Abstract. I study social learning in networks with information acquisition and choice. Bayesian agents act in sequence, observe the choices of their connections, and acquire information via sequential search. Complete learning occurs if search costs are not bounded away from zero and the network is sufficiently connected and has identifiable information paths. If search costs are bounded away from zero, complete learning is possible in many stochastic networks, including almost-complete networks, but even a weaker notion of long-run learning fails in many other networks. When agents observe random numbers of immediate predecessors, the rate of convergence, the probability of wrong herds, and long-run efficiency properties are the same as in the complete network. The density of indirect connections affects convergence rates. Network transparency has short-run implications for welfare and efficiency. Simply letting agents observe the shares of earlier choices reduces inefficiency and welfare losses.
Learning while Bargaining: Experimentation and Coasean Dynamics – new version in preparation
- Best Graduate Paper Award at the Lisbon Meetings in Game Theory and Applications 2018
- Finalist for the LAGV Prize at ASSET 2018
Abstract. I study bargaining with incomplete information in markets with search and learning. At every instant, a seller with no commitment power makes price offers to a privately informed buyer. Gains from trade are ex ante uncertain: possibly superior outside options may stochastically arrive on either or both sides of the transaction. Arrivals of outside options are public and learning about their existence is common. Delay is always present in equilibrium: different (groups of) buyer types trade at different times. With independent private values, delay is efficient. In contrast, with (endogenously) interdependent values, the timing of agreements is inefficient; excessive delay as wells as excessive hurry are equally plausible equilibrium outcomes. The type of inefficiency depends on the sign of the intrapersonal externality among the seller’s multiple selves. Prices increase or decrease over time depending on which party has a higher option value of waiting to learn. When the seller can clear the market in finite time at a positive price, prices and the seller’s payoff are higher than the competitive ones. This, however, need not be at odds with efficiency. The connection between bargaining externalities and market unraveling is also explored.
Abstract. We develop a method to identify and estimate a game’s primitives when agents interact repeatedly but, because of the environment’s complexity, may not know or understand key features of the interaction. Instead of relying on equilibrium assumptions, we impose an asymptotic ε-regret (ε-AR) condition on the observed play. According to ε-AR, the time average of the counterfactual increase in past payoffs, had different actions been played, becomes small in the long run. Under the ε-AR assumption, we (partially) identify the structural parameters of the stage game. We do so in two steps. First, we prove that the time average of play satisfying ε-AR converges to the set of Bayes (coarse) correlated ε-equilibrium predictions of the stage game. Then, we use the static limiting model to obtain consistent estimates of the parameters of interest. We apply the method to study pricing behavior on an e-commerce platform and show that it yields useful bounds on the distribution of sellers’ marginal costs.
Selected Work in Progress
A Mediator Approach to Mechanism Design with Limited Commitment (with Takuro Yamashita)
Social Learning and the Value of Searching: Identification of Search Cost Distributions (with Emanuele Tarantino)
(Robust) Identification in Repeated Games (with Lorenzo Magnolfi)