Michael G. Baumann has expertise related to modeling the effects of mergers. He has analyzed the likely competitive effects of mergers in a variety of industries.

Is a Relevant Market Irrelevant?

A paper proposing a new method for the antitrust agencies to use in screening mergers between manufacturers of differentiated products was released last November by Joseph Farrell and Carl Shapiro, who soon after became chief economists at the FTC and DOJ Antitrust Division respectively. While the paper, “Antitrust Evaluation of Horizontal Mergers: An Economic Alternative to Market Definition,” presents their personal views and does not necessarily reflect the official position of the agencies, it provides some insight into how the authors are likely to approach these types of mergers.

Farrell and Shapiro claim that a new screening mechanism is needed for mergers involving differentiated products because it may be hard for the antitrust agencies to define the relevant market and compute market shares following the methodology in the Horizontal Merger Guidelines. They argue that market boundaries are unclear when dealing with differentiated products and that what really matters is the proximity of the products.

Farrell and Shapiro suggest screening differentiated product mergers for anticompetitive effects using what they term upward pricing pressure (UPP). Under a unilateral effects theory of competitive harm, the merger gives the merged entity a unilateral incentive to raise price. UPP measures the incentive to raise price. It does not measure actual price changes, which depend on shapes of demand curves and responses of the other firms. The authors argue that actual price changes are too hard to measure to use in screening.

The key factors determining UPP are the merging firms’ prices, marginal costs and diversion ratios. These are the same factors identified for determining unilateral effects many years ago. Before the merger, neither merging party would increase price because if it did, the loss of sales to other firms would make the price increase unprofitable. The diversion ratio is the share of those lost sales that would go to the other merging party. After the merger, that share of sales would be recaptured by the firm rather than lost. As a result, the merger would increase the incentive to increase price. The increase in incentives depends on the price-cost margin enjoyed by the other firm, which indicates how profitable it would be to recapture those sales, and the diversion ratio.

For tractability, the authors assume a particular model of competitive behavior, Bertrand, and assume constant marginal costs and constant diversion ratios. Given these assumptions, the UPP is equal to price minus marginal cost multiplied by the diversion ratio. They assume that the results from their simplified model are not misleading – but the analysis does not necessarily accurately reflect all industries.

There will always be a positive UPP if the two firms have a positive price-cost margin and there is any substitution between their products. Recognizing that the screening mechanism cannot forbid all mergers with a positive UPP, Farrell and Shapiro propose incorporating some standard level of efficiencies into the analysis – a standard deduction. For example, any merger could be assumed to reduce costs by a given percent, say 5%. A merger would be presumed to raise prices if the UPP was greater than that percent times marginal costs.

The proposed methodology is not meant to determine the competitive effects of a merger but to provide an initial screening of mergers with possible unilateral effects and establish a level of presumptive harm that the merging parties would then have to overcome without the need to define a relevant market. Thus, it would replace the current practice of computing market shares, calculating the HHI, and determining if the merger falls into the range that raises significant competitive concerns and a presumption of enhanced market power.

The authors argue that it is easier to measure the price-cost margin and diversion ratio than to define a relevant market and compute market shares. But the variables used in the proposed screening analysis are not always readily observable and often have to be guesstimated. Because marginal costs are hard to measure, typically they are approximated by short-run variable manufacturing costs. The use of this proxy, however, may misstate the true price-cost margin. Available accounting data often will not reflect all economic marginal costs, and that will result in overstating the measured margins and the UPPs.

The authors suggest using business documents, survey data, information about consumer switching patterns, econometric methods, or market shares to estimate the diversion ratio. Some of these methods require quantifying qualitative discussions. Others involve processing large amounts of data that may not be available during the screening process. Their suggestion that diversion ratios can be estimated based on market shares brings the analysis close to defining a relevant market. The authors claim that calculating market shares does not necessarily require doing so in a “relevant antitrust market” if all products in the “market” are about equally close substitutes for the product of the merging firm and if one can estimate the fraction of sales that would be lost to those firms in the “market” rather than to firms outside the “market.” But determining to what alternatives a firm loses sales if it raises price comes close to defining a relevant market.

The proposed screen is likely to catch many mergers in its net. For example, if the standard deduction for efficiencies is 5%, there are 10 identical firms whose products are all equally substitutable, and when a firm raises price all lost sales go to other firms in the market, then a merger between any two firms would violate the screen if the price-variable cost margin were greater than 31%.

While the proposed screening mechanism creates a presumption of harm, this presumption can be overcome. The merging parties might be able to show that the estimated UPP is wrong because estimates of margins, diversion ratios, or efficiencies used by the agency are wrong. The parties also might show that the calculated UPP does not indicate any actual price increase because the basic assumptions of the analysis do not apply or because entry or repositioning will negate the UPP. Or the parties could do a full price effect analysis, which might show no anticompetitive harms.

The antitrust agencies probably will still define markets in investigations of differentiated product mergers. Nonetheless, the issues spotlighted in Farrell and Shapiro’s paper are likely to become extremely important, now that they are the chief economists of the two antitrust agencies.

Additional Articles in Summer 2009 Issue of Economists Ink

The CCC/Mitchell Decision and the Standards for Preliminary Injunctions Against Mergers

Investment Incentives and Merger-Specific Efficiencies

EI News and Notes