Merger Simulation

• Pages: 269-309

269

CHAPTER XI

MERGER SIMULATION

A. Introduction to Merger Simulation

Merger simulation can generate a quantitative prediction of the likely

unilateral price effects of a proposed merger.1 Merger simulation has

been used extensively at the antititrust agencies, including in the

EchoStar transaction discussed in Chapter VI, and in the proposed

WorldCom/Sprint merger discussed in Appendix 1. In addition, merger

simulation played a prominant role in the district court decision declining

to enjoin the Oracle/PeopleSoft merger.2

The basic idea is to combine what can be observed easily, such as

prices and shares, with reasonable assumptions about the behavior of

market participants in a manner that allows the calculation of the implied

unilateral effects. Merger simulation can illuminate important factors in

determining the unilateral effects of a merger, as well as how and to what

degree they matter. This chapter discusses the strengths and weaknesses

of merger simulation generally and of specific models used in merger

simulation.

1. As explained below, combining the merging firms, and hence

internalizing the competition among their merging products, may alter

profit incentives and cause the merged firm to raise prices or reduce

output without any sort of coordination with competitors.

2. United States v. Oracle Corp., 331 F. Supp. 2d 1098, 1122 (N.D. Cal.

2004) (“Despite the problems with qualitative analyses, modern

econometric methods hold promise in analyzing differentiated products

unilateral effects cases. Merger simulation models may allow more

precise estimations of likely competitive effects and eliminate the need to,

or lessen the impact of, the arbitrariness inherent in defining the relevant

market.”). But see id. at 1172-73 (finding that government’s case was

“devoid of any thorough econometric analysis” such as calculation of

diversion ratios or cross-elasticities of demand).

Econometrics in Antitrust

270

Merger simulation has been used mainly with differentiated

consumer products, and this chapter examines three models of consumer

demand that have been used in that context. The first four sections of

this chapter provide non-technical introductions to (1) the concept of

merger simulation and its proper use, (2) unilateral merger effects with

differentiated consumer products and the role of demand elasticities, (3)

three models of consumer demand that have been used in merger

simulation with differentiated consumer products, and (4) the advantages

and disadvantages of these three models in merger simulation.

Because economic modeling never fully captures real-world

competitive processes, the ultimate test for the reliability of merger

simulation is how well it predicts the effects of actual mergers. To date,

little is known about that, nor about the accuracy of any other method for

predicting the competitive effects of mergers. For now, the reliability of

merger simulation is judged on the basis of the “fit” between the model

and the industry.

Merger simulation with differentiated consumer products employs

the Bertrand oligopoly model. The Bertrand model assumes competitors

interact just once, each maximizing its short-run profit, with price as the

sole dimension of competition. Bertrand equilibrium is reached when all

competitors are happy with their prices, given rivals’ prices.3 The

Bertrand model is best employed where its assumptions comport

reasonably well with the factual setting of the industry in which the

merging parties compete.4

Assessing the fit between the Bertrand model and real world

economic conditions is largely a matter of evaluating how well it

explains the past. One important aspect of fit is the degree to which the

predicted price-cost margins, which characterize the intensity of

competition, correspond to observed margins. Also important is how

well the model would have predicted responses to significant cost

changes or new product introductions occurring in the recent past.

Finally, it is important to evaluate the role of nonprice competition, e.g.,

3. For a discussion of the Bertrand model, see DENNIS W. CARLTON &

JEFFREY M. PERLOFF, MODERN INDUSTRIAL ORGANIZATION 166-72 (3d

ed. 2000).

4. For further discussion of when the Bertrand model fits the industry well

enough to offer useful predictions, see Gregory J. Werden, Luke M.

Froeb & David T. Scheffman, A Daubert Discipline for Merger

Simulation, ANTITRUST, Summer 2004, at 89.

Merger Simulation

271

advertising and product positioning, as well as the strategic behavior in

repeated competitor interaction, all of which are outside the model.

Merger simulation holds such things as product characteristics and

advertising constant, which can result in misleading predictions,

particularly if aspects of marketing strategy interact in important ways

with pricing. Merger simulation also examines only the unilateral

incentive that may exist for the merged firm to raise price.5 If some form

of pricing coordination (e.g., overt collusion) is likely to exist before the

merger, or arise as a result of the merger, merger simulation may not

offer any useful insights to the merger’s competitive effects.

If the Bertrand model reasonably approximates competitor behavior

before and after a merger, simulation may be helpful in gaining insight

into likely postmerger price changes. Merger simulation can provide

reasonable, if rough, estimates of a differentiated products merger’s price

effects. But significant weight should be placed on the predictions of a

merger simulation only if the modeling assumptions and the simulation

predictions are consistent with the picture of the competitive landscape

painted by the totality of the evidence. Merger simulation can usefully

complement a traditional, fact-intensive analysis of consumers,

competitors, and the institutional setting of an industry, but cannot

substitute for it.

B. Unilateral Effects Analysis and the Role of Demand Elasticities

The basic idea of unilateral effects from differentiated products

mergers is straightforward: Prior to a proposed merger, the producers of

Brands A and B presumably are happy with their prices, given the prices

of rival brands. Thus, a small price increase would reduce profits

because the revenues lost from the resulting sales reduction would

exceed the cost reduction associated with producing less. If Brands A

and B compete, and the price of either is raised, some of its customers

would switch to the other product. The merger of the producers of

Brands A and B would alter the profit-maximization calculus, because

some of the sales lost from increasing the price of either brand would be

5. Unilateral effects may occur when products produced by the merging

parties are close substitutes in consumption. The merger may create a

significant incentive to raise the prices of products that had been sold

separately, if a substantial proportion of the sales lost by one merging

product due to a higher price flow to a product that the merger brings

under common ownership.