Common Sense and Simplicity in Empirical Industrial Organization∗

This paper is a revised version of a keynote address delivered at the inaugural International Industrial Organization Conference in Boston, April 2003. I argue that new econometric tools have facilitated the estimation of models with realistic theoretical underpinnings, and because of this, have made empirical I.O. much more useful. The tools solve computational problems thereby allowing us to make the relationship between the economic model and the estimating equations transparent. This, in turn, enables us to utilize the available data more effectively. It also facilitates robustness analysis and clarifies the assumptions needed to analyze the causes of past events and/or make predictions of the likely impacts of future policy or environmental changes. The paper provides examples illustrating the value of simulation for the estimation of demand systems and of semiparametrics for the estimation of entry models. ∗I would like to thank Chris Snyder, Lars-Hendrik Röller, and F.M. Scherer both for organizing the conference and for inviting me to give a keynote talk, and the NSF for financial support. Matt Gentzkow and Chris Snyder provided helpful comments on earlier drafts.