We consider two-player repeated games, where the players observe their own payoffs with a positive probability. Typically, a player observes neither the other's actions nor her payoff.  We show that knowing her own payoff is sufficient to obtain any strictly efficient payoff  by sequential equilibrium, when costly communication is available and the players are  sufficiently patient.

For a constant ε>0, we prove a poly(N) lower bound on the (randomized) communication complexity of ε-Nash equilibrium in two-player N×N games.For n-player binary-action games we prove an exp(n) lower bound for the (randomized) communication complexity of (ε,ε)-weak approximateNash equilibrium, which is a profile of mixed actions such that at least (1-ε)-fraction of the players are ε-best replying. 

Risks related to events that arrive randomly play important role in many real life decisions, and models of learning and experimentation based on two-armed Poisson bandits addressed several important aspects related to strategic and motivational learning in cases when events arrive at jumps times of the standard Poisson process. At the same time, these models remain mostly abstract theoretical models with few direct economic applications.

A long-standing tradition models legal enforcement as being distinct from community enforcement, whereby individuals follow the law because legal enforcement constrains the actions that players may take or fixes their payoffs from their actions.

We provide an axiomatic characterization of an income segregation index in school districts. One axiom requires that single-school districts be the least segregated of all districts. A second axiom requires that any reorganization of a subdistrict that raises its segregation, raises the districtwide segregation as well. A third axiom requires an intuitive decomposition by subdistricts into within-district and between-district terms.

Koszegi and Rabin’s (2006, 2007, 2009) model of expectations-based reference-dependent preferences offers a unified explanation for a diverse body of evidence across different domains. However, almost a decade of direct lab tests of the model has generated mixed evidence: in only a subset of (what appear to be) similar experimental setups are lagged-probability-beliefs treatments found to affect behavior as (apparently) predicted by the theory. The present paper aims to investigate why.

Online reviews could, in principle, greatly improve the match between consumers and products. However, the authenticity of online user reviews remains a concern; firms have an incentive to manufacture positive reviews for their own products and negative reviews for their rivals. In this paper, we marry the diverse literature on economic subterfuge with the literature on organizational form. We undertake an empirical analysis of promotional reviews, examining both the extent to which fakery occurs and the market conditions that encourage or discourage promotional reviewing activity.

Two  firms produce substitute goods with unknown quality. At each stage the firms set prices and a consumer with private information and unit demand buys from one of the fi rms. Both  firms and consumers see the entire history of prices and purchases. Will such markets aggregate information? Will the superior  rm necessarily prevail? We adapt the classical social learning model by introducing strategic dynamic pricing. We provide necessary and sufficient conditions for learning. In contrast to previous results, learning can occur when signals are bounded.

We study a discounted repeated inspection game with two agents and one principal. Both agents may profit by violating certain rules, while the principal can inspect on at most one agent in each period, inflicting a punishment on an agent who is caught violating the rules. The goal of the principal is to minimize the discounted number of violations, and he has a Stackelberg leader advantage. We characterize the principal's optimal inspection strategy.

We prove that every multiplayer quitting game admits a sunspot epsilon-equilibrium for every epsilon > 0, that is, an epsilon-equilibrium in an extended game in which the players observe a public signal at every stage. We also prove that if a certain matrix that is derived from the payoff s in the game is a Q-matrix in the sense of linear complementarity problems, then the game admits a Nash epsilon-equilibrium for every epsilon > 0.

The talk will consider evolutionary games on graphs, which is a generalization of matrix games to graphs, using the solution concept of evolutionary stable strategies. In evolutionary games on graphs, there is a graph and a 2x2 matrix. Each node of the graph has a strategy of the matrix game - r or b - and initially all nodes are r. Then, a uniformly random node is made to follow b and the following is repeated until the graph is all one strategy again:1. Find the average payoff of each node when playing against their neighbors in the graph (depending on their current strategies).2.

We show that a simple decentralized dynamic, where players update theirbids proportionally to how useful the investments were, leads to growth ofthe economy in the long term (whenever growth is possible) but also createsunbounded inequality, i.e. very rich and very poor players emerge. Weanalyze several other phenomena, such as how the relation of a player withothers influences its development and the Gini index of the system.Joint work with Ruta Mehta and Noam Nisan.