تحقیقات تجربی درباره مدل های تئوری بازی انتخاب گسسته ورود: تصویر

We discuss the empirical implementation of discrete game theoretic models of firm entry. After presenting a simple model of entry that underlies much of existing empirical analysis, we discuss the major problems that must be tackled when empirically implementing theoretical models, and econometric methods used in the literature. Finally, we present results from a reduced form estimation of a sequential move entry game, using data from the UK hamburger market.

The last decade has witnessed important advances both in our ability to translate game theoretic models involving discrete choices into econometric models, and in developing relevant estimation methods. Several contributions combine these advances,1 promising a better understanding of the issues, and firmer answers to some fundamental questions. The objective of this paper is to outline some of the key ideas involved in game theoretic models of entry, then to present some results using data on the UK fast food duopoly. The remainder of the paper is structured as follows. In Section 2, we discuss a static two-firm entry model, the various issues that arise in the construction of an econometric model, the modelling choices made thus far in the literature, and the econometric methods used. In Section 3 we present an application to the UK fast food market, and some reduced form results. The fourth and last section concludes.

We have attempted to illustrate how empirical work based on discrete game-theoretic models has evolved in the last decade, in part through the medium of our example. The seminal papers of Bresnahan and Reiss largely set the agenda. The usefulness of simulation estimators, following Berry (1992), removed one obstacle from further development, and all later work uses simulation estimators to some extent to solve the technical problems of potentially endogenous entry. One clear obstacle to swift progress is the availability of good data, since most recent analyses are based on data sets constructed and collected by the researchers themselves. However, the most important task ahead is to move from using static models of entry, and allow for dynamics.