Archives For Merger Guidelines Symposium

The stated purpose of the DOJ/FTC Horizontal Merger Guidelines is to “reduce the uncertainty associated with enforcement of the antitrust laws.”  The Guidelines have had limited success in achieving this goal. They generally succeed in two important dimensions and fail in one. First, the Guidelines lay out a five step analysis consisting of market definition, market shares, potential adverse effects, entry, and efficiencies. The procedure is generally consistent with practice, it is a reasonable way to parse the set of issues, and thus provides some real guidance. Second, the Guidelines explain how the Agencies view the economic issues associated with each of these five steps and define broad standards. This is also useful.  The Guidelines describe, for example, that the Agencies evaluate entry on a timely, likely and sufficiency standard and explain the meaning of these standards. This type of content helps parties understand the Agencies’ approach and can reduce uncertainty. There are places where these standards are appropriate and reflect practice and other places where they can be improved.

The Guidelines fall short, however, in setting out the detailed methods of analysis and their relation to the standards for each of these steps. In a few areas, such as market share analysis, the Guidelines are very detailed in the methods and their implications for the standards while in most areas, the Guidelines provide few details. When the Guidelines are detailed, the practice tends to not follow the details (this is usually good, because the details are flawed).  In other areas, the Guidelines provide few details – they do not tell us how to estimate cost reductions, predict the likelihood of entry, or predict likely changes in coordination.

A revision of the Guidelines could attempt to correct details where they exist and add them where they are lacking. I think this would be a mistake because the task is hopeless. A more effective approach to revising the Guidelines would be to eliminate the misleading and flawed details and clarify and improve the definition, description and reasoning of the broader standards.  Such a revision could help the Guidelines achieve their purpose by eliminating misinformation and updating standards and explanations to reflect current practices and research advances.

A few examples of important potential changes along these lines are below:

  • Eliminate the presumptions of anticompetitive impacts from market share analysis and revise the presumptions of no competitive problem to align with practice. The Guidelines imply an anticompetitive presumption of a merger between a 5% and 10% firm in a market with a share distribution of 20-20-20-20-10-5-5. Fortunately, this seems grossly inconsistent with practice. The anticompetitive presumptions provide little guidance for two reasons. First they do not reflect true presumptions. Second, and more important, they play little role because the additional analysis of likely effects almost always trumps any presumption. It is hard to know if these presumptions deter potential efficiency-enhancing mergers, but they should be eliminated. The greatest value of market share analysis is to provide a presumption of no anticompetitive impact for mergers below certain thresholds. This helps promote the goal of reducing uncertainty, so long as the thresholds are consistent with the Agencies’ practices.
  • Explain and emphasize how the market definition exercise and effects analysis depend on the how pricing incentives change with a merger. In a static product differentiation model, this boils down to how elasticities change with a merger. Such a revision would avoid some of the most egregious misuses of critical loss analysis.
  • The Guidelines segment investigations into coordinated and unilateral effects and then further divides unilateral effects into product differentiation and capacity markets.  I find this segmentation problematic for reasons that require more detailed explanation than is appropriate for this forum. Among the problems with the current segmentation are:  (1) it pushes the analysis of differentiated product markets towards static models that may be inappropriate, (2) there seems to be a gap for markets with limited differentiation where capacity does not play an important role, including some bid markets, and (3) the definitions of unilateral and coordinated effects are murky and fail to deal with situations which are not coordination in the sense described by the Guidelines, but are situations where anticipated responses by competitors affects decisions.
  • One final, unrelated recommended revision is that the Guidelines include an explicit caveat that they apply only to merger analysis. In particular, the Guidelines’ market definition methodology is generally inappropriate in Section 2 cases, largely because cellophane fallacy issues that are avoided in mergers by analyzing changes from current prices. A caveat could help reduce their misuse.

The Horizontal Merger Guidelines have brought discipline to the unruly world of merger analysis; but have also accommodated advances in our understanding of the myriad ways in which firms compete and how mergers affect such competition.  However, in cases where there is better information about the effects of the merger than there is about the relevant market, I would change the Guidelines to allow analysis that bypasses market delineation.

My attorney colleagues would immediately point me to section 7 of the Clayton Act that seems to demand market definition because of its reference to a “line of commerce” and “section of the country.”  Indeed, Judge Brown in Whole Foods said that the FTC’s proposal to dispense with market definition was “in contravention of the statute itself.”

However, I would naively point them to section 1 of the Sherman Act that dispenses with market definition in establishing market power or monopoly power; and in establishing anticompetitive effects under the rule of reason.  Why should it be different for mergers?

For consummated mergers, like the FTC’s Evanston case, effects were proven directly; and in many unilateral effects cases, “more direct” proof of effects is possible. In the Oracle case, for example, the court encouraged the use of merger simulation instead of reliance on unreliable market share data.  If we view market delineation as a means to the end of predicting merger effects, and we have better information about the end, why bother with the means?

1. Do the Guidelines need revision?

The Horizontal Merger Guidelines have served a very valuable purpose by making horizontal merger analysis much more sensible than it was prior to the 1980s.There is much less disagreement about horizontal merger policy than there is about vertical antitrust policy so some vertical guidelines would be especially welcome. Nevertheless, despite the lack of widespread disagreement about horizontal merger policy, it is desirable, though not critical, to improve the horizontal guidelines, making sure that they are up to date and relevant. I would be reluctant to change them significantly because they have served a valuable purpose by providing a consistent and durable framework for merger analysis primarily by focusing attention on the right questions to ask in merger analysis, namely how will price and competition be affected post merger.

2. If yes, What would be the first change you would recommend?

If the guidelines are revised,there are several areas for improvement. The main one would be an elimination (or clarification) of the distinction between unilateral and coordinated effects analysis. The guidelines fail to distinguish clearly between unilateral and coordinated effects. The guidelines suggest that there is one theory for unilateral and another for coordinated effects. That is just not so. In terms of economic theory, they both rely on identical economic theory which is non-cooperative game theory.

The way unilateral effects have often been implemented is as part of a merger simulation using a static model with an assumption of Bertrand competition. There’s no reason in theory why you can’t expand a model like that to be dynamic over time. And if it is dynamic then it looks like much of the description of what the guidelines would call coordinated effects where one firm interacts with another over time.  (The discussion of unilateral effects in the guidelines suggests keeping the price of all other firms constant in doing the analysis. This is somewhat of a detail.  It is of course conservative to ignore the fact that other firms may respond to a price increase by their own price increase but there is no need to ignore the behavior of other firms.) Distinguishing between coordinated and unilateral on the basis of whether a model is dynamic or static seems like a peculiar use of terminology.

There is however a relevant distinction that could be made based on whether the competitive structure– the competitive game– will change after a merger. This change could happen if for example post merger information flows between firms became more transparent so as to change what economists call the competitive game. Note that this is different from simply changing the number of firms that are playing any specific competitive game. One could define coordinated to refer to this case of the game changing post merger, but that is not how it is currently defined.

You often see unilateral effects being used when, in order to find an antitrust harm, the market definition would require such a narrow market definition that people who are prosecuting the case feel uncomfortable with it. They therefore prefer to use a broad market definition but to rely on unilateral effects.  But this use of unilateral effects just serves to undermine the market definition in the guidelines and I find it unhelpful.

Welcome From the Agencies

Joe Farrell —  26 October 2009

I think this symposium is a terrific idea and thank Josh Wright for organizing it.  As Chairman Leibowitz and AAG Varney said, we want to have as open a process as possible as we explore whether to update the HMG and, if so, how.   There can hardly be a more open place for a symposium than on the blogosphere.  I anticipate that there will be a wide array of responses and comments posted here, and we will be reading with great interest.

There have been 17 years of learning in both the legal and economic realms since the last major Guidelines revision.  Are there advances in research or evolutions in best practice that should be included in the Guidelines in order to make them more useful to the business and antitrust community?

Although we issued 20 questions, with subparts to many of them, I like how this symposium will get right down to the heart of the matter by focusing on two critical questions to start the ball rolling:  “should the HMG be revised?”, and “if yes, what is the most important revision that you would recommend and why?”

And, as a reminder to all.  The FTC and DOJ are soliciting comments, which are due by November 9, and holding public workshops.  We hope that those participating in this event will also contribute in a more formal way to the agencies’ process through filing public comments at and through participating in our workshops either here in Washington (December 3 and January 26) or at NYU (December 8), Northwestern University (December 10), or Stanford (January 14).