« Back to Results
Atlanta Marriott Marquis, International C
Hosted By:
Econometric Society
sectional outcomes when data are interpreted as the long-run result of a history of game plays. Agents interact repeatedly in a game of incomplete information, but the econometrician only observes a snapshot of this activity. We remain agnostic on the details of the process and only impose a minimal behavioral assumption describing an optimality condition for the long-term outcome of agents' interaction. In particular, we assume that play satisfies a property of "asymptotic no regret" (ANR). This condition is weaker than the no-ex post-regret property of pure-strategy Nash equilibria. It only requires that the time average of the counterfactual increase in past payoffs, had different actions been played, becomes approximately zero in the long run. A large class of well-known dynamics has the ANR property. Since we do not fully specify what the behavior of agents is or what they do to play according to this minimal requirement, we depart from the current literature on empirical dynamic games that typically imposes the Markov perfect (or related) solution concept. We show that, under the ANR assumption, it is possible to partially identify the structural parameters of agents' payoff functions. We establish our result in two steps. First, we prove that the time average of play that satisfies ANR converges to the set of "Epsilon-BCE of the underlying static game. To do so,we extend to incomplete information environments prior results on dynamic foundations for equilibrium play in static games of complete information. Second, we show how to use the limiting model to obtain consistent estimates of the parameters of interest.
model of two-sided, one-to-one matching with transfers. The data
scenario we consider involves many independent markets, each of which
consists of a finite set of agents with discrete types. We use
exogenous variation in the market composition or the market-level
state variables to point- or set-identify the match surplus, based
only on aggregate matching patterns (defined by the distribution of
matched agent types). Our method allows the agents to have preferences
over unmeasured types of partners, and does not impose any parametric
restriction on match-specific unobserved heterogeneity. It also
reduces the computational costs relative to existing methods that are
based on individual-specific assignments. Simulation results show our
approach leads to informative confidence sets and speedy improvements
as sample size increases.
Methodological Advances in IO
Paper Session
Friday, Jan. 4, 2019 10:15 AM - 12:15 PM
- Chair: Eduardo Souza-Rodrigues, University of Toronto
How Do Physicians React to Patient Costs?
Abstract
In response to rising health costs, insurers have increased the share of health spending that their enrollees must shoulder out-of-pocket. We study whether physicians know the relative out of pocket costs patients face, and whether they respond to these demand-side incentives by altering their treatment recommendations. Using individual claims data from patients covered by all private insurers in the state of Oregon, we examine how physicians’ treatment patterns change both over time, as high deductible health plans become more common, and in the cross-section, as a function of the relative generosity of employer-provided plans and individual market plans. In particular, the creation of individual insurance marketplaces in 2014 under the Affordable Care Act shifts the typical health plan characteristics in that year. Using the moment inequality approach from Dickstein and Morales (2018), we test for heterogeneity in information across physicians, based on their training, mix of patients, and experience in specific disease categories.Dynamic Foundations for Empirical Static Games
Abstract
We propose the use of Bayes correlated equilibrium (BCE) as a restriction on cross-sectional outcomes when data are interpreted as the long-run result of a history of game plays. Agents interact repeatedly in a game of incomplete information, but the econometrician only observes a snapshot of this activity. We remain agnostic on the details of the process and only impose a minimal behavioral assumption describing an optimality condition for the long-term outcome of agents' interaction. In particular, we assume that play satisfies a property of "asymptotic no regret" (ANR). This condition is weaker than the no-ex post-regret property of pure-strategy Nash equilibria. It only requires that the time average of the counterfactual increase in past payoffs, had different actions been played, becomes approximately zero in the long run. A large class of well-known dynamics has the ANR property. Since we do not fully specify what the behavior of agents is or what they do to play according to this minimal requirement, we depart from the current literature on empirical dynamic games that typically imposes the Markov perfect (or related) solution concept. We show that, under the ANR assumption, it is possible to partially identify the structural parameters of agents' payoff functions. We establish our result in two steps. First, we prove that the time average of play that satisfies ANR converges to the set of "Epsilon-BCE of the underlying static game. To do so,we extend to incomplete information environments prior results on dynamic foundations for equilibrium play in static games of complete information. Second, we show how to use the limiting model to obtain consistent estimates of the parameters of interest.
Estimation of Aggregate Matching Models
Abstract
We study the identification and inference of an econometricmodel of two-sided, one-to-one matching with transfers. The data
scenario we consider involves many independent markets, each of which
consists of a finite set of agents with discrete types. We use
exogenous variation in the market composition or the market-level
state variables to point- or set-identify the match surplus, based
only on aggregate matching patterns (defined by the distribution of
matched agent types). Our method allows the agents to have preferences
over unmeasured types of partners, and does not impose any parametric
restriction on match-specific unobserved heterogeneity. It also
reduces the computational costs relative to existing methods that are
based on individual-specific assignments. Simulation results show our
approach leads to informative confidence sets and speedy improvements
as sample size increases.
Identification and Estimation of Firm Level Markups from Production Data
Abstract
No abstract available yet.JEL Classifications
- L0 - General
- C5 - Econometric Modeling