Pakes outline

Econometric Institute and Princeton University Press organize the intensive PhD-course:

"Econometrics and Theory in Empirical Industrial Organization"

Prof. Ariel Pakes (Harvard University)


Applied Analysis of Imperfectly Competitive Markets: Problems and Prospects.

1. Overview

Static vs. Dynamic Analysis.

1.1 Static Analysis: Existing Techniques.

  • Demand analysis.
  • Production and cost functions.
  • Equilibrium assumptions.

1.2 Dynamic Analysis: Recent Contributions.

  • Equilibrium notion.
  • Estimating additional primitives.
  • Computing equilibrium policies.
  • Other issues; asymetric information, dynamic consumers and producers, unobserved serially correlated state variables.

1.3 Problems with Equilibrium Notions.

  • Specification of information sets.
  • Behavioral assumptions.
  • Markets  with a small number of agents on each side (buyer-seller networks). Housing markets.
  • Statics vs. dynamics in price and quantity setting games.
  • Multiple Equilibria
    - In estimation; consistency vs. efficiency issues.
    - In understanding of historical events.
    - In policy analysis.

2. Equilibrium Notions and Applied Work.

2.1 What are we actually using?

  • The notion of self-confirming equilibria.
  • Transient vs. recurrent points.
  • Heterogeneous beliefs and asymmetric information.

2.2 Two Period Games.

  • What do they assume?
  • The notion of a "rest point", and its implications for structuring data.

2.3 The Role of Inequality Estimators.

What do we need to assume?

  • Equilibrium selection.
  • Information sets.
  • Behavioral assumptions.

What can we learn?

  • Parameters.
  • Distribution of disturbances.
  • Information sets.

2.4 Policy Analysis and Multiple Equilibria.

  • Ennumeration.
  • The role of learning theory.

Verbal Outline of Topics

The Lectures consider the applied analysis of imperfectly competitive markets. They follow the standard undergraduate program in first analyzing price setting conditional on the outputs and cost functions of the firms (static analysis), and then analyzing investment decisions and how they determine the evolution of the costs and the outputs produced by those firms (dynamic analysis). The lectures begin by reviewing recent advances in this analysis. This includes advances in;  the estimation of demand systems, costs systems, and the impacts of investments, as well as in the computation of dynamic equilibrium. The other primitive of the analysis relies on assumptions on the nature of equilibrium. Here standard notions of full information Nash equilibria have been typically assumed, and there has been very little investigation of alternatives.

We consider first what assumptions are actually used in the empirical analysis of equilibrium outcomes. This includes examining the content of the standard full information Nash assumptions as it is used in empirical work (i.e. are we actually using the full power of these assumptions, or are we only using something weaker). It also includes the issue of whether we can tell whether full information is an appropriate assumption for the problem at hand, and what can be done if it is inappropriate. In this context we examine; (i) the notion of using a "rest point" to dynamic system (a point which seems stable in the sense that no actor wants to deviate from it) to structure data, and (ii) the use of inequality estimators to unravel the appropriate assumptions. Finally we consider the issue of multiple equilibria and how this possibility effects empirical work.