Revising the horizontal merger guidelines

Journal of Competition Law and Economics

Volumn 6, Issue 3, 2010, Pages 619-652

  Carlton, Dennis W ^a  

^a UNIVERSITY OF CHICAGO   (United States)

Author keywords

[No Author keywords available]

Indexed keywords

EID: 77956932230     PISSN: 17446414     EISSN: 17446422     Source Type: Journal    
DOI: 10.1093/joclec/nhq007     Document Type: Article

References (39)

1. U.S. DEP'TOFJUSTICE & FED. TRADE COMM'N, HORIZONTAL MERGER GUIDELINES (revised 1997), available at http://www.usdoj.gov/atr/public/guidelines/horiz_book/hmgl.html [hereinafter HORIZONTAL MERGER GUIDELINES].
2. ANTITRUST MODERNIZATION COMM'N, REPORT AND RECOMMENDATIONS, at 48 (2007). I was one of the Commissioners.
3. Dep't of Justice & Fed. Trade Comm'n, Commentary on the Horizontal Merger Guidelines (Mar. 2006).
4. Fed. Trade Comm'n, Horizontal Merger Investigation Data, Fiscal Years 1996-2007 (2008) [hereinafter FTC Horizontal Merger Investigation Data].
5. See, e.g., Joseph Farrell & Carl Shapiro, Antitrust Evaluation of Horizontal Mergers: An Economic Alternative to Market Definition, 10(1) THE B.E.J. THEORETICAL ECON. art. 9 (2010).
6. Dennis Carlton, Why We Need to Measure the Effect of Merger Policy and How to Do It, 5 COMPETITION POL'Y INT'L 77 (2009)
7. See, e.g., Joseph Farrell & Carl Shapiro, Improving Critical Loss Analysis, 7 ANTITRUST SOURCE 1 (2008).
8. David Scheffman & Joseph Simons, The State of Critical Loss Analysis: Let's Make Sure We Understand the Whole Story, 3 ANTITRUST SOURCE 1 (2003)
9. I use the word "appear" because, in practice, unilateral effects are often estimated using a merger simulation model that allows for the nonmerged firms to alter their prices post-merger
10. Fixed cost reductions result in real resource savings which benefit society. However, there is no need to debate whether merger review should be based on consumer surplus or total surplus (including both producer and consumer surplus) as long as the role of fixed cost savings in fostering dynamic competition is properly recognized.
11. To cite one example, Murphy and Topel find that advances in medical technology and health care have resulted in very large benefits to consumers. They conclude that "[c]umulative gains in life expectancy after 1900 were worth over $1.2 million to the representative American in 2000, whereas post-1970 gains added about $3.2 trillion per year to national wealth, equal to about half of GDP."
12. Kevin Murphy & Robert Topel, The Value of Health and Longevity, 114 J. POL. ECON. 871 (2006).
13. See Michael Katz & Howard Shelanski, Mergers and Innovation, 74 ANTITRUST L.J. 1 (2007) for a discussion of cases involving innovation markets.
14. See Dennis Carlton & Robert Gertner, Intellectual Property, Antitrust and Strategic Behavior, in INNOVATION POLICY AND THE ECONOMY 3 (A. Jaffe, J. Lerner & S. Stern eds., National Bureau of Economic Research 2003).
15. Dennis Carlton, Market Definition: Use and Abuse, 3 COMPETITION POL'Y INT'L 3 (2007)
16. FED. TRADE COMM'N & DEP'TOFJUSTICE, HORIZONTAL MERGER GUIDELINES FOR PUBLIC COMMENT, at 1 (Apr. 20, 2010) [hereinafter PROPOSED GUIDELINES].
17. See, e.g., Craig Peters, Evaluating the Performance of Merger Simulation: Evidence from the U.S. Airline Industry, 49 J.L. & ECON. 627 (2006).
18. In particular, the Proposed Guidelines raise the HHI cutoff from 1000 to 1500 for "moderately concentrated" markets and from 1800 to 2500 for "highly concentrated" markets. In moderately concentrated markets, the proposed Guidelines maintain the standard that mergers that induce HHI changes of 100 or more "potentially raise significant competitive concerns and often warrant scrutiny." However, for highly concentrated markets, the proposed Guidelines raise from 50 to 100 the change required for mergers to warrant scrutiny and from 100 to 200 the change required for a merger to create a presumption of enhanced market power.
19. For an example, note that, in the pharmaceutical industry (an industry with important nonprice competition to introduce new and improved products), in markets with post-merger HHIs between 3000 and 5000 and merger-induced changes in HHI between 200 and 1200 (which would be presumed to enhance market power under the Proposed Guidelines), there were nearly as many cases (10) in which the merger was approved with no conditions as cases (11) in which relief was sought. In contrast, for oil mergers, of the 23 markets with HHI levels and changes in this range, relief was sought in 22.
20. To the extent that research on this topic occurs, it should also evaluate concentration measures other than HHI that are used by the agencies, such as the number of "significant competitors in the market."
21. In particular, in the context of unilateral effects, the Guidelines state that "the elimination of competition between two firms that results from their merger may alone constitute a substantial lessening of competition." In the context of coordinated effects, the Guidelines state that "a merger may diminish competition by enabling or encouraging post-merger coordinated interaction among firms."
22. Throughout this section, I use the term "UPP" to refer to the general methodology described in Farrell and Shapiro's paper and the term "UPP index" to refer to the specific mathematical formula computed as part of the methodology
23. The notation here borrows from Richard Schmalensee, Should New Merger Guidelines Give UPP Market Definition?, 1 GCP: THE ANTITRUST CHRON. 1 (2009).
24. See Glen Weyl, The First-Order Approach to Merger Analysis (Work in progress, Harvard University 2010), available from the author on request
25. This critique applies not only to UPP but also more generally to merger simulation models that rely on a specific model of oligopoly such as static Bertrand competition. This critique also highlights the need for empirical work on the accuracy of predictions based on UPP. The extent to which, in practice, deviations from Bertrand competition may bias results based on UPP (or merger simulation) cannot be known without substantial testing.
26. For example, in a long-run equilibrium in a perfectly competitive market with identical firms, price equals marginal cost, which equals average total cost, and thus exceeds average variable cost.
27. See DENNIS W. C ARLTON & JEFFREY M. PERLOFF, MODERN INDUSTRIAL ORGANIZATION 63 (4th ed. 2005).
28. 4 are prices for Firms 1 through 4, respectively. The demand curves for other firms are symmetric to that of Firm 1. The marginal cost for each firm is $75. Competition is assumed to be static Bertrand. Hence, at the pre-merger equilibrium, the following conditions hold: each product has a price of $100, each firm sells a quantity of 25 units, the own-price elasticity for each product is equal to 24, and the cross-price elasticity between all products is equal to 1.
29. See Luke Froeb, Steven Tschantz & Gregory J. Werden, Pass-through Rates and the Price Effects of Mergers, 23 INT'L J. INDUS. ORG. 703 (2005).
30. The fact that different demand curves can lead to substantially different price predictions is well understood, having been previously documented in the context of merger simulations.
31. See, e.g., Philip Crooke, Luke Froeb, Steven Tschantz & Gregory Werden, Effects of Assumed Demand Form on Simulated Postmerger Equilibria, 15 REV. INDUS. ORG. 205 (1999).
32. The PCAIDS demand system, a variant of the Almost Ideal Demand System (AIDS), was proposed for use in merger simulations by Roy Epstein & Daniel Rubinfeld, Merger Simulation: A Simplified Approach with New Applications, 6 ANTITRUST L.J. 883 (2001).
33. 4), with the demand curves for all other firms symmetric to that of Firm 1. For both Case A and Case B, the marginal cost for each firm is $75. Competition is again assumed to be static Bertrand. Hence, in the pre-merger equilibria for both Case A and Case B, the following conditions hold: each product has a price of $100, each firm sells a quantity of 25 units, the own-price elasticity for each product is equal to 24, and the cross-price elasticity between all products is equal to 1.
34. As discussed above, the UPP index for Product 1 is precisely the same as a cost change only when the prices of other products are held fixed, so that there are no feedback effects. However, to conform with standard merger simulation methods, the results in Table 2 allow the prices of all products to change. My conclusions are unaffected if I instead model a case in which only the price of Product 1 is allowed to change, with the prices of all other products fixed at their pre-merger levels, thus eliminating feedback effects.
35. Note that this result implies that it is entirely possible for there to be situations in which the UPP index is higher for potential merger A than potential merger B even though the predicted price change (using merger simulation) is larger for potential merger B. This result does not require that the mergers under consideration occur in different industries with different demand curves. For example, even if one restricts attention to a specific type of demand curve (for example, PCAIDS), different combinations of own-and cross-price elasticities can generate the same value for the UPP index but yield different predictions of price changes following a merger.
36. The only change from Case B of Table 2 is that I assume that, due to efficiencies from the merger, the marginal cost of Product 1 falls by 10 percent.
37. Some researchers have suggested using this modification to UPP as a way to include the effect of cross-product efficiencies.
38. As part of his analysis, Peters also studies the simpler method of predicting the post-merger price based on historical relationships between prices and market concentration. Interestingly, he finds that, in many cases, this simpler method yields results that are fairly similar to more sophisticated merger simulation methods.
39. As described there, such retrospective studies would need to control for post-merger changes in industry conditions and for the "selection bias" induced by the fact that only consummated mergers can be studied, but there are standard econometric techniques for implementing such controls.