Collusion and reciprocity in infinitely repeated games

This paper extends the standard industrial organization models of repeated interaction between firms by incorporating preferences for reciprocity. A reciprocal firm responds to unkind behavior of rivals with unkind actions (destructive reciprocity), while at the same time, it responds to kind behavior of rivals with kind actions (constructive reciprocity). The main finding of the paper is that, for plausible perceptions of fairness, preferences for reciprocity facilitate collusion in infinitely repeated market games, that is, the critical discount rate at wish collusion can be sustained tends to be lower when firms have preferences for reciprocity than when firms are selfish. The paper also finds that the best collusive outcome that can be sustained in the infinitely repeated Cournot game with reciprocal firms is worse for consumers than the best collusive outcome that can be sustained in the infinitely repeated Cournot game with selfish firms.

Repeated games played in a network

Delayed perfect monitoring in an infinitely repeated discounted game is modelled by letting the players form a connected and undirected network. Players observe their immediate neighbors' behavior only, but communicate over time the repeated game's history truthfully throughout the network. The Folk Theorem extends to this setup, although for a range of discount factors strictly below 1, the set of sequential equilibria and the corresponding payoff set may be reduced. A general class of games is analyzed without imposing restrictions on the dimensionality of the payoff space. This and the bilateral communication structure allow for limited results under strategic communication only. As a by-product this model produces a network result; namely, the level of cooperation in this setup depends on the network's diameter, and not on its clustering coefficient as in other models.

The Limits of Discrete Time Repeated Games:Some Notes and Comments

This paper studies the limits of discrete time repeated games with public monitoring. We solve and characterize the Abreu, Milgrom and Pearce (1991) problem. We found that for the "bad" ("good") news model the lower (higher) magnitude events suggest cooperation, i.e., zero punishment probability, while the highrt (lower) magnitude events suggest defection, i.e., punishment with probability one. Public correlation is used to connect these two sets of signals and to make the enforceability to bind. The dynamic and limit behavior of the punishment probabilities for variations in ... (the discount rate) and ... (the time interval) are characterized, as well as the limit payo¤s for all these scenarios (We also introduce uncertainty in the time domain). The obtained ... limits are to the best of my knowledge, new. The obtained ... limits coincide with Fudenberg and Levine (2007) and Fudenberg and Olszewski (2011), with the exception that we clearly state the precise informational conditions that cause the limit to converge from above, to converge from below or to degenerate. JEL: C73, D82, D86. KEYWORDS: Repeated Games, Frequent Monitoring, Random Pub- lic Monitoring, Moral Hazard, Stochastic Processes.

Some Notes and Comments on the Efficient use of Information in Repeated Games with Poisson Signals

In the present paper we characterize the optimal use of Poisson signals to establish incentives in the "bad" and "good" news models of Abreu et al. [1]. In the former, for small time intervals the signals' quality is high and we observe a "selective" use of information; otherwise there is a "mass" use. In the latter, for small time intervals the signals' quality is low and we observe a "fine" use of information; otherwise there is a "non-selective" use. JEL: C73, D82, D86. KEYWORDS: Repeated Games, Frequent Monitoring, Public Monitoring, Infor- mation Characteristics.

A Review on Repeated Games and Reputations with Incomplete Information

In this project we review the effects of reputation within the context of game theory. This is done through a study of two key papers. First, we examine a paper from Fudenberg and Levine: Reputation and Equilibrium Selection in Games with a Patient Player (1989). We add to this a review Gossner’s Simple Bounds on the Value of a Reputation (2011). We look specifically at scenarios in which a long-run player faces a series of short-run opponents, and how the former may develop a reputation. In turn, we show how reputation leads directly to both lower and upper bounds on the long-run player’s payoffs.

A Review on Repeated Games and Reputations with Incomplete Information

In this project we review the effects of reputation within the context of game theory. This is done through a study of two key papers. First, we examine a paper from Fudenberg and Levine: Reputation and Equilibrium Selection in Games with a Patient Player (1989). We add to this a review Gossner’s Simple Bounds on the Value of a Reputation (2011). We look specifically at scenarios in which a long-run player faces a series of short-run opponents, and how the former may develop a reputation. In turn, we show how reputation leads directly to both lower and upper bounds on the long-run player’s payoffs.

A Folk Theorem for Games when Frequent Monitoring Decreases Noise

This paper studies frequent monitoring in an infinitely repeated game with imperfect public information and discounting, where players observe the state of a continuous time Brownian process at moments in time of length _. It shows that a limit folk theorem can be achieved with imperfect public monitoring when players monitor each other at the highest frequency, i.e., _. The approach assumes that the expected joint output depends exclusively on the action profile simultaneously and privately decided by the players at the beginning of each period of the game, but not on _. The strong decreasing effect on the expected immediate gains from deviation when the interval between actions shrinks, and the associated increase precision of the public signals, make the result possible in the limit. JEL: C72/73, D82, L20. KEYWORDS: Repeated Games, Frequent Monitoring, Public Monitoring, Brownian Motion.

A model for the evolution of reinforcement learning in fluctuating games

Many species are able to learn to associate behaviours with rewards as this gives fitness advantages in changing environments. Social interactions between population members may, however, require more cognitive abilities than simple trial-and-error learning, in particular the capacity to make accurate hypotheses about the material payoff consequences of alternative action combinations. It is unclear in this context whether natural selection necessarily favours individuals to use information about payoffs associated with nontried actions (hypothetical payoffs), as opposed to simple reinforcement of realized payoff. Here, we develop an evolutionary model in which individuals are genetically determined to use either trial-and-error learning or learning based on hypothetical reinforcements, and ask what is the evolutionarily stable learning rule under pairwise symmetric two-action stochastic repeated games played over the individual's lifetime. We analyse through stochastic approximation theory and simulations the learning dynamics on the behavioural timescale, and derive conditions where trial-and-error learning outcompetes hypothetical reinforcement learning on the evolutionary timescale. This occurs in particular under repeated cooperative interactions with the same partner. By contrast, we find that hypothetical reinforcement learners tend to be favoured under random interactions, but stable polymorphisms can also obtain where trial-and-error learners are maintained at a low frequency. We conclude that specific game structures can select for trial-and-error learning even in the absence of costs of cognition, which illustrates that cost-free increased cognition can be counterselected under social interactions.

On Limits to the Use of Linear Markov Strategies in Common Property Natural Resource Games

We derive conditions that must be satisfied by the primitives of the problem in order for an equilibrium in linear Markov strategies to exist in some common property natural resource differential games. These conditions impose restrictions on the admissible form of the natural growth function, given a benefit function, or on the admissible form of the benefit function, given a natural growth function.

Three essays on game theory and bargaining

A PhD Dissertation, presented as part of the requirements for the Degree of Doctor of Philosophy from the NOVA - School of Business and Economics

Penalizing Cartels: The Case for Basing Penalties on Price Overcharge

In this paper we set out the welfare economics based case for imposing cartel penalties on the cartel overcharge rather than on the more conventional bases of revenue or profits (illegal gains). To do this we undertake a systematic comparison of a penalty based on the cartel overcharge with three other penalty regimes: fixed penalties; penalties based on revenue, and penalties based on profits. Our analysis is the first to compare these regimes in terms of their impact on both (i) the prices charged by those cartels that do form; and (ii) the number of stable cartels that form (deterrence). We show that the class of penalties based on profits is identical to the class of fixed penalties in all welfare-relevant respects. For the other three types of penalty we show that, for those cartels that do form, penalties based on the overcharge produce lower prices than those based on profit) while penalties based on revenue produce the highest prices. Further, in conjunction with the above result, our analysis of cartel stability (and thus deterrence), shows that penalties based on the overcharge out-perform those based on profits, which in turn out-perform those based on revenue in terms of their impact on each of the following welfare criteria: (a) average overcharge; (b) average consumer surplus; (c) average total welfare.

Public Monitoring with Uncertainty in the Time Repetitions

This paper study repeated games where the time repetitions of the stage game are not known or controlled by the players. We call this feature random monitoring. Kawamori's (2004) shows that perfect random monitoring is always better than the canonical case. Surprisingly, when the monitoring is public, the result is less clear-cut and does not generalize in a straightforward way. Unless the public signals are sufficiently informative about player's actions and/or players are patient enough. In addition to a discount effect, that tends to consistently favor the provision of incentives, we found an information effect, associated with the time uncertainty on the distribution of public signals. Whether payoff improvements are or not possible, depends crucially on the direction and strength of these effects. JEL: C73, D82, D86. KEYWORDS: Repeated Games, Frequent Monitoring, Random Public Monitoring, Moral Hazard, Stochastic Processes.

Stochastic uncoupled dynamics and Nash equilibrium

In this paper we consider dynamic processes, in repeated games, that are subject to the natural informational restriction of uncoupledness. We study the almost sure convergence to Nash equilibria, and present a number of possibility and impossibility results. Basically, we show that if in addition to random moves some recall is introduced, then successful search procedures that are uncoupled can be devised. In particular, to get almost sure convergence to pure Nash equilibria when these exist, it su±ces to recall the last two periods of play.

Tacit Collusion under Fairness and Reciprocity

This paper departs from the standard profit-maximizing model of firm behavior by assuming that firms are motivated in part by personal animosity-or respect-towards their competitors. A reciprocal firm responds to unkind behavior of rivals with unkind actions (negative reciprocity), while at the same time, it responds to kind behavior of rivals with kind actions (positive reciprocity). We find that collusion is easier to sustain when firms have a concern for reciprocity towards competing firms provided that they consider collusive prices to be kind and punishment prices to be unkind. Thus, reciprocity concerns among firms can have adverse welfare consequences for consumers.

The Ghost in the Machine: Inferring Machine-Based Strategies from Observed Behavior

We introduce a procedure to infer the repeated-game strategies that generate actions in experimental choice data. We apply the technique to set of experiments where human subjects play a repeated Prisoner's Dilemma. The technique suggests that two types of strategies underly the data.