3 - Estimation of (Dynamic) Games: A Discussion

Summary

Introduction

The last 10 years witnessed a number of advances in empirical industrial organization (IO). Exciting new methods for estimating dynamic models and especially dynamic games were developed. These new methods enable researchers to use a wider set of empirical tools. The new methods had positive effects on the well-received literature on estimating static games. At the same time, there have been improvements in single-agent dynamic-model estimation.

The two papers provide excellent surveys covering selected setups in detail. The setups include models with strategic interaction in static and dynamic settings and dynamic single-agent models with a focus on storable-and durable-goods demand models. The single-agent and multi-agent dynamic frameworks enable researchers to explicitly model dynamic linkages in state variables or in payoffs. The setups also nest the static games (or single-agent) setting that arises as a special case when agents are myopic and assign no value to the future.

Three types of dynamic linkages can be distinguished: (1) a strategic linkage arises when a firm's (or consumer's) strategy is allowed to depend on past choices of (all) firms (or consumers); (2) an informational linkage arises when informational uncertainty is resolved gradually over time; and (3) a third type of linkage arises when the current payoff depends on state variables influenced by prior actions. The surveyed literature studies the third type of linkage.