Skip to main content

Market Definition Under The Merger Guidelines: Some Modest Proposals

Market Definition Under the Merger Guidelines:
Some Modest Proposals

Willard K. Tom*

Comments Submitted to the DOJ/FTC Merger Workshop
February 17, 2004

The Merger Guidelines and its 1982 and 1984 predecessors represent a major advance over the state of the law and enforcement practice prior to 1982. I know this because I was a brand new lawyer in the Antitrust Division in 1979, and I well remember my sense of befuddlement at being presented with my first merger matter. A disciplined framework for thinking about market definition, in particular, would have been most welcome. The Guidelines provided that framework.

In the face of that signal contribution, the comments here might seem like mere caviling. In the spirit of continuous improvement, however, I offer a few modest suggestions, grouped under two headings:

  • More explicit recognition of the role of uncertainty, and
  • More explicit recognition that the assessment of competitive effect is the underlying purpose of market definition.
  1. More Explicit Recognition Of The Role Of Uncertainty
    1. The Hypothetical Monopolist Test Tempts Staff to Reverse the Proper Burden of Proof

It is a truism that the government bears the burden of proof in a merger case. The very nature of the hypothetical monopolist test, however, tempts staff into reversing that burden. This is so even though the Guidelines explicitly state that they do not attempt to assign the burden of proof or the burden of coming forward with evidence.

The agencies "begin with each product . . . and ask what would happen if a hypothetical monopolist . . . imposed at least a 'small but significant and nontransitory' increase in price" [SSNIP]. "If, in response to the price increase, the reduction in sales of the product would be large enough that a hypothetical monopolist would not find it profitable to impose such an increase in price, then the Agency will add to the product group the product that is the next-best substitute for the merging firm's product." It is a natural human tendency to place the burden of proof on those who argue that something would happen, not on those who say it would not happen. Put another way, the hypothesized price increase is treated as fact, and the hypothesized diversion to other products is treated as speculation. The result will be a bias in favor of narrower markets--a boon to the merging parties if their products are not in the same market as a result, but an impediment to them otherwise.

I am not suggesting that this is a huge or frequent problem. Oftentimes, the market definition issue is not a close one, or the issue turns on issues such as whose econometric model is more convincing. Often, too, neither side explicitly defines a market, at least not in discussions with each other. The staff's Merger Screening memo may contain somewhat off-the-cuff market definitions, and associated HHIs, but nothing of substance turns on them. Nonetheless, the bias is there, subtle as it is, and in rare cases, it may actually make a difference.

At the risk of being branded an antediluvian, let me suggest a possible role for the Brown Shoe "practical indicia" here. If, as a practical matter, business people in the ordinary course of business treat the market as being broader, someone applying the Guidelines test should stop and think twice, or maybe three times, before concluding that the market really is significantly narrower.

    1. "Too Many Notes"

A second uncertainty-related issue springs not from the hypothetical monopolist test itself, but from the fact that product and geographic market determinations are only two of many steps that an analyst theoretically takes to make a judgment on the ultimate question of overall competitive effect. Without in any way denigrating the masterful symphony that is the Guidelines, and with apologies to the playwright Peter Schaffer, I am reminded of the scene in the movie Amadeus in which Emperor Joseph II comments that Mozart's opera has "too many notes." Perhaps the Guidelines contain "too many steps." If merger analysis consists of a long series of steps, all of which must be answered in a particular way for the merger to be anticompetitive, and all of which are inherently uncertain, it is easy to reach the wrong result if one treats each step as fully answered and concluded before moving on the next.

A simple example will illustrate the point. Suppose that in the proposed merger of A and B, C is a firm (perhaps in an adjacent market) that might be able to enter quickly and without sunk costs, and D is a firm that might be able to enter within two years after incurring substantial sunk costs. Suppose there is a 51% probability that, absent the merger, if A were to raise the price of its product, customers would turn to B in sufficient quantities to defeat the price increase. Suppose further that there is a 49% probability that C is in fact an uncommitted entrant that could defeat the price increase, and suppose further that there is a 49% probability that the entry of D would be timely, likely, and sufficient to defeat a price increase. The probability of an anticompetitive effect from the merger is 51% x 51% x 51%, or about 13%. An inexperienced analyst, however, could easily reason as follows: "It is more likely than not that A and B collectively constitute a market. It does not appear likely that C is an uncommitted entrant. Nor does it appear likely that entry by D would prevent an anticompetitive effect. Therefore, we should challenge the merger." Given that the staff and the parties have asymmetric access to the witnesses and documents of customers and other third parties, and given the staff's proper reticence in disclosing information obtained from third parties, it can be extremely difficult to know when the staff has inadvertently fallen into this logical trap.

    1. "Customers Say They Wouldn't Switch"

The information asymmetry issue comes into play in a context other than the layered uncertainties of a multi-step analysis. This was truer in the early days of the Merger Guidelines, when it was more common for the staff to ask customers the "hypothetical price increase" question outright. Today, the focus seems to be more on either concrete past behavior or on objective circumstances such as the timing and costs of switching products. In consumer products, where scanner data are available, the hypothetical question neither needs to be nor can easily be asked directly in any event. Ultimate consumers are simply too numerous; if one were to try to ask the question directly, one would engage survey experts who would, one hopes, explain the limitations and pitfalls of such research and try to protect against the most obvious errors. With intermediate goods, however, staff often begins by doing telephonic interviews of the principal customers. Here, the danger of badly worded questions eliciting misleading information about the unknowable is at its greatest. At the same time, the difficulty of staff and parties joining issue on the implications of conversations to which the parties cannot be privy is also at its greatest. To the extent that staff says, "customers say they wouldn't switch," the likelihood of a productive dialogue is quite small. The discussions are off to a much more productive start if the conversation begins, "customers say they wouldn't switch for the following four objective reasons."

  1. More Explicit Recognition That The Assessment Of Competitive Effect Is The Underlying Purpose Of Market Definition.
    1. When Should the Analysis Depart from 5%?

The Guidelines state that "what constitutes a 'small but significant and nontransitory' increase in price will depend on the nature of the industry, and the Agency at times may use a price increase that is larger or smaller than five percent." I suggest that--on the principle that the assessment of competitive effect is the underlying purpose of market definition--there is a rational basis for departing upwards from 5% in particular cases. I also suggest that, although there is a principled basis for departing downwards from 5% in particular cases, one should never do so, and that one should accordingly delete the sentence in Section 1.0 that states: "The 'small but significant and non-transitory' increase in price is employed solely as a methodological tool for the analysis of mergers: it is not a tolerance level for price increases."

Let us takes the upward departure first. Imagine a situation in which a price increase of 5% would lead to the loss of x units of sales. Suppose that the profit on those x units is greater than the increased profits on the sales that the seller would continue to make. Therefore, a 5% price increase would not be profitable. Now suppose that the next tranche of customers is highly price-inelastic. Even with a price increase of 10%, the seller still loses no more than x units of sales. But this time, with a 10% instead of a 5% price increase, the increased margin on the sales that are kept is great enough to outweigh the lost profits from the sales that are lost. Now the price increase is profitable. Because the purpose of the market definition exercise is to protect consumers from being harmed, why would we not want to guard against this 10% price increase?

The same argument could be applied, of course, to price increases smaller than 5%. A price increase of 1% might cause a small enough loss of unit sales to be profitable, whereas a 5% price increase would not. If we knew exactly what would happen with perfect foresight--if the world resulting from the merger and the but-for world could be known without any uncertainty--I would be perfectly comfortable defining the market based on such a 1% price increase. I suggest, however, that, on average, the uncertainties of life swamp any effort to make such fine distinctions. In order to have some reasonable benchmarks--not only for planning and counseling, but also to avoid unnecessary interference in procompetitive mergers--I propose that we accept 5% as a tolerance level.

    1. It Is Time to Reconcile Unilateral Effects With Market Definition.

There are some confusing redundancies between market definition and competitive effect in the Guidelines. For the most part, they do no harm. Nonetheless, there are confusing. If one is going to revisit the Guidelines anyway, one might want to consider eliminating such redundancies. Probably the leading example is trying to make sense of unilateral effects in light of the sentence in Section 1.11 that reads: "In performing successive iterations of the price increase test, the hypothetical monopolist will be assumed to pursue maximum profits in deciding whether to raise the prices of any or all of the additional products under its control." This appears to mean that the sustainability of 5% price increase in only a single one of the hypothetical monopolist's products is sufficient to define a market. If that is so, then market definition vitiates the need for a unilateral competitive effects analysis. Unilateral effects analysis says that a merger is anticompetitive if, as a result of a merger of A and B, the post-merger entity can raise the price of at least one product without accommodating reactions by other competitors. This may occur, for example, because the market is differentiated and A and B are each other's closest competitors. But if the sustainability of a 5% price increase in only a single one of A's and B's products is sufficient to define a market--as the above-quoted sentence from Section 1.11 seems to say--then by definition, a unilateral effect consisting of a price increase of 5% or more is also a merger to monopoly.

Three possibilities come to mind for reconciling market definition and competitive effects in this context:

  • The two sections are simply redundant.
  • The market definition section allows the agencies to evade the 35% benchmark in the competitive effects section.(1)
  • The competitive effects section allows the agencies to attack a price increase of less than 5%, assuming the agencies do not want to achieve the same goal by varying the 5% benchmark for market definition.

Whatever the explanation is, some clarification appears desirable.

    1. Bring Supply Substitution Back Into Market Definition.

I hesitate to talk about a case that I know only from the outside. Both the staff and the lawyers for the parties saw many documents that I will never see and talked to witnesses to whom I will never talk. So take this discussion for what it is: a "methodological tool" for talking about the Guidelines, not a comment upon the outcome in a particular case. With that caveat, can "intense mints" really be a market? Would not makers of regular mints easily be able to carry out line extensions into the "intense mints" space? Or was this a case in which, once the market was defined, it was all too easy to skip over steps such as uncommitted entry, product repositioning, or committed entry?

I don't know, and the answer may be as simple as the fact that when the staff has the leverage to extract a consent order, all doubts about the effect of the merger tend to be resolved against the parties. But I do sometimes wonder whether the need to obtain a current market share to plug into the spreadsheet in order to generate an HHI number leads staff to systematically underweight the competitive influence of firms whose current market share is zero.


As I said at the outset, my comments on the market definition sections of the Merger Guidelines are minor ones. The Guidelines represented a significant advance, and their achievement has withstood the test of more than two decades. If the agencies are inclined to do some fine-tuning, however, I hope these modest suggestions prove useful.


* Partner, Morgan, Lewis & Bockius LLP. These comments reflect only the views of the author and not necessarily those of Morgan Lewis or any of its clients.

1. Dropping the 35% benchmark would probably be a good idea in any event; I have yet to understand how it is helpful to a unilateral effects analysis.

Updated December 19, 2023