Auction Models

• Pages: 225-245

225

CHAPTER IX

AUCTION MODELS

A. Introduction

When applying for a mortgage, selecting a bottle of Bordeaux, or

giving roses for a special occasion, one indirectly participates in an

auction. Prices of treasury bills, wine futures, and wholesale flowers are

all set at auction. Like traditional market mechanisms, auctions allocate

goods to buyers who are willing to pay the most for them. However,

unlike traditional markets, in which slight price adjustments are usually

met by commensurately small changes in sales, the winner-take-all

nature of auctions can amplify the consequences of mergers and cartels

in markets characterized by auction conditions.

In auction markets, antitrust legal disputes often raise questions like:

is this merger anticompetitive, or by how much did this conspiracy raise

prices? Comparing two states of the world, one before and one after a

merger, or with and without a conspiracy, provides information for

answering these questions. In some cases, natural experiments, like the

collapse of a price-fixing conspiracy, allow observations of both states.

Straightforward estimation using “reduced-form” econometric models

allows a comparison of prices across the competitive and collusive

regimes while controlling for costs and other confounding factors.

Where a good natural experiment is not available, merger and

conspiracy effects can sometimes be simulated with theoretical models

of competition calibrated to observable data. Such “structural models”

are well developed for traditional price-setting markets. For example, in

the WorldCom-Sprint merger case discussed in detail in Appendix I,

both the Department of Justice and interested third parties developed

demand estimates for residential long distance phone service illustrating

that Sprint and MCI were close substitutes. The demand estimates were

used as inputs to a structural model of competition aimed at

Econometrics in Antitrust

226

demonstrating that the proposed merger would substantially raise prices.1

Subsequently, the Justice Department challenged the merger and

WorldCom abandoned its acquisition plans.

WorldCom and Sprint also bid against each other in auctions to

supply telephone services to businesses. To estimate the merger’s effect

on these customers, an analogous structural model of competitive

bidding could be utilized. Just as traditional firms select prices to

maximize profit in light of consumer demand, auction participants

choose bids to maximize profit given their costs of providing the service

or the value of winning the contract. The goal of structural estimation is

to reverse this process and uncover the latent distribution of costs from

observed bids, similar to attempting to uncover characteristics of demand

from observed prices in traditional markets.

This chapter reviews the relatively young empirical literature on

auctions that is especially relevant to issues in antitrust cases.2 The

chapter begins with a primer on auction formats and different

information environments. This is followed by an examination of

reduced-form estimation of conspiracy effects and structural estimation

in the context of merger analysis. A discussion of the role of economic

analysis of auctions in the courtroom concludes the chapter.

B. Primer on Auctions

1. Introduction

Since firm behavior varies depending on the type of auction, one

must understand how observed price and bidding data reflect bidders’

underlying values or costs. Different auction formats generate varying

1. See Opposition of SBC Communications, Inc., CC Docket No. 99-333

(FCC filed Feb. 16, 2000), Ex. B (Decl. of Prof. Jerry A. Hausman).

2. See generally Robert C. Marshall and Michael J. Meurer, Bidder

Collusion and Antitrust Law: Refining the Analysis of Price Fixing to

Account for the Special Features of Auction Markets, 72 ANTITRUST L.J.

83 (2004). For reviews of earlier empirical literature not specifically

focused on antitrust concerns, see Robert G. Hansen, Empirical Testing of

Auction Theory, 75 AM. ECON. REV. 862 (1985); Jean-Jacques Laffont,

Game Theory and Empirical Economics: The Case of Auction Data, 41

EUR. ECON. REV. 1 (1997). For a review of auction design, see Paul D.

Klemperer, What Really Matters in Auction Design, 16 J. ECON. PERSP.

169 (2002).