Tymofiy Mylovanov and Thomas Troger, ``Optimal Auction Design and Irrelevance of Privacy of Information''. It's folk wisdom that a seller in an auction can't improve their revenue by withholding her own valuation; unfortunately, I have found formal literature on this topic to be relatively unintuitive. This paper derives a simple economic condition that characterizes when the seller loses nothing by revealing her information in the private-value environment (and you only have to read page 2).
Tomasz Sadzik and Ennio Stacchetti, ``Continuous Time Contracting: Hidden Action and Hidden Information''. This paper shows that Sannikov's elegant continuous time principal-agent results (REStud, 2008) that have given rise to a growing literature are sensitive to how one takes the limit in the corresponding discrete time game. A thought provoking paper.
Rüdiger Bachmann and Giuseppe Moscarini ``Business Cycles and Endogenous Uncertainty''. The first contribution of this paper is to introduce experimentation into a general equilibrium monopolistically competitive industry. They use the model to show that a bad shock makes firms more likely to experiment since their value function becomes more convex near the exit point. Hence negative aggregate shocks can increase price volatility. This is a nice argument, but I'm not sure I'm convinced. Is price experimentation so important in a world with focus groups and marketing consultants? If we're thinking about experimentation from launching new products, this might be more important in booms. Nevertheless, the authors deserve credit for asking these questions.
Godfrey Keller and Sven Rady ``Undiscounted Bandit Problems''. This considers experimentation problems where agents are patient, introducing exogenous information revelation to ensure the game ends in finite time. They derive the optimal experimentation rule and show that it is merely depends on the myopic payoff and the full-information payoff. This is an amazingly simply result, but I'm afraid I have no intuition for it.
Andrew Rhodes ``Multiproduct Pricing and the Diamond Paradox''. The Diamond paradox says that any search costs will lead firms to charge the monopoly price. This paper points out that the logic depends on all customers being inframarginal once they have arrived at the store - an assumption that does not hold when firms sell multiple products.
Volke Nocke and Michael Whinston ``Merger Policy with Merger Choice''. A firm has the option to merge with one of N competitors. These competitors differ in their current costs, so the mergers have different anti-competitive effects (the planner does not want the two most efficient firms to merge unless the savings are huge). Assuming the planner can see the current costs of the firms and the costs of the proposed merger (but not the other options), they solve for the optimal acceptance policy. A very elegant model.
Atilla Ambrus, Eduardo Azevedo and Yuichiro Kamad ``Hierarchical cheap talk''. Suppose you wish to communicate some information to the CEO, but have to go via your boss who may filter the content. This is clearly a huge problem in bureaucracies - just watch The Wire - but seems very hard to analyse. Not so! This paper shows we can look at equilibria which are `filter proof', that the order of the intermediaries does not matter, and the bias of the sender is more important than the bias of the intermediaries.
Sylvain Chassang, Gerard Padro i Miquel and Erik Snowberg ``Selective Trials, Information Production and Technology Diffusion''. Can we think about designing experiments using mechanism design? This is particularly important when agents choose actions that interact with the treatment. For example, this allows the authors to ask whether blind trials are a good idea. A really innovative idea.
Steven Callander, ``Searching for Good Policies''. A nice model where voters faces a continuum of policies and try to find which is best. One contribution of the paper is that payoffs are determined by the realisation of a Brownian motion with drift, and try to find the where it crosses the x-axis.
Andy Atkeson, Christian Hellwig and Guillermo Ordonez, ``Optimal Regulation in the Presence of Reputation Concerns''. This paper presents a reputational model with monopolistic competition. I'm not sure how interesting the regulation question is, but the model could become a workhorse in applications.
Jonathan Thomas and Tim Worrall, ``Dynamic Relational contracts with Consumption Constraints''. This model considers a repeated two-sided holdup problem. Each period consists of two phases: (1) both agents choose actions; (2) the agents split the output, or can quit and get some outside option. When only one party invests, payments are back-loaded and the action converges to the efficient one. When both choose actions, only one agent can have his consumptions back-loaded. Moreover, this agent may overinvest to increase the amount of back-loading. This may help explain why young lawyers work 90 hours a week...
Willie Fuchs and Vinicius Carrasco, ``Dividing and Discarding: A Procedure for Taking Decisions with Non-transferable Utility''. Two agents wish to take a common decision. The agents have private values distributed U[0,1], and quadratic loss functions. The paper shows the utilitarian-welfare maximising decision is remarkably simple. The initial state space is [0,1]. The agents then say if they are above/below the midpoint. If they disagree, the planner implements the midpoint; if they agree, the planner reduces the effective state space to [0,1/2] or [1/2,1]. Then repeat.
Charles Roddie, ``Repeated Signalling and Reputation''. This paper considers a repeated signalling model where both the receiver and the sender move simultaneously. Each period, the sender's type may change (with probability &epsilon), leading to an equilibrium where the agent continuously signals. In the limit, as &epsilon &rarr 0, the equilibrium converges to the Stackelberg equilibrium of the static game. In many ways this is more attractive than the Fudenberg-Levine argument using commitment types. The one problem is that it relies on high types signalling when there is a small probability of them being low. This is not realistic if output is a noisy function of the chosen action: see Brendan Daley and Brett Green, ``Market Signaling with Grades''.
.
Papers will be removed when published.
Date: Aug 2011