strategies. However, there are other sets of Nash equilibria, whose payoffs may be very different from those desired by the social choice function. With monotonic dynamics, all these sets of equilibria contain limit points of the evolutionary dynamics. Furthermore, even if the dynamics converge to the “right” set of equilibria (i.e., the one which contains the solution of the mechanism), it may converge to an equilibrium which is worse in welfare terms. In contrast with this result, any interior solution of the best-reply dynamics converges to the equilibrium whose outcome the planner desires. Journal of Economic Literature Classification Numbers: C72, D70, D78.     

Legislative Bargaining and Coalition Formation

The finite horizon version of D. P. Baron and J. Ferejohn's [1989, Amer. Polit. Sci. Rev.83, 1181-1206] legislative bargaining model is investigated. With three or more periods, a continuum of divisions is supportable as subgame perfect equilib- ria. There exist equilibria where coalitions larger than a minimal winning coalition receive strictly positive shares. With sufficiently patient players and a sufficiently long horizon, any interior distribution is supportable as an equilibrium. In contrast, a generic uniqueness result applies when introducing heterogenous time preferences. The unique backwards induction equilibrium in the perturbed game is nonstationary, and neither the original (symmetric) nor the perturbed game provides guidance for equilibrium selection in the infinite game. Journal of Economic Literature Classification Numbers: C73, C78, D72, D78, H49.     

Violence against women, social learning, and deterrence

We develop a simple model of sexual and domestic violence. By assumption, the potential victim’s threat to report if she is victimized is not credible, which implies that the only sequential equilibrium involves violence. However, a realistic social learning process converges to a non-sequential equilibrium without violence from all nearby states if the expected punishment for offenders whose victims report to the police is sufficiently high. A policy to increase the sentences for sexual and domestic violence convictions could therefore substantially reduce such violence in the long run, even if it is powerless to make women’s threats to report credible.

How Proper Is Sequential Equilibrium?

A strategy profile of a normal form game is proper if and only if it is quasi-perfect in every extensive form (with that normal form). Thus, properness requires optimality along a sequence of supporting trembles, while sequentiality only requires optimality in the limit. A decision-theoretic implementation of sequential rationality, strategic independence respecting equilibrium (SIRE), is defined and compared to proper equilibrium, using lexicographic probability systems. Finally, we give tremble-based characterizations, which do not involve structural features of the game, of the rankings of strategies that underlie proper equilibrium and SIRE.Journal of Economic LiteratureClassification Numbers: C70, C72.     

Non-Additive Beliefs and Strategic Equilibria

This paper studies n-player games where players' beliefs about their opponents' behaviour are modelled as non-additive probabilities. The concept of an “equilibrium under uncertainty” which is introduced in this paper extends the equilibrium notion of Dow and Werlang (1994, J. Econom. Theory64, 305–324) to n-player games in strategic form. Existence of such an equilibrium is demonstrated under usual conditions. For low degrees of ambiguity, equilibria under uncertainty approximate Nash equilibria. At the other extreme, with a low degree of confidence, maximin equilibria appear. Finally, robustness against a lack of confidence may be viewed as a refinement for Nash equilibria. Journal of Economic Literature Classification Numbers: C72, D81.     

A dynamic model of expectations equilibrium

This paper describes a dynamic information adjustment process which achieves (rational) expectations equilibria for stochastic exchange environments. An informational temporary equilibrium is an exchange equilibrium in which agents' expectations are conditioned on their initial information and market data generated in previous informational temporary equilibria. An equilibrium is a temporary equilibrium which reveals no further information. This process leads to an equilibrium even when the data observed by agents is insufficient to permit the existence of an expectations equilibrium.     

On the beliefs off the path: Equilibrium refinement due to quantal response and level-k

The extensive form game we study has multiple perfect equilibria, but it has a unique limiting logit equilibrium (QRE) and a unique level-k prediction as k approaches infinity. The convergence paths of QRE and level-k are different, but they converge to the same limit point. We analyze whether subjects adapt beliefs when gaining experience, and if so whether they take the QRE or the level-k learning path. We estimate transitions between level-k and QRE belief rules using Markov-switching rule learning models. The analysis reveals that subjects take the level-k learning path and that they advance gradually, switching from level 1 to 2, from level 2 to equilibrium, and reverting to level 1 after observing opponents deviating from equilibrium. The steady state therefore contains a mixture of behavioral rules: levels 0, 1, 2, and equilibrium with weights of 2.9%, 16.6%, 37.9%, and 42.6%, respectively.     

Capital as a commitment: Strategic investment to deter mobility

This paper analyzes how an early entrant in a market can exploit its head start by strategic investment. The analysis is based on Spence's paper, Investment strategy and growth in a new market, (Bell J. Econ., 10 (1979), 1–19). We frist study the investment game in the no-discounting case, which embodies the key features of mobility deterrence. We establish the existence of a set of perfect equilibria and suggest that one particular equilibrium is most reasonable. This equilibrium, also valid with discounting, involves the follower firm being forever deterred from investing to its steady-state reaction curve, in contrast to Spence's proposed solution.     

Adaptive learning in extensive form games and sequential equilibrium

Summary. This paper studies adaptive learning in extensive form games and provides conditions for convergence points of adaptive learning to be sequential equilibria. Precisely, we present a set of conditions on learning sequences such that an assessment is a sequential equilibrium if and only if there is a learning sequence fulfilling the conditions, which leads to the assessment. Received: November 5, 1996; revised version: May 28, 1997     

Equilibrium selection in common-value second-price auctions

This note considers equilibrium selection in common-value second-price auctions with two bidders. We show that for each ex post equilibrium in continuous and undominated strategies, a sequence of “almost common-value” auctions can be constructed such that each of them possesses a unique undominated and continuous equilibrium and the corresponding sequence of equilibria converges to that ex post equilibrium. As an implication, no equilibrium selection of this model based on perturbations seems to be more convincing than others.     

Subgame perfect reaction function equilibria in discounted duopoly supergames are trivial

This paper is concerned with infinitely repeated duopoly games with discounting. A question which has been open since Friedman's (Rev. Econ. Stud. 35 (1968), 257–272) reaction function article is settled for a general class of games. The question is wheter nontrivial reaction function equilibria can be subgame perfect. This question is answered in the negative. Such equilibria must be trivial in the sense of prescribing the stage game noncooperative equilibrium actions in every period, independent of prior history.