Click here for teaching materials (login required).

Teaching materials for Game Theory Principles

My CV

Listen to Florian Blume playing "All Along the Watchtower" (starts after 8 seconds)  and watch Jimi Hendrix  play.

Listen to Florian Blume playing "Freebird."

See Edda Blume's artwork. 

Here are some of my papers:

"Noisy Talk"   (with Oliver Board and Kohei Kawamura) 

"All Nash Equilibria of the Multi-Unit Vickrey Auction" (with Paul Heidhues, Jonathan Lafky, Johannes Muenster and Meixia Zhang

"A Learning-Efficiency Explanation of Structure in Language"

"Decentralized Learning from Failure" (with April Franco)

We study decentralized learning in organizations. Decentralization is captured through Crawford and Haller’s  attainability constraints on strategies. We analyze a repeated game with imperfectly observable actions. A fixed subset of action profiles are successes and all others are failures. The location of successes is unknown. The game is played until either there is a success or the time horizon is reached. We partially characterize optimal attainable strategies in the infinite horizon game by showing that after any fixed time, agents will occasionally randomize while at the same time mixing probabilities cannot be uniformly bounded away from zero.

"Private Monitoring in Auctions" (with Paul Heidhues) 

We study infinitely repeated first-price auctions in which a bidder only learns whether or not he won the object. While repetition of the stage-game equilibrium is the unique Nash equilibrium in public strategies, with patient bidders there are simple Nash equilibria in private strategies that improve on bid rotation. Sequential rationality is appropriately captured by essentially perfect Bayesian equilibrium (EPBE), which ignores behavior after irrelevant histories. Our main result is the construction of EPBEa that improve upon bid rotation. Assuming symmetry, the exclusionary schemes of Skrzypacz and Hopenhayn, including asymptotically effcient ones, are supported as EPBEa.

"Modeling Tacit Collusion in Auctions" (with Paul Heidhues)

We study tacit collusion in repeated auctions in which bidders can only observe past winners and not their bids. We adopt a stringent interpretation of tacit collusion as collusion without communication about strategies that we model as a symmetry restriction on repeated game strategies: Strategies cannot discriminate among initially nameless bidders until they have become named through winning an auction. We obtain three classes of results: (1) Completely refraining from using names, i.e. strengthening the symmetry constraint, rules out collusion altogether, and even if naming is permitted, as per our definition of tacit collusion, the lack of communication limits collusive strategies and payoffs among impatient bidders. (2) If communication is allowed, there are sustained improvements over bid rotation and competitive bidding among patient bidders. (3) These gains extend to tacit collusion among patient bidders. However, whether tacit or not, collusion need not be efficient.

"Learning and Experiments: The Bootstrap to the Rescue" (with Doug DeJong, George Neumann and Gene Savin)

Using asymptotic critical values, Blume et al. (2002) tested the parameters of stimulus-response (SR) and belief-based learning (BBL) learning models with experimental data from sender-receiver games. Using these same models, we carry out a Monte Carlo investigation of the true levels of the tests with asymptotic critical values. The results show that there are substantial differences between the empirical and nominal levels of the tests. The bootstrap often reduces the distortions in the levels of tests that occur when asymptotic critical values are used. In our Monte Carlo investigation, the bootstrap essentially eliminates the differences between the empirical and nominal levels of the tests for sample sizes typically found in practice. Because increasing the number of subjects in laboratory experiments is often an impractical method of increasing the sample size, the bootstrap provides a practical method for controlling the level of tests in experimental economics.

"Cognitive Forward Induction and Coordination without Common Knowledge: Theory and Evidence" (with Uri Gneezy) (pictures and instructions for this paper)

This paper investigates optimal play in coordination games where cognition plays an important role. The standard approach to games of incomplete information, which assumes common knowledge of the structure, fails in such games. We develop a theory for such games and experimentally test its predictions. Our main experimental results are that (1) with a cognitively demanding coordination task players play differently when playing against themselves rather than against another player, and (2) given the opportunity, players will signal cognition by choosing a more complex task, a phenomenon which we term "cognitive forward induction."