York Annual Symposium on Game Theory 2013 13-14 June 2013 Keynote Speakers
Total Page:16
File Type:pdf, Size:1020Kb
York Annual Symposium on Game Theory 2013 13-14 June 2013 Keynote Speakers: Vincent Crawford (Oxford) Fuhito Kojima (Stanford) Ariel Rubinstein (Tel Aviv & NYU) Larry Samuelson (Yale) Department of Economics and Related Studies, University of York, York, UK York Annual Symposium on Game Theory 2013 Programme 13-14 June 2013 • ARRC Auditorium (R/C/014), University of York Day 1: Thursday 13 June 2013 09:00 Registration and Welcome 09:15 Vincent Crawford (Oxford): Efficient Mechanisms for Level-k Bilateral Trading 10:15 Break (Coffee and Tea) 10:35 Oleg Rubanov (LBS): Asymptotic Full Revelation in Cheap Talk with a Large Number of Senders Alex Teytelboym (Oxford): Auctions for complements: an experimental investigation Makoto Shimoji (York): A Non-Equilibrium Approach to the Revenue Comparison of Discrete Private-Value Auctions 12:20 Lunch 13:50 Indrajit Ray (Birmingham): Cheap Talk in BoS with Private Information: Theory and Experiments Jiawen Li (York): Do to No One What You Yourself Dislike? – An Experimental Study of the Golden Rule 15:00 Break 15:20 Tim Worrall (Edinburgh): Relational Contracts: the Risk Averse Case Luigi Siciliani (York): Hospital Mergers: A Spatial Competition Approach 16:30 Break (Coffee and Tea) 16:50 Fuhito Kojima (Stanford): Efficient Matching under Distributional Constraints: Theory and Applications 18:00 Departure for Symposium Dinner The Symposium is financed by the Research and Impact Support (RIS) Fund at the Department of Economics and Related Studies, University of York, UK. York Annual Symposium on Game Theory 2013 Day 2: Friday 14 June 2013 09:15 Ariel Rubinstein (Tel Aviv and NYU): Strategic Tournaments 10:15 Break (Coffee and Tea) 10:35 Bernhard von Stengel (LSE): Equilibria in the Challenge Tournament David Wettstein (BGU): The Optimal Design of Rewards in Contests Paul Schweinzer (York): Auctioning risk: The all-pay auction under mean-variance preferences 12:20 Lunch 13:50 Maria Montero (Nottingham): Proportional Payoffs in Majority Games Suresh Mutuswami (Leicester): Reordering An Existing Queue Zaifu Yang (York): A Competitive Partnership Formation Process 15:35 Break (Coffee and Tea) 15:55 Larry Samuelson (Yale): Investment and Matching 17:00 Farewell The Symposium is financed by the Research and Impact Support (RIS) Fund at the Department of Economics and Related Studies, University of York, UK. York Annual Symposium on Game Theory 2013 Abstracts Vincent Crawford: Efficient Mechanisms for Level-k Bilateral Trading This paper revisits Myerson and Satterthwaite’s (1983; “MS”) classic analysis of mechanism design for bilateral bargaining, replacing equilibrium by a structural nonequilibrium model based on “level-k” thinking, which experiments suggest often reliably out-predicts equilibrium in applications that do not permit learning. The revelation principle fails for level-k models, because the transition from a non-incentive-compatible mechanism to a level-k-incentive-compatible mechanism that seeks to implement the same outcome changes the relationship between level-k players’ beliefs and the correct beliefs (which is not invariant to how the game is framed), altering the incentive constraints. As a result, outcomes implementable by a non-incentive-compatible mechanism may not be implementable by an incentive-compatible mechanism, or vice versa; and the choice of mechanism must trade off the factors that enhance incentive-efficiency in an equilibrium analysis against the benefits of relaxing incentive constraints by influencing level-k players’ beliefs. Even so, restricting attention to level-k-incentive-compatible mechanisms and homogeneous populations of levels allows extensions of MS’s results characterizing incentive-efficient mechanisms to a wide class of level-k models. Without that restriction, or with heterogeneous populations of levels requiring multidimensional screening, the analysis becomes more difficult but some progress is possible. Oleg Rubanov: Asymptotic Full Revelation in Cheap Talk with a Large Number of Senders Strategic information transmission, commonly referred to as “cheap talk”, was first introduced by Crawford and Sobel (1982), who showed that there is only limited amount of information that can be transmitted between an informed sender and an uninformed receiver. Previous attempts to achieve full revelation by introducing several senders relied on unreasonable off- the-equilibrium-path beliefs. This paper shows that in situations with one receiver and N senders, even if senders are limited to sending only two types of signals and the state space is fixed, there is robust asymptotic full revelation, so that the receiver gets to know the state of the world almost perfectly, as the number of senders goes to infinity. Alex Teytelboym: (joint with D. Marszalec): Auctions for complements: an experimental investigation We examine effects of exposure and possibilities for combinatorial bidding in auctions for complements. In a two-item three-bidder setting, we compare four auctions - first-price, Vickrey, Vickrey-Nearest Rule and Reference Rule - analysing both bidder- and auction-level data. We find that in all treatments the first-price auction is revenue-dominant and no The Symposium is financed by the Research and Impact Support (RIS) Fund at the Department of Economics and Related Studies, University of York, UK. York Annual Symposium on Game Theory 2013 less efficient than the other auctions. Exposure only harms global bidders in the first-price auction. Opportunities for combinatorial bidding reduce global bidders' profits in Vickrey, Vickrey-Nearest Rule and Reference Rule auctions - we call this the “combinatorial curse”. Finally, we do not find any pronounced differences in efficiency between Vickrey-Nearest and Reference Rule auctions. Makoto Shimoji: A Non-Equilibrium Approach to the Revenue Comparison of Discrete Private-Value Auctions We analyze discrete private-value first-price and second-price auctions via dominance arguments and compare their revenues. Indrajit Ray: Cheap Talk in BoS with Private Information: Theory and Experiments We consider a Battle of the Sexes game with incomplete information (two types, High and Low, for each player) and allow cheap talk regarding players' private information (types) before the game is played. We prove that the unique fully revealing symmetric cheap talk equilibrium exists for a range of prior probability of the High-type; this equilibrium has the desirable type-coordination property: when the players' types are different, it fully coordinates on the ex-post efficient pure Nash equilibrium. Such type-coordination can also be obtained as a partially revealing cheap talk equilibrium in which only the High-type is not truthful, while the Low-type is truthful. We then run an experiment with a version of the Battle of the Sexes game with two-sided private information, allowing a possible round of either one-sided, two-sided or no cheap talk before the game is played. We compare these treatments to study coordination (on either of the two pure Nash equilibrium) and truthful revelation of information. We find that the unique symmetric equilibrium in the two-sided cheap talk game is played when they players fully reveal their information. Jiawen Li (joint with M. Costa-Gomes and Y. Ju): Do to No One What You Yourself Dislike? – An Experimental Study of the Golden Rule This paper reports an experimental test of whether people do not treat the others in ways that they themselves would not like to be treated, which we call role-reversal consistency. In the experiment, subjects play both roles of a modified version of the standard ultimatum game. We find that over three quarters of the subjects in the experiment are role-reversal consistent. Furthermore, we find a higher level of role-reversal consistency in a treatment where each subject gets to know either if others accept the way she treats them or how others treat her before playing the reverse role, than in a treatment where such information is unavailable. We also find that role-reversal consistent subjects are more accurate predictors of whether others accept the way they treat them and earn more money than role-reversal inconsistent subjects. The Symposium is financed by the Research and Impact Support (RIS) Fund at the Department of Economics and Related Studies, University of York, UK. York Annual Symposium on Game Theory 2013 Tim Worrall: Relational Contracts: the Risk Averse Case This paper presents a dynamic version of the hold-up problem where agents are risk averse and the environment is uncertain. In each period agents invest in a joint project but also have outside options generated by the opportunity to expropriate part of the current output. The question asked is how to structure the investments and division of the surplus over time so as to avoid expropriation. A companion paper has considered the case where agents are risk neutral. In that paper it is shown quite generally how payoffs and actions of one agent are backloaded in an initial phase and that there is convergence with probability one to a stationary phase in which both agents underinvest or efficiently investment depending on the state of nature and other parameter values. With risk- averse agents such extreme backloading is unlikely to remain optimal. In the current paper we consider the case where agents are risk averse but have quasi-linear preferences. We show that there is convergence to a unique invariant distribution independent of the initial conditions. We show that where