Consumer savings from merger enforcement: A review of the antitrust agencies' estimates

Antitrust Law Journal

Volumn 69, Issue 3, 2001, Pages 921-951

  Nelson, Philip ^a   Sun, Su ^a  

^a Economists Incorporated   (United States)

Author keywords

[No Author keywords available]

Indexed keywords

EID: 0442279260     PISSN: 00036056     EISSN: None     Source Type: Journal    
DOI: None     Document Type: Review

References (65)

1. GPRA requires the agencies to report on the results or outcomes of their enforcement efforts using performance indicators. The agencies chose to produce consumer savings estimates as one way of meeting this requirement.
2. The FTC has reported that its merger enforcement effort saved consumers $1.2 billion in Fiscal 1999. See, e.g., Federal Trade Commission Strategic Plan, Fiscal Years 2000-2005 at 7 (Sept. 2000) [hereinafter FTC Strategic Plan]. The DOJ estimates that its merger enforcement efforts saved consumers $2.551 billion in Fiscal 1999 and $4.094 billion in Fiscal 1998. Antitrust Division Congressional Submission for Fiscal Year 2001 at 76 [hereinafter Antitrust Division Congressional Submission].
3. Each year the agency will "estimate the dollar savings to consumers resulting from the FTC's successful merger and nonmerger challenges." FTC Strategic Plan, supra note 2, at 29. In response to our questions, the DOJ staff indicated that they will "report estimates following the end of each fiscal year."
4. For example, in its Strategic Plan for Fiscal 2000-2005 (which is mandated by GPRA) the FTC indicates that it has a goal of saving consumers over $5 billion during the 2000-2005 period "by stopping anticompetitive mergers and business practices that would otherwise cause price increases or reduce innovation." Of this $5 billion in savings, $4 billion is expected to come from savings from mergers that were successfully challenged. FTC Strategic Plan, supra note 2, at 1, 31. The DOJ has not published a projection. See, e.g., Antitrust Division Congressional Submission, supra note 2, at 76.
5. Based on the interviews, it is clear that the agencies have met to discuss their different approaches and recognized their differences. See also Department of Justice, Office of the Attorney General, Fiscal Year 2002, Annual Performance Plan 113 (Oct. 2000) ("Both agencies track success rate and consumer savings for this crosscutting activity [merger enforcement] - albeit using somewhat different, tailored approaches - as a means to assessing program performance in the merger area.").
6. FTC staff who were interviewed include David Balto, Denis Breen, and Claudia Higgins. DOJ staff who were interviewed include Gregory Werden, Norman Familant, and William Spencer.
7. Government Performance Results Act of 1993, § 1115.
8. Id. § 1116.
9. The GPRA process provides for review by agency Inspector Generals and/or GAO. While the DOJ did not report critical review of its estimates, the FTC indicated that the Inspector General at the FTC did review the FTC's methodology for estimating consumer savings from mergers and that the Inspector General commented favorably on it (e.g., indicated that it was auditable and conservative). They revised non-merger estimates in response to somewhat less positive comments.
10. Prepared Statement of the Federal Trade Commission presented by Robert Pitofsky, Chairman, Before the Subcommittee on Antitrust, Business Rights, and Competition, Committee on the Judiciary, U.S. Senate 11-12 (Mar. 22, 2000) [hereinafter Pitofsky Senate Testimony]. See also Prepared Statement of the Federal Trade Commission presented by Robert Pitofsky, Chairman, Before the Committee on the Judiciary, U.S. House of Representatives 5, (Apr. 12, 2000) [hereinafter Pitofsky House Testimony].
11. Pitofsky Senate Testimony, supra note 11, at 12 n.21. See also Pitofsky House Testimony, supra note 11, at 13 n.24.
12. Pitofsky Senate Testimony, supra note 11, at 12n.21. See also Pitofsky House Testimony, supra note 11, at 13 n.24.
13. Federal Trade Commission, Performance Report, Fiscal Year 1999, at 20 n.1.
14. Antitrust Division Congressional Submission, supra note 2, at 76. The fiscal 1998 estimate was released in the Department of Justice, Antitrust Division, FY 2000 Congressional Budget Submission at 64, and in the Final Report of the International Competition Policy Advisory Committee to the Attorney General and Assistant Attorney General for Antitrust 88 (2000). The DOJ also estimated savings from non-merger antitrust enforcement efforts. For civil non-merger matters, the DOJ estimates that consumer savings were $62.6 million in Fiscal 1998 and $10.3 million in Fiscal 1999. For criminal matters, the DOJ estimates that consumer savings were $232 million in Fiscal 1998 and $245 million in Fiscal 1999. Antitrust Division Congressional Submission for Fiscal Year 2001 at 82, 88.
15. The DOJ includes cases in which it believes that the merger was modified or abandoned because of the DOJ's investigation. It does not include cases that the DOJ lost in district court or cases that did not involve a thorough DOJ investigation. It does include cases where it believes one of the following events took place: (1) merger was abandoned due to Division actions before compulsory process initiated; (2) merger was abandoned due to Division actions after compulsory process initiated without case filed; (3) merger was modified because it was "fixed first" without a case being filed; (4) merger was modified because of a consent decree; (5) merger was resolved with some form of fix prior to conclusion of trial; (6) merger was stopped or modified because of a successful litigation effort with no pending appeals.
16. Antitrust Division Congressional Submission, supra note 2, at 79.
17. The DOJ on rare occasions asks an industry expert for an estimate of the price effect. It will use these estimates directly, especially when it has no better data to estimate likely price increases. This is particularly likely to be the case when the DOJ is attempting to estimate the price effects that will result when there is one less bidder on a prospective defense procurement. For these cases, it is not uncommon for the DOJ to ask the Department of Defense for its opinion about the likely price effects.
18. Antitrust Division Congressional Submission, supra note 2, at 79.
19. The DOJ uses somewhat different methodologies for estimating consumer benefits of its civil non-merger and criminal enforcement efforts. Specifically, in 1998 for civil non-merger matters, the DOJ appears to adopt the FTC's approach and assumes that, but for its enforcement effort, there would be a 1% increase in prices. Like the PTC, the DOJ justifies this assumption by arguing that it is a "conservative" estimate of the likely price effect. Antitrust Division Congressional Submission, supra note 2, at 84. Discussions with the DOJ staff indicate that it continues to rely on 1% as a conservative estimate when it does not have the data to develop a better estimate and when it has not concluded that it should use a zero estimate because it does not expect immediate consumer savings. With respect to criminal matters, the DOJ either employs a direct estimate, or assumes a 10% price increase. Because the data that are needed to produce a direct estimate are often missing, the DOJ often uses the 10% figure, which is based on the figure in the U.S. Sentencing Guidelines. See U.S. SENTENCING GUIDELINES MANUAL, Nov. 1, 1997, § 2R1.1 Application Note 3, at 227; Antitrust Division Congressional Submission, supra note 2, at 90.
20. Simulation models can use a variety of different demand systems. Common demand systems are: linear, logit, log linear (constant elasticity), and almost ideal demand system (AIDS). Interviews suggest that where the simulation method is used, the DOJ's estimates typically use linear demand. However, the DOJ may also have used logit models in some cases. Gregory Werden writes: "In both published papers and actual antitrust investigations, I have used the logit demand system." He cites a footnote that suggests that a logit system was used in mergers involving white bread and cosmetics. Simulation models have also been used in frozen sea food and tissue. Gregory J. Werden, Differentiated Products Mergers, 5 GEO. MASON L. REV. 363-78 nn.76 & 4 (1997).
21. When DOJ economists were asked for details on how they estimated consumer savings in vertical mergers, they could not remember exactly how they made these estimates in particular calculations. However, they did indicate that when the main concern about a merger was horizontal, they did not add to their consumer savings estimate by adding savings that might have resulted from ancillary vertical effects.
22. Antitrust Division Congressional Submission, supra note 2, at 79.
23. While the Merger Guidelines express HHIs as numbers that lie between 0 and 10,000 because market shares are treated as whole numbers (rather than decimals), this formula assumes that market shares are calculated from market shares that are reported as decimals. As a result, an HHI of 1800 under the Merger Guidelines would be 0.18 in the formula.
24. Robert Willig, Merger Analysis, Industrial Organization Theory, and Merger Guidelines, in BROOKINGS PAPERS: MICROECONOMICS 288, equation 5 (after correction of typo) (1991).
25. In fact, in some cases (including the differentiated products market cases described below), the DOJ assigns a value of zero to the benefits when they don't believe that they have the information to develop an accurate estimate.
26. The DOJ was asked to provide a list of the mergers that identified which mergers were determined to involve differentiated products and which were determined to involve homogeneous products so that their categorization of mergers could be reviewed and so that it would be possible to determine how often the different formulas were used. While the staff agreed to see if it could be turned over, there was concern about whether it would involve the provision of confidential information and no list is available at this time. However, statements in the interviews suggested that a simulation model was used in less than half of the cases. Moreover, the DOJ staff indicated that simulation models were used "when we have data to estimate elasticities and generally is not used otherwise." This suggests that if DOJ had a differentiated products case that settled early (e.g., before DOJ had enough data to estimate elasticities), some other approach might have been used. However, the DOJ economists did not believe that they were in a position to say precisely what was done a couple of years ago in dozens of separate cases.
27. DOJ economists indicated that they typically ran simulation models when they were concerned about a merger in a differentiated product market. Without consulting their records, they could not be sure that a simulation model was run in every case.
28. Philip Crooke, Luke Froeb, Steven Tschantz & Gregory Werden, The Effects of Assumed Demand Form on Simulated Postmerger Equilibria, 15 REV. INDUS. ORG. 205 (1999).
29. For a review of these models, see id.
30. Gregory Werden, Luke Froeb & Timothy Tardiff, The Use of Logit Model in Applied Industrial Organization, J. Bus. ECON. (1994); Gregory Werden & Luke Froeb, The Effects of Mergers in Differentiated Products Industries: Logit Demand and Merger Policy, 10 J.L. & ORG. 407 (1994); Gregory Werden, Luke Froeb & Timothy Tardiff, The Use of the Logit Model in Applied Industrial Organization, 3 INT'L L. ECON. & Bus. 83 (1996).
31. Gregory Werden, Luke Froeb & Timothy Tardiff, The Use of Logit Model in Applied Industrial Organization, J. Bus. ECON. (1994); Gregory Werden & Luke Froeb, The Effects of Mergers in Differentiated Products Industries: Logit Demand and Merger Policy, 10 J.L. & ORG. 407 (1994); Gregory Werden, Luke Froeb & Timothy Tardiff, The Use of the Logit Model in Applied Industrial Organization, 3 INT'L L. ECON. & Bus. 83 (1996).
32. Gregory Werden, Luke Froeb & Timothy Tardiff, The Use of Logit Model in Applied Industrial Organization, J. Bus. ECON. (1994); Gregory Werden & Luke Froeb, The Effects of Mergers in Differentiated Products Industries: Logit Demand and Merger Policy, 10 J.L. & ORG. 407 (1994); Gregory Werden, Luke Froeb & Timothy Tardiff, The Use of the Logit Model in Applied Industrial Organization, 3 INT'L L. ECON. & Bus. 83 (1996).
33. Crooke et al., supra note 32, at 2-3.
34. When linear demand is used and market shares are highly asymmetric, it may be necessary to impose non-negativity constraints to avoid situations where negative quantities are predicted. Id. at 5.
35. They write: "In calculating consumer savings across our enforcement areas, key input measures, such as market elasticities of demand, if not actually estimated in the investigation or case, often had to be estimated based on anecdotal information and observations. These values are both conservative and consistently estimated over time." Antitrust Division Congressional Submission, supra note 2, at 46.
36. FTC Strategic Plan, supra note 2, at 29.
37. The FTC staff indicated that the FTC periodically conducts studies that evaluate the effectiveness of its merger enforcement effort. For example, it has studied postmerger price effects in markets where the FTC let a merger proceed but was concerned that it might have been overly permissive. The FTC has also explored whether its remedies have been effective. However, none of these studies is directly on point.
38. An FTC goal is to achieve orders, litigation victories, and abandoned transactions in 80% of the cases in which the FTC finds reason to believe the law has been violated. FTC Strategic Plan, supra note 2, at 31. Because the FTC will have a 100% record for situations in which it issues an order or in which the transaction is abandoned, this implies that they anticipate losing more than 20% of the cases that go to district court. This projection aligns with recent history. For example, the FTC set a Fiscal 1999 goal of a positive outcome in 80% of the enforcement actions brought by the agency to challenge anticompetitive merger practices. It reports that it was successful in 80% of the matters it challenged. There were 23 proposed consent orders, the parties abandoned their transactions in 10 instances, and 2 transactions were restructured after the FTC investigation. Federal Trade Commission, Performance Report, Fiscal Year 1999, supra note 13, at 20. entire FTC maintaining competition mission required 469 full-time-equivalent workers and $55 million in Fiscal 1999 and 514 full-time-equivalent workers and $59 million in Fiscal 2000.
39. Antitrust Division Congressional Submission, supra note 2, at 74. The DOJ indicated that it employed 250 full-time-equivalent workers on mergers in Fiscal 1999, which led to expenditures of $31 million. In Fiscal 2000, the DOJ's Antitrust Division budgeted 813 full-time-equivalent workers at a cost of $110 million for the promotion of competition. In Fiscal 1999, it budgeted 800 full-time-equivalents at a cost of $101 million. Id. at 49B.
40. There were 4,642 HSR fillings in fiscal 1999. The FTC estimates that fewer than 3% of these filings receive Second Requests. Over 85% of the Second Requests are resolved with the parties having to complete their document production. Over 60% of the investigations involve the production of fewer than 20 boxes of documents. Most of the merger filings are for smaller transactions, with only 273 involving billion-dollar mergers. Pitofsky House Testimony, supra note 10, at 2, 6.
41. An informal ABA survey of private law firms in the early 1990s revealed that some Second Requests involved private expenditures of over a million dollars. Based on our work on mergers, it appears likely that expenditures are even higher today.
42. The agencies and the merging parties often disagree on the size of these savings. As the DOJ staff point out, they sometimes find the efficiency claims presented by the merging parties to be "wildly exaggerated and without foundation," which implies that they do not perceive the social losses resulting from this type of deterrence effect to be as large as others might.
43. In particular, they would like to target Second Requests so that "at least 50% of HSR requests for additional information result in enforcement actions." FTC Strategic Plan, supra note 2, at 28.
44. Antitrust Division Congressional Submission, supra note 2, at 49.
45. Id. at 46; see also id. at 79.
46. Federal Trade Commission, Performance Report, Fiscal Year 1999, supra note 13, at 19.
47. It is difficult to dispute the claim that, but for antitrust enforcement, firms would have an incentive to merge in markets where postmerger price increases could be sustained because entry barriers are high.
48. The DOJ explicitly recognizes that one of the challenges in measuring the benefits of legal enforcement efforts is the difficulty of measuring deterrence effects. See, e.g., Antitrust Division Congressional Submission, supra note 2, at 36. While the DOJ has made some effort at measuring deterrence, the main study of deterrence that it has undertaken involves the deterrence effect of its criminal enforcement effort, not its merger effort. Moreover, there was a low response rate to the survey on which this study was based. This study concluded that, if the Division stopped enforcing Section 1, there would be a 150 percent increase in the number of conspiracies. See Antitrust Division Congressional Submission, supra note 2, at 49A.
49. The size of the bias is only ignorable when the price change is very small. In this sense, the DOJ's estimates may have a larger bias than the FTC's estimates because the DOJ's estimated price increases are usually much larger than 1%.
50. For a discussion of the goals of antitrust enforcement, see Robert Lande, Wealth Transfers as the Original and Primary Concern of Antitrust: The Efficiency Interpretation Challenged, 34 HASTINGS L.J. 61 (1982).
51. Discussions also indicated that direct estimates are likely to be a higher percentage in the future because the FTC staff are more aware of the need for these estimates for GPRA reports.
52. RISK MANAGEMENT ASSOCIATION, ANNUAL STATEMENT STUDIES 2000-2001 (2000). For example, according to the RMA (which uses IRS data) the net-profit-before-tax-to-sales ratio for grocery stores has averaged between 1.3% and 1.7% during the last three years ending 3/31/00. Similarly, the net-profit-before-tax-to-sales ratio for gasoline stations has varied between 1.3% and 1.4%.
53. The FTC staff stated that "even in markets, such as gasoline, where margins might be thought to be relatively thin, our experience is that 1% is entirely appropriate."
54. For Fiscal 1998, the DOJ indicates that the dollar volume of affected commerce was $35.303 billion, and that the savings to the consumer from their policies were $4.094 billion. In addition, the DOJ intervened in bank mergers that involved deposits of $3.529 billion. If one excludes banking matters from the denominator, this implies that consumer savings were about 11.6% of affected commerce in Fiscal 1998. If one includes banking matters and weights them by their deposits (which may give them too much weight), one finds that consumer savings were at least 10.5% of revenues. Similarly, in Fiscal 1999, the DOJ indicates that the dollar volume of affected commerce was $15.459 billion, that DOJ intervened in banking matters involving deposits worth $1.992 billion, and that DOJ estimates that the savings to the consumer from its policies were $2.551 billion. If one excludes the banking matters from the denominator, this implies that consumer savings were about 16.5% of affected commerce. If one includes banking mergers weighted by deposits, this percentage falls to 14.6%. Antitrust Division Congressional Submission, supra note 2, at 76.
55. As explained earlier, this equation is: equation presented increase in price. As a result, the results for a Stackelberg model can be very different from those that are observed if a Cournot or a Bertrand model is employed.
56. For a discussion of how differences in demand assumptions can affect postmerger price estimates, see Crooke et al., supra note 32.
57. While the DOJ considers product repositioning in its decision to issue a complaint in a differentiated product market, the DOJ may issue a complaint when there is some repositioning, but it is too limited to prevent a postmerger price increase. As a result, a model that completely excludes the possibility of postmerger product repositioning may lead to overestimates of the size of the postmerger price increases, even if this product repositioning is not sufficient to prevent a postmerger price increase.
58. FTC Strategic Plan, supra note 2, at 7.
59. "Data Limitations: In calculating consumer savings across enforcement areas, key input measures, such as market elasticities of demand, if not actually estimated in the investigation or case, often had to be estimated based on anecdotal information or observations." U.S. Department of Justice, Fiscal Year 2001 Summary Performance Plan at 56.
60. "These [missing and estimated] values [used in DOJ's estimates] are both conservatively and consistently estimated over time." See id.
61. The performance measures that the DOJ employs have evolved over time. For example, the DOJ initially did not include consumer savings estimates. In Fiscal 2001, the DOJ was expected to use a number of performance measures, including savings to U.S. consumers (combining all three of its enforcement areas), success rates for merger and non-merger programs, and volume of commerce affected by its successful criminal enforcement efforts. The DOJ has historically reported high success rates (e.g., reported to have been 98% in 1998 and 97% in 1999). U.S. Department of Justice, FY 99 Annual Accountability Report at III-10-III-11.
62. Antitrust Division Congressional Submission, supra note 2, at 79.
63. Some of the agency personnel that were interviewed indicated that they did not believe that year-to-year comparisons made sense. However, the reported data do not contain clear statements discouraging this type of analysis. and McKesson/Amerisource Health drug wholesaler mergers in 1998. Final Report of the International Competition Policy Advisory Committee, supra note 15, at 88. See FTC v. Staples, Inc. 970 F. Supp. 1066 (D.D.C. 1997); FTC v. Cardinal Health, Inc., 12 F. Supp. 2d 34 (D.D.C. 1998).
64. Antitrust Division Congressional Submission, supra note 2, at 49.
65. The agencies' staff appear to recognize this limitation.