Updating the Merger Guidelines:
12 November 2009
In September, Assistant Attorney General Christine Varney and Federal Trade Commission Chairman Jon Leibowitz announced that our two Agencies were initiating a process to review and possibly update the Horizontal Merger Guidelines ("Guidelines"). AAG Varney explained in a speech why we were undertaking this project,(1) and the Agencies invited outside input by posing a series of Questions for Public Comment ("Questions").(2) The formal closing date for the public comments was this past Monday, November 9th, and we are just beginning the process of reviewing them. We continue to welcome thoughtful comments, which will be reviewed to the extent possible and hopefully will be useful to us as we think about possible revisions to the Guidelines.
We also announced that we will be holding a series of public workshops to explore the Guidelines, structured around the questions we have posed to the public and additional issues that may arise based on the comments we receive.(3) Invitations to participate in the workshops will be made in part based on the quality of comments received by the deadline.
As the Economics Deputy AAG at the Antitrust Division, I am a member of the joint DOJ/FTC Working Group that is charged with reviewing the Guidelines. The Working Group includes my fellow Deputy AAGs Molly Boast and Phil Weiser on the DOJ side, and Joseph Farrell, Rich Feinstein, and Howard Shelanski on the FTC side. Now that we have received the public comments and the public workshops are approaching, I would like to take the opportunity afforded by the Fall Forum to continue the dialogue between the Agencies and the public regarding possible updates to the Guidelines. I hope my remarks today, which reflect the views of the Working Group, will help make the upcoming workshops as informative and productive as possible for the Agencies and the public.
We are now in the process of organizing five workshops that will take place during December and January. All workshops are open to the public and will be webcast.
The first workshop will take place here in Washington, at the FTC, on Thursday December 3rd, just three weeks from today. We are planning a number of panels that will discuss the use of direct evidence to assess competitive effects (Question 2), market definition and the hypothetical monopolist test (Questions 3, 4, 5, and 6), and unilateral effects and product differentiation (Question 10), among other topics. We also envision an overview panel that will discuss the role played by the Guidelines. Rich Feinstein and I are organizing this workshop.
The second workshop will be held at New York University on Tuesday December 8th. We are planning panels on market concentration and the structural presumption (Questions 7 and 9), failing firms and minority interests (Questions 16 and 17), remedies (Question 18) and working with international and state authorities. Howard Shelanski and Phil Weiser are organizing this workshop.
The third workshop will be held at Northwestern University in Chicago on Thursday December 10th. We are planning panels on the use of direct evidence of competitive effects (Question 2), unilateral effects and differentiated products (Question 10), entry and repositioning (Question 13), and efficiencies (Question 14), among other topics. Molly Boast and Rich Feinstein are organizing this workshop.
After a break for the holidays, we will resume with a workshop on Thursday January 14th at Stanford University. We are planning panels on the use of direct evidence of competitive effects (Question 2), unilateral effects and differentiated products (Question 10), price discrimination and large buyers (Questions 11 and 12), and market dynamics and innovation (Questions 8 and 15). Joseph Farrell and I are organizing this workshop.
The final workshop will be held back here in Washington on Tuesday January 26th. We plan panels on market concentration and the structural presumption (Questions 7 and 9), price discrimination and large buyers (Questions 11 and 12), entry and repositioning (Question 13), and remedies (Question 18), as well as a wrap-up panel. Howard Shelanski and Phil Weiser are organizing this workshop.
Let me now turn to the substantive issues that will be addressed during these workshops. Perhaps the best place to begin is by making clear the outer bounds of the Guidelines review project envisioned by the Agencies. In particular, if the Guidelines are revised, we anticipate:
Nonetheless, a number of meaningful revisions could be made while retaining these basic aspects of the Guidelines. For a number of topics, possible revisions are anticipated in the "Commentary on the Horizontal Merger Guidelines," issued by the Agencies in March 2006 ("Commentary").(4)
It is no secret that the structural presumption in merger law has weakened considerably during the 46 years since the Supreme Court issued its ruling in Philadelphia National Bank.(5) And the Guidelines have evolved to reflect this long-term trend. The 1968 Guidelines were heavily focused on market concentration, and the 1982 Guidelines continued to place great weight on market shares. The 1984 Guidelines put more emphasis on a variety of additional factors relevant to assessing a merger's competitive effects.
The 1992 Guidelines placed less weight on market concentration than their predecessors. However, the exercise of defining the relevant market and measuring market shares remains central. In the Overview (Section §0.2) the Guidelines state:
In some contrast, the 2006 Commentary emphasizes that the Agencies take a flexible and integrated approach to evaluating competitive effects, using whatever evidence and methodologies are informative:
The Commentary gives a more accurate picture of how the Agencies currently conduct merger investigations. In cases where the relevant market is fairly clear, measuring shares in that market is a simple and informative first step in screening transactions. However, many investigations focus, at least initially, on evidence about likely competitive effects that is not based on inferences drawn from increases in market concentration. These investigations do not start by defining the relevant market and measuring the level and post-merger change in the HHI.
The workshops will explore whether the Guidelines should be updated to reflect the fact that investigations often do not begin with, or focus on, market definition and concentration. This was the intent behind our Question 1.
Notwithstanding the decline of the structural presumption, the Agencies continue to rely on measures of market concentration, both to decide which mergers warrant the additional scrutiny associated with a second request, and to decide which mergers to challenge. We do not anticipate changing this basic reliance on the structural presumption in the foreseeable future. Nonetheless, as the importance of market concentration in merger law has declined over the decades, our investigations have focused more on direct evidence of competitive effects, and in some cases we infer the relevant market using the same evidence that leads us to conclude there are likely to be adverse competitive effects. Overall, this reflects a healthy trend towards more effective, more sophisticated, and hopefully more accurate merger enforcement. We hope the workshops will provide useful information on whether the Guidelines should be updated to reflect the more flexible approach taken, with its greater emphasis on such direct evidence of competitive effects. We also are very interested in obtaining further input on how such evidence is best evaluated, as reflected in Question 2.
Several other jurisdictions have revised their merger guidelines in recent years. The European Commission issued new guidelines in 2004.(6) Canada also issued new guidelines in 2004.(7) The U.K. is in the process of updating its guidelines, having issued new draft guidelines in April 2009,(8) and having received public input into that process. We hope to learn from these agencies about the benefits, and hazards, of modernizing merger guidelines.
In addition, members of the OECD regularly share their experience with merger review through the auspices of Working Party 3,(9) and the ICN Merger Working Group, currently co-chaired by our very own Phil Weiser, provides another venue where officials and practitioners from various jurisdictions can share best practices.(10)
We have invited a number of foreign officials to participate in the upcoming workshops. We also have invited a representative of the National Association of Attorneys General to participate. We welcome their input on how updating our Guidelines might facilitate the more effective handling of mergers that are reviewed by multiple jurisdictions.
I now turn to a number of specific areas where we have posed Questions for Public Comment. My remarks today do not address in a comprehensive manner all the areas where we seek public comment or might make revisions. I do not explicitly address here issues of geographic market definition (Question 6), price discrimination markets (Question 11), dynamic markets and innovation (Questions 8 and 15), efficiencies (Question 14), minority interests (Question 16), the failing firm defense (Question 17), or remedies (Question 18), though we will be looking at all of these issues.
The hypothetical monopolist test has been part of the Guidelines since 1982. Under this test, a market is defined as a set of products for which a hypothetical, profit-maximizing monopolist would impose at least a small but significant non-transitory increase in price (SSNIP) on at least some of the products sold by the merging firms. As noted above, we do not anticipate fundamentally changing this basic component of the Guidelines. There are, however, a number of aspects of the test that warrant a fresh look and may benefit from alterations.
While the hypothetical monopolist test is well established in antitrust circles, its implications for relevant antitrust markets, especially those involving differentiated products, are not widely appreciated. Updated Guidelines could explain more clearly the implications of the test, which may be counter-intuitive.
Updated Guidelines could make more explicit that the hypothetical monopolist test often leads to properly defined relevant antitrust markets that do not include the full range of functional substitutes from which customers can choose. Question 3 addresses this issue. Again, the Commentary anticipates an important concept that might be included in updated Guidelines:
Updated Guidelines could explain these important points, which continue to cause confusion.
Likewise, updated Guidelines could explain more clearly that relevant antitrust markets identified using the hypothetical monopolist test do not neatly partition products into various markets. To the contrary, the test generates one relevant market starting from each product sold by the merging firms, and these markets need not be the same: they can overlap, they can be nested, or they can be disjoint. (Remember those Venn diagrams from high school?). For example, suppose Firm A sells Product A, Firm B sells Product B, Firm C sells Product C, and all three of these products compete directly against each other. If Firms A and B propose to merge, the Agencies construct the relevant market starting with Product A and the relevant market starting with Product B. Call these "Market A" and "Market B." These may well be different markets. Likewise, if Firms B and C propose to merger, the Agencies will construct "Market B" and "Market C." If Market A and Market C are quite distinct, the structural analysis associated with the first merger can differ a great deal from that associated with the second merger, despite the fact that Products A, B, and C all compete against each other, which might, intuitively, suggest that a single relevant market containing all three products would apply to both mergers. This feature of the test is quite counter-intuitive and can create confusion, especially in markets with differentiated products. These ideas could be developed using real-world or hypothetical examples. We hope to learn more at the workshops about whether updating the Guidelines could help clarify this implication of the hypothetical market test, perhaps using hypothetical or real-world examples.
The hypothetical monopolist test is sometimes implemented using "critical loss analysis." This method calculates the magnitude of lost sales necessary to make a price increase unprofitable for the hypothetical monopolist, which is the so-called "critical loss." Critical loss analysis then seeks to determine whether the sales that would actually be lost due to the price increase, the so-called "actual loss," are greater or less than the critical loss. As usually practiced, this method is best suited to markets in which the various suppliers have roughly equal price/cost margins.
The Guidelines are silent on critical loss analysis as such. This is unsurprising, since critical loss analysis was just emerging as a method when the current Guidelines were drafted.(11) But the method has often been used over the intervening years, and addressing it in the Guidelines might be beneficial. Question 3 raises this issue.
Critical loss analysis is often performed by asking whether imposing a SSNIP would increase or decrease profits for the hypothetical monopolist. This involves a "break-even" profit calculation. The Guidelines (§1.11) ask a facially similar, but ultimately different question: whether the profit-maximizing price increase is at least a SSNIP, not whether a SSNIP would raise or lower profits for the hypothetical monopolist. Clarification of this distinction could be useful. Another problem with critical loss analysis is that it typically only considers a small price increase; in some cases, a large price increase may be profitable even if a small one is not.(12) Regardless of the size of the price increase, the role of pre-merger price/cost margins in the analysis could be explained.(13) The higher the pre-merger margin, the smaller is the critical loss, simply as a matter of arithmetic. But higher margins also constitute strong evidence that the "actual loss" is smaller. The Guidelines could discuss how the Agencies assess the "actual loss," including the role of pre-merger margins in making this assessment. Alternatively, the Guidelines could refrain from explicating these issues related to critical loss.
Questions 4 and 5 address several "technical adjustments" that could be made to the hypothetical monopolist test employed by the Guidelines. We welcome further input on the pros and cons of making such changes.
The Guidelines (§1.11) instruct that products be added to the candidate market in the order of "next-best substitutes." In some cases, adding products in this order requires more information about substitution patterns than is available to the Agencies. Plus, when the order in which products should be added using this procedure is unclear, the analysis can get bogged down in details that may be of little relevance for the ultimate assessment of competitive effects. Are the benefits of using the "next best substitutes" ordering worth these costs?
The Guidelines (§1.11) state: "The Agency generally will consider the relevant product market to be the smallest group of products that satisfies this test." This "smallest market" principle, strictly followed, can lead to certain well-known problems when the products sold by the merging firms are not next-best substitutes. Consider a merger between rival Products A and B. Suppose that Product C is the closest substitute to each of these two products. Suppose further that, using the normal SSNIP, and starting with Product A, Products A and C form a relevant market. Suppose as well that, starting with Product B, Products B and C form a relevant market. The Guidelines could thus fail to identify this merger as horizontal, even though Products A and B are substitutes, even if the merger substantially raises concentration in a relevant market consisting of Products A, B, and C, and even in situations where the merger could lead to significant unilateral or coordinated effects. This problem could be skirted through the device of using a larger-than-normal SSNIP. Employing this solution is problematic, however, because it is unclear what guidance could be given to indicate what larger-than-normal SSNIP size will be used in any given case. This problem could be avoided by dropping or easing up on the "smallest market" principle, as suggested in Question 4. Is this change worthwhile? Are there other principled ways to avoid this problem?
Question 5 raises some additional technical issues regarding the size of the SSNIP. The Guidelines (§1.11) state that "the Agency, in most contexts, will use a price increase of five percent lasting for the foreseeable future." Should the Agencies provide further guidance on when different SSNIP sizes will be used, or generally provide for the use of larger SSNIP sizes? Should the SSNIP size used in most contexts be increased to 10%, which would lead to broader markets? Should the Guidelines provide further explanation of the base price from which the SSNIP is calculated?
The Guidelines (§1.41) state: "Market shares will be calculated using the best indicator of firms' future competitive significance." This principle does not appear to be controversial, but in practice a great deal of effort can revolve around the "correct" way to measure market shares. For example, merging parties sometimes argue that all firms should be counted equally in markets where suppliers engage in negotiations and bidding for business. Debate can also ensue over whether market shares should be measured in units vs. revenues, or sales vs. capacities. Measuring shares can be especially tricky in dynamic markets, where assessing "future competitive significance" can be controversial. Questions 7 and 8 address these issues.
Once the market shares are measured and the level and change in the HHI is computed, the Guidelines (§1.51) specify various thresholds that can either establish safe harbors or be used to create presumptions that a merger will create or enhance market power or facilitate its exercise. Question 9 asks whether these thresholds accurately reflect current Agency practice, whether they should be adjusted, and if so, to what values.
The 1992 Guidelines introduced unilateral effects explicitly into the analysis. This was a major advance. In a significant proportion of merger investigations, the Agencies pursue a unilateral effects theory of harm. This proportion is especially high in cases involving highly differentiated products. Unilateral effects theories are also very common in markets for intermediate products where the customers are themselves businesses and prices are set through negotiations. In many of these cases, intellectual property rights are important sources of product differentiation, and suppliers must anticipate that prices will significantly exceed marginal costs in order to have the incentive to incur the fixed costs necessary to conduct the R&D leading to these intellectual property rights. Further complicating the analysis, the relationship between market definition under the hypothetical monopolist test and unilateral competitive effects can be confusing. Question 10 provides some details on the issues related to unilateral effects that we hope will be addressed in the upcoming workshops.
The basic economic theory underlying unilateral effects with differentiated products goes back to 19th century work by Bertrand. This theory is a standard part of any undergraduate course on industrial organization economics, but applying these ideas in practice to assess the likely effects of horizontal mergers is far from straightforward. A great deal of learning has taken place since unilateral effects were introduced into the Guidelines in 1992. The Agencies and private parties have accumulated considerable experience with a range of techniques, including simple illustrative calculations of unilateral effects, sophisticated merger simulation, and analysis of product repositioning. One of the most important contributors to this learning is a long-standing member of the Antitrust Division's Economic Analysis Group, Greg Werden.(14)
Economists have long recognized that the exercise of defining relevant markets and measuring market concentration is more closely aligned with theories of coordinated effects than with theories of unilateral effects. In coordinated effects cases, the hypothetical monopolist exercise, which identifies a set of products that could profitably be cartelized, is directly relevant. Market definition and market concentration can thus frame and inform the analysis of whether the proposed merger will significantly increase the danger that such coordination will succeed.
Market definition and market concentration are less relevant to the theory of competitive harm in cases involving unilateral effects. This point is not made explicit in the Guidelines, but it is clearly noted in the Commentary:
While unilateral effects are not naturally diagnosed by looking at market shares and changes in market concentration, except in special cases, there are some simple and informative diagnostics regarding such effects: the profitability of the competing products sold by the merging firms, as measured by their price/cost margins, and the extent to which they compete directly against each other, as measured by the diversion ratios between them.
These ideas are not novel. They can be found in a variety of published papers going back more than a decade.(15) I published an article in 1996, during my previous tour of duty as Economics Deputy AAG, sketching out these ideas.(16) But the logic and operation of unilateral effects is not fully articulated in the Guidelines, and it continues to generate some confusion, raising the question of whether it could be more clearly explained and more extensively developed. Again, the Commentary goes beyond the Guidelines and provides a roadmap for at least part of what updated Guidelines might say:
We look forward to learning more about these issues at the upcoming workshops. The role of price/cost margins can be discussed, along with the role of diversion ratios. The use of more sophisticated techniques, including merger simulation, can also be addressed.
Unilateral effects also differ from coordinated effects in that the merger certainly will eliminate independent competition between the two merging firms, which is the source of unilateral effects, whereas a merger may or may not have any effect on coordination with non-merging parties. Because some unilateral effects are "inevitable," it is important for the Agencies to have safe harbors within their unilateral effects analysis so that we can close investigations promptly when no significant unilateral effects are likely to be found.
Just about one year ago, before joining the Antitrust Division, I wrote a paper with Joseph Farrell that discussed the treatment of unilateral effects in markets with differentiated products.(17) We emphasized the role of pre-merger price/cost margins and diversion ratios in diagnosing unilateral effects for such mergers. For this class of mergers, we argued that margins and diversion ratios could be more informative than changes in the HHI. Our analysis relied heavily on work by Greg Werden and by Dan O'Brien and Steve Salop.(18) We constructed a measure of "upward pricing pressure" by multiplying together the margin and the diversion ratio. We suggested that a merger generating substantial upward pricing pressure would tend to lead to higher prices, unless this pressure was offset by other factors such as repositioning, entry, or efficiencies.(19) The economic logic underlying our analysis is the same as that underlying the quote from the Commentary given above. The Agencies look closely at diversion ratios and margins when diagnosing unilateral effects in markets with differentiated products, and have done so for some time. We hope to learn at the workshops whether a more detailed explanation in the Guidelines of how this is done would be helpful.
The Guidelines do not explicitly address the implications of large buyers. In many markets, large buyers are able to negotiate more favorable terms than smaller buyers. Question 12 asks whether the Guidelines should be revised to discuss the implications of large buyers.
In practice, merging parties commonly argue that the merged entity would not be able profitably to raise price because it will be selling to large, powerful buyers. The Commentary expresses skepticism about this argument:
Even if large buyers are able to negotiate more favorable terms than smaller buyers, what further evidence is required to establish that they are immune from harm due to the loss of competition resulting from the merger? Is the role of large buyers different in cases involving coordinated effects, where they might be able to disrupt coordination, than in cases involving unilateral effects? And, even if large buyers are protected, under what circumstances should antitrust analysis attend to the interests of smaller buyers?
Section V of the European Commission's merger guidelines is entitled "Countervailing Buyer Power." Should our Guidelines be revised to incorporate some of the ideas in that Section of the EC's merger guidelines?
Section 3 of the Guidelines describes the entry analysis used by the Agencies, stating: "A merger is not likely to create or enhance market power or to facilitate its exercise, if entry into the market is so easy that market participants, after the merger, either collectively or unilaterally could not profitably maintain a price increase above premerger levels." The Guidelines continue: "Entry is that easy if entry would be timely, likely, and sufficient in its magnitude, character and scope to deter or counteract the competitive effects of concern." We do not envision changing this basic approach to entry analysis.
Within this framework, however, meaningful changes could be made. For example, how well does the "minimum viable scale" approach to the likelihood of entry in Section 3.3 work in practice? Question 13 asks specifically about the distinction made in the Guidelines between "uncommitted" entry, which is discussed in Section 1.32, and "committed" entry, which is discussed in Section 3. During the workshops we would like to learn more about how useful this distinction is in practice and whether it should be retained.
The Agencies, and the joint DOJ/FTC Working Group engaged in reviewing the Guidelines, are looking forward to the workshops that will be held during December and January. I encourage you to attend the workshops and continue to provide the Agencies with your views on whether, and how, we should update the Horizontal Merger Guidelines.
1. Christine A. Varney, "Merger Guidelines Workshops," September 22, 2009, available at http://www.justice.gov/atr/public/speeches/250238.htm.
2. See http://www.ftc.gov/bc/workshops/hmg/hmg-questions.pdf.
3. See http://www.ftc.gov/bc/workshops/hmg/.
4. See http://www.justice.gov/atr/public/guidelines/215247.htm.
5. 374 U.S. 321 (1963).
6. "Guidelines on the Assessment of Horizontal Mergers under the Council Regulation on the Control of Concentrations Among Undertakings," February 2004, available at
7. "Merger Enforcement Guidelines," Canadian Competition Bureau, available at
8. "Merger Assessment Guidelines," Office of Fair Trading and Competition Commission, Consultation Document, April 2009, available at
9. For example, one of the topics for discussion at the June 2009 OECD meetings was the standard for review used for horizontal mergers in different jurisdictions.
10. One of this group's projects for 2009-10 is convergence on substantive merger analysis, including the development of additional recommended practices on market definition and the analysis of failing firms and exiting assets.
11. To the best of my knowledge, critical loss analysis dates back to a 1989 article by Barry Harris and Joseph Simons, "Focusing Market Definition: How Much Substitution is Necessary," Research in Law and Economics, Richard O. Zerbe, Jr., ed.
12. On this point, see Gregory Werden, (2005), "Beyond Critical Loss: Tailored Application of the Hypothetical Monopolist Test," Competition Law Journal, 4, 69.
13. Some of my academic work has been critical of approaches that lead to broader markets in the presence of higher pre-merger price/cost margins, all else equal. Michael Katz and Carl Shapiro, "Critical Loss: Let's Tell the Whole Story," Antitrust, (Spring 2003). Michael Katz and Carl Shapiro, "Further Thoughts on Critical Loss," Antitrust Source, (March 2004). Joseph Farrell and Carl Shapiro, "Improving Critical Loss Analysis," Antitrust Source, (February 2008). See also Daniel O'Brien and Abraham Wickelgren, "A Critical Analysis of Critical Loss Analysis," Antitrust Law Journal, (2003). For a recent application, see Kevin Murphy and Robert Topel, "Critical Loss Analysis in the Whole Foods Case," Global Competition Policy, (March 2008).
14. See, for example, Gregory Werden and Luke Froeb, (1994), "The Effects of Mergers in Differentiated Product Industries," Journal of Law, Economics and Organization, 10(2), 407-426, and Luke Froeb, Steven Tschantz, and Gregory Werden, (2005), "Pass-Through Rates and the Price Effects of Mergers," International Journal of Industrial Organization, 23, 703-715.
15. For an overview of this literature, see Gregory Werden and Luke Froeb (2008), "Unilateral Competitive Effects of Horizontal Mergers," in P. Buccirossi, ed., Advances in the Economics of Competition Law, MIT Press. See also Jonathan Baker and David Reitman, "Research Topics in Unilateral Effects Analysis," October 2009, prepared for inclusion in the Research Handbook on the Economics of Antitrust, Einer Elhauge, ed.
16. Carl Shapiro, "Mergers with Differentiated Products," Antitrust, Spring 1996, 23-30.
17. Joseph Farrell and Carl Shapiro, "Antitrust Evaluation of Horizontal Mergers: An Economic Alternative to Market Definition," available at http://faculty.haas.berkeley.edu/shapiro/alternative.pdf.
18. See Gregory Werden (1996), "A Robust Test for Consumer Welfare Enhancing Mergers Among Sellers of Differentiated Products," Journal of Industrial Economics, 44, 409-413 and Daniel O'Brien and Steven Salop (2000) "Competitive Effects of Partial Ownership: Financial Interest and Corporate Control," Antitrust Law Journal, 67:559.
19. We suggested that a merger generating substantial upward pricing pressure could be presumed to raise prices. This presumption could then be rebutted based on a more compete analysis of competitive effects, encompassing evidence of repositioning, entry, and efficiencies. This is not dissimilar to how the Agencies currently evaluate unilateral effects in markets with differentiated products.