1. Srihari Govindan and Robert Wilson Axiomatic Justification of Stable Equilibria
Refinements usually judge equilibria by the plausibility of the beliefs that support them. This is an extensive form criterion. On the other hand, a traditional viewpoint is that rationality-based theories of behavior should depend only on the strategic form. This paper embraces both views. Invariance, the requirement that solutions depend only on the reduced strategic form, together with a version of backward induction, imply Kohlberg-Mertens stability.
2. Yeon-Koo Che and Ian Gale Revenue Comparisons for Auctions when Bidders Have Arbitrary Types
This paper introduces a clever approach to compare the expected revenue from different auction designs, assuming bidders in each auction use symmetric equilibrium strategies. Roughly, the idea is replace each bidder with a twin who is risk-neutral and has a single-dimensional value distribution but whose equilibrium bidding behavior would be the same. The expected revenue from the original auction must equal the second-order statistic of the twin's value distribution. This approach is used to extend the revenue equivalence theorem to discrete value distributions and to greatly generalize the revenue ranking of first and second price auctions with risk-averse bidders.
3. Marco Ottaviani and Peter Sorenson Noise, Information and the Favorite-Longshot Bias
In pari-mutuel (i.e. horse race) betting, the payout odds on a horse are determined by the fraction out of the total betting pool wagered on the horse. There is a well-documented regularity, the favorite-longshot bias: the odds for longshots overstate, and the odds for favorites understate, the true probability of winning. This paper provides a simple and elegant explanation based on a model of privately informed bettors. Because the odds are determined only after betting closes, bettors do not know the payout odds when they bet. After the betting closes the revelation of odds aggregates information but by then bets cannot be changed. In particular, those who bet on the horse which turned out to be the longshot realize ex post that the probability of winning is lower than they thought, hence the bias.
4. Philip J. Reny On the Existence of Monotone Pure Strategy Equilibria in Bayesian Games
Phil Reny provides new techniques for proving existence of monotone pure strategy equilibria Bayesian games with multidimensional types and actions. It clarifies the role of single-crossing, and uses a weaker version than previously employed. The paper leaves open questions as it has a continuity assumption that while allowing for a wide variety of games with finite strategy sets, does not admit discontinuous games with continuum action spaces, as in many auction models. Nevertheless, the set of games covered is of substantial interest and more general than in previous results, the arguments deepen our understanding of what is needed for existence, and the use of contractability is clever and looks like to be useful beyond the current work.
5. Yann Bramoulle and Rachel Kranton Strategic Experimentation in Networks
Bramoulle and Kranton study the play of local public goods games when players are linked by a network. Players derive payoffs from their own and immediate neighbors actions, and the authors discuss applications to experimentation where players learn and benefit from the actions of immediate neighbors, but not indirect neighbors. Nevertheless, indirect neighbors’ play affect direct neighbors’ choices, as actions are strategic substitutes. While tractability is a challenge, the authors are able to deduce some interesting patterns of behavior and specialization as a function of network architecture.