FEDERAL TRADE COMMISSION HEARINGS ON ANTITRUST IN A GLOBAL ECONOMY OCTOBER 23 - 24, 1995

THE USE OF INNOVATION MARKETS IN ANTITRUST EVALUATION OF MERGERS

Statement Of Judy Whalley Howrey & Simon

Reduced innovation long has been identified as a possible detrimental effect of market power.[1] Generally, however the potential for adverse impact on innovation, a dynamic force and not easily measured, has been treated only off-handedly in merger review, in favor of more easily quantified price and output effects. Recently, the antitrust agencies have begun to consider new ways to evaluate more directly the impact of mergers on innovation. Given the importance of innovation to competition and consumer welfare, in my view they are right to do so.

Both agencies have brought merger cases recently which have focused on a concern for diminished innovation.[2] Assistant Attorney General Bingaman, Commissioner Varney and former Commissioner Yao, in particular, have endorsed the need to consider the impact on innovation in evaluating the competitive effects of mergers.[3]

In their new 1995 Guidelines for the Licensing of Intellectual Property, the agencies adopted an approach to analyzing the impact on innovation of licensing arrangements that utilizes "innovation markets."[4] Former Antitrust Division Deputy Assistant Attorneys General Gilbert and Sunshine have authored a very thoughtful and thought provoking article outlining a rigorous methodology for defining such innovation markets and utilizing them to analyze a merger's likely competitive effects.[5] The approach envisions a separately defined market for innovation in which the merging parties are horizontal innovation competitors even though they currently may not be horizontal competitors in all or, conceivably, any of the end product markets that potentially would benefit from their competing to innovate. The primary benefit of such an approach, as articulated by Gilbert and Sunshine, is to capture all substantial and foreseeable adverse impacts on innovation, even those that would otherwise not be identified because the parties are not currently direct competitors in some relevant end product market. This approach has been endorsed by Commissioner Varney.[6]

This focus on the impact of mergers on the pace, direction and volume of innovation is well justified. Innovation is critical to the long-term success of American industry and the health of the American economy. Concern has developed, particularly over the last two decades, over whether American industry is keeping pace with worldwide innovation. In a marketplace that is increasingly global, America cannot afford to lose the innovation race, either in newly emergent high tech industries or in older smoke stack industries that have formed the bedrock of the American economy.

The methodology outlined by Gilbert and Sunshine to identify and utilize innovation markets promises to be a useful tool in the antitrust enforcement arsenal for evaluating mergers. As identified in their original article, and in their recent reply to resulting criticisms, the use of innovation markets permits the identification of adverse welfare effects in output markets that could not be reached by more traditional analysis of horizontal competition in output markets, potential competition analysis or vertical merger analysis. While there may be only limited numbers of mergers where adverse effects from a reduction of competition in innovation would "fall through the cracks" without the analytical tool provided by innovation markets; nonetheless, if the tool enables enforcers to prevent such adverse effects, it should be welcomed. The analytical framework set forth by Gilbert and Sunshine appears to be theoretically sound and they have identified at least two circumstances in which innovation markets should be used -- the GM-ZF situation (where only one of the merging parties competes in a U.S. output market, but both compete to innovate on products sold by one in the U.S.) and the situation presented in Example 4 of the DOJ/FTC Guidelines for the Licensing of Intellectual Property (where neither merging party is presently a competitor because the relevant output product is not yet produced). Equally or perhaps more importantly, the methodology outlined should prove useful in analyzing effects on innovation in more traditional horizontal and potential competition markets.

However, several factors urge caution in adopting and implementing this innovation market approach.

First, unlike the relationship between increasing concentration and increases in price/decreases in performance which is generally, although not totally, accepted in the economic community, the relationship between concentration and the pace and level of innovation is less well understood and clearly is more complex. Strong proponents exist in the economic community for both a positive and an inverse relationship between concentration and the pace and level of innovation.[7] There is no consensus when, or if, an increase in concentration will have detrimental effects on innovation. Surely it is the case that reaching agreements on innovation output and policing any such agreement would be most difficult. A good case can be made that antitrust enforcers should focus on transactions that create unilateral power rather than those that are theorized to increase coordinated interaction to suppress innovation. In any event, the presumption that increased concentration will lead to a decrease in price competitiveness that underlies traditional merger analysis is not readily transferable to an evaluation of the effects of increased concentration on innovation.

In addition to this uncertainty with regard to the general relationship between concentration and innovation, the factors that affect the role, direction and pace of innovation in a particular industry are not as well understood as the factors that effect the likely price/output behavior in an industry. For example, in some industries innovation primarily comes from the fringe, while in other industries it comes primarily from the market leaders. Would a decrease in innovation from market leaders as a result of a merger lead to an increase in innovation from the fringe? Would such fringe innovation be more or less effective than the innovation that might otherwise have come from market leaders? Are the incentives for process innovation different than those for product innovation? How are incentives to innovate affected by a large installed base? How does the history of previous innovation in an industry illuminate, if at all, the nature of innovation that is likely to come.

A discussion of any of these topics and its impact on the definition, analysis and importance of innovation markets could take up all of the afternoon's time. The key point I would like to make is that while there is some uncertainty and debate over the many factors that, in any given industry, influence the likelihood a merger would lead to supracompetitive prices and a reduction in output, the uncertainty is substantially greater over the factors that determine whether a merger will reduce innovation. This uncertainty counsels caution in moving to challenge transactions because of a concern they will adversely impact innovation. Some have suggested that this uncertainty argues against any antitrust challenge being mounted because of an alleged impact on innovation[8]; however; evaluation of the impact of mergers on innovation should not be abandoned. The importance of innovation is too great and the impact of mergers on innovation has been ignored or downplayed for too long. Rather, I would urge that higher standards be imposed for bringing such cases and that rigor be used in evaluating the impact of a merger on innovation. Courts and enforcers should impose a higher standard of certainty about the likely impact on innovation of a particular merger than they would about the likely impact on price competition. Only in cases where there is a high degree of certainty of an adverse impact from the specific transaction should a challenge be mounted or sustained. While somewhat eroded, traditional merger analysis still benefits from the Philadelphia National Bank presumption; enforcers may not be required to prove just how a merger that substantially increases concentration will result in decreased price competition in a particular market. In the case of an innovation market, I believe proof of why, in the specific market in question, the merger will result in decreased innovation should be required, with no presumption of adverse effects. Specific proposals for implementing such higher standards follow.

With any new theory or approach in antitrust enforcement there is a risk of excessive adoption or application. A new theory or approach is proffered that has merit and is an advance beyond previous thinking. Having received an enthusiastic reception, suddenly it seems applicable in a widespread range of cases and becomes the most exciting and important approach to use. Everyone wants to try it out in his or her case - - - with the result that it may be applied in cases beyond those it fairly can address.

Eventually the pendulum swings back and balance is achieved. The theory finds its proper and useful place. Adoption and blind application of Chicago School thinking on the role of entry is a good example of such a phenomena. Enforcement agencies in the initial burst of enthusiasm for a more sophisticated analysis of the role and importance of entry over-applied the theory, with the result that, in my view, too few cases were brought. It is important that enthusiasm for innovation market analysis not lead to marginal cases being brought, cases that may in fact chill or deter innovation, but in any event would deny the procompetitive effects that generally flow from mergers.

- SPECIFIC PROPOSALS FOR USE OF INNOVATION MARKETS -

1. Circumstances in which Utilization of an Innovation Market is Warranted.

Gilbert and Sunshine have indicated, as do the 1995 Guidelines for the Licensing of Intellectual Property, that the circumstances in which it is necessary or appropriate to adopt an innovation market approach are limited - most often concerns about the impact of a merger on innovation can be captured and addressed either in evaluating the competitive impact of the merger in more traditional horizontal product markets or through potential competition analysis.[9] Perhaps even more importantly, Gilbert and Sunshine note that "defining innovation markets is more likely to be feasible when innovation requires specific assets and the population of firms possessing those assets can be reasonably identified."[10] Further, they note the "sources of innovation may not be reasonably identified unless innovation would be unlikely in the absence of specialized assets."[11] I believe this point is critical not only in determining whether an innovation market may be defined but it is also in assessing the impact on innovation of a merger under more traditional horizontal competition or potential competition analysis. The sources for innovation are potentially very diverse (see discussion below on entry) and the incentives/pressures to innovate are strong. Only when specialized assets are required for innovation and are in short supply is there likely to be a detrimental impact on innovation, however analyzed.

2. Geographic Markets

Gilbert and Sunshine suggest in their article that the logical presumption is that the relevant geographic market for innovation is the world (assuming no trade or regulatory barriers would prevent R&D from being disseminated.)[12] Such a presumption should be explicitly adopted by the enforcement agencies. Even where the output markets are national or local, it should be presumed that any market for innovation is worldwide. Ideas generally are not subject to the constraints that may limit the flow of products and services -- such as tariffs, shipping costs, availability of distribution or services, brand name recognition, etc. Ideas developed outside the U.S. by companies or individuals not participating in U.S. markets for goods or services may be disseminated here in a number of ways: shipment of products or provision of services implementing the technology from the home facilities of the innovator, entry of the innovator into U.S. production, sale of the innovation to fringe competitors or upstream or downstream market participants in the U.S. or sponsorship of a new U.S. entrant. Only in extraordinary circumstances should evaluation of a merger's effects on innovation exclude non-U.S. sources of innovation (such as where introducing of the innovation into the U.S. could not be accomplished by any of the above means).

3. Competitive Effects

Consistent with limiting challenges to those transactions where competitive harm is sufficiently likely to occur to outweigh the limits of our understanding about market factors affecting innovation, concern about competitive effects in innovation markets should be focused almost exclusively upon unilateral rather than coordinated effects.

The ultimate competitive concern with the impact of a merger on innovation is that because of the merger less money/effort will be invested in research and development or alternative research tracks will be abandoned and, as a result, innovation will be reduced -- less innovation will occur and/or it will occur more slowly. This could occur because the acquisition reduces the merged firm's incentives to innovate and/or increases its ability to deter innovation (unilateral effects) or because it increases the incentives and/or abilities of the firms remaining in the market to reach a coordinated outcome that reduces each firm's investment in research and development or limits research tracks (coordinated effects).

Achieving a coordinated outcome on research and development is sufficiently unlikely that the uncertainties inherent in evaluating the impact of mergers on innovation will almost invariably outweigh the likelihood of coordinated effects. Absent extraordinary circumstances (such as clear evidence of pre-existing coordination of research and development efforts, or acquisition of a maverick innovator with substantial evidence indicating an intention to acquire the maverick in order to eliminate its disruptive innovation efforts and restore stability of research and development among the remaining market participants), challenges to mergers based on the transaction increasing the risks of coordinated reductions in innovation are unwarranted. As noted by Gilbert and Sunshine, reaching a coordinated outcome in an innovation market is particularly difficult.[13] To accomplish a coordinated outcome (through either tacit or explicit collusion) requires the companies in the industry be able to agree upon a coordinated outcome, monitor each other's compliance with the "agreement" and punish any deviations.[14] An anticompetitive outcome as a result of coordinated effects on innovation is unlikely because:

1. It would be difficult to reach "agreement"

a) companies' abilities to innovate may be very different;
b) companies' approaches (technical and otherwise) to innovation may be very different;
c) companies' gains from innovation may be very different;
d) secrecy in the innovation process makes resolving these differences and achieving trust very difficult;
e) the rewards from innovation (cheating) may be very great and may be sustainable (particularly if the innovation "leap frogs" others); and
f) because innovation may come from nonmarket participants or upstream or downstream participants and may spring on the marketplace before coordinating competitors could respond, with significant disadvantages to coordinating companies over both the short and longer runs, the risks of coordinating on innovation may be particularly great;

2. because research and development is generally conducted confidentially, detecting cheating in innovation is difficult until "the cat is out of the bag" and the innovation is implemented. In the case of process innovation, cheating may not be detectable even after implementation.

3. There is no easy way to mete out punishment once innovation is detected after implementation.

In contrast, a merger may increase the merged company's incentives and abilities to compete in innovation (1) by giving it a larger market share over which to enjoy the benefits of the innovation and to spread its costs, (2) by eliminating unnecessary redundancies in research and development and freeing more dollars for productive research efforts, thus enabling it to achieve economies of scale in research and development and, by (3) enabling it to combine complementary technologies, etc. These benefits only strengthen the incentives to innovate and make it less likely the merged entity would find it attractive to coordinate and reduce research and development given the difficulties of reaching and sustaining the coordinated outcome.[15]

4. Unilateral Effects and Entry

In evaluating whether unilateral competitive effects could occur, the ultimate issue is whether the merged firm has the ability unilaterally to reduce industry research and development and slow or deter innovation. The merged firm's ability and incentives to restrict its own research and development must be considered and only when it can be established that it would be profitable for the merged firm to restrict its innovation output should a challenge be mounted. To a large extent that conclusion must depend on the ability of other firms in the market, or outside it, to make up the loss in research and development. It will not be in the interest of the merged firm, no matter how big, to restrict its innovation only to lose the innovation race and its position in the end product markets because others stepped forward to innovate when it did not. Thus critical to any government challenge should be a determination that others cannot effectively replace the merged firm's innovation output either through expansion of existing research and development efforts or through entry. While it is unclear whether the government or the merging parties bear the burden of proof on the issue of entry in traditional horizontal merger analysis,[16] given the risks of over-enforcement associated with innovation markets, the government should bear the burden of proof on entry or expansion in innovation markets as a part of its burden of establishing the likelihood the merger will have adverse effects on innovation.

Entry in innovation may come from a variety of sources - the companies in the market, companies in the same product but other geographic markets, from upstream or downstream market participants, from the halls of academe or from government sponsored or conducted research, among others. Excluding the possibility of entry from such a broad array of potential sources will be a substantial burden for the government. However, the government could successfully prove entry is not likely by demonstrating, as Gilbert and Sunshine suggest was true in the GM/ZF case, that the merged firm has sole access to critical inputs in the innovation process (in that case production experience and related production capability necessary to test innovations), or by showing other firms have identifiable disadvantages in the innovation process that cannot be remedied in a reasonable period of time or have such disadvantages in the end use markets that their innovations could not be disseminated in any reasonable time (and no access to others who do not have such disadvantages in end use markets), or by proving the merged companies' headstart in a lengthy regulatory process insures no innovator could catch them in any reasonable time. The two years for entry under the Guidelines may be good general rule of thumb for a reasonable period of time, but where innovation efforts take a substantial period of time, it may be appropriate to lengthen the period for measuring likelihood of entry.

The above criteria will restrict the circumstances in which innovation market analysis can be used be, but these are the only circumstances in which I think our present state of knowledge about the innovation process warrants a challenge. Otherwise we will risk actually deterring innovation and lessening the potential competitiveness of American companies.

Further, I think it is important to continue to evaluate the nature of incentive for process as opposed to product innovations. The incentives to innovate on process may be stronger and less likely to be affected by the potential to exercise market power in downstream markets than the incentives to innovate on products.

Unless innovation is very costly or renders installed equipment obsolete, it is likely to be to a company's advantage to improve its process and reduce its unit costs even if it has market power in downstream end use markets. The incentives for product innovation may not be as strong for a company with market power downstream. Product innovations may be much more disruptive of the exercise of market power. Gains to the company from product innovations are less predictable given issues of customer acceptance (whereas the company, as it innovates, can quickly and easily determine the value of process improvements). If further study supports a conclusion that incentives are stronger for process innovation and unilateral restrictions on process innovation are correspondingly less likely, challenges based on innovation markets should be restricted to those involving identifiable impacts on product innovation.

5. Efficiencies

While not an identified topic for our discussions of innovation markets, I would also like to raise some issues with respect to the treatment of efficiencies in innovation markets. Gilbert and Sunshine, Varney and Yao and DeSanti have written that, as in traditional market merger review, even if it is concluded a merger may have an anticompetitive effect after defining markets, analyzing concentration and evaluating competitive effects, it is still necessary to weigh the efficiencies associated with the transaction, before reaching a final conclusion as to the competitive impact of the transaction. All would limit the evaluation of efficiencies to those associated with the innovation market, i.e., efficiencies in research and development.[17] It is not clear why efficiency considerations should be so limited. Why not consider all efficiencies associated with end use markets to which the innovation market relates. Traditionally, efficiencies in markets other than the one(s) in which the potential anticompetitive effects would occur were not considered by the enforcement agencies in merger review, in large part because enforcers found in most transactions the assets involved in the problematic overlap could be spun off, permitting the balance of the transaction to go forward capturing efficiencies associated with other markets. However, where efficiencies were produced in markets which utilized common production assets with markets that presented antitrust issues, at least the Antitrust Division has indicated it would consider efficiencies in both markets.[18]

Innovation markets are likely to be similarly intertwined with the end product markets - i.e., if there were a competitive problem in an innovation market the company may utilize joint assets for both innovation and the end product market and it may not be possible to "spin off" just the assets associated with the innovation market. In that circumstance at least, it is appropriate to consider efficiencies associated with serving the end product market as well as efficiencies associated with the research and development in assessing the impact of the merger.

 

Endnotes:

[1] United States v. Aluminum Co. of Amer., 148 F.2d 416, 427 (2d Cir. 1945)

[2] See, for example, Roche Holdings Ltd. No. C3315 (filed Nov. 28, 1990); Sensormatic Electronic Corp., File No. 941-0126 (filed Jan. 4, 1995); Wright Medical Technology, Inc., No. C-3564 (filed March 23, 1995); U.S. v. General Motors Corp. No. 93 53-530 (D. Del. filed Nov. 16, (1994). U.S. v. Flow Int'l Corp. No. 94-71320 (S.D. Mich. filed April. 4, 1994)

[3] Assistant Attorney General Bingaman, "The Role of Antitrust in Intellectual Property," Address to the Federal Circuit Conference, June 16, 1994; Commissioner Varney, "Antitrust and the Drive to Innovate: Innovation Markets in Merger Review Analysis." Antitrust, Vol. 9, No. 3, Summer 1995, Commissioner Dennis A. Yao and Susan S. DeSanti, "Innovation Issues Under the 1992 Merger Guidelines," Antitrust Law Journal, Vol. 61. p. 505.

[4] 1995 Antitrust Guidelines for the Licensing of Intellectual Property, at 11.

[5] Gilbert and Sunshine, "Incorporating Dynamic Efficiency Concerns in Merger Analysis: the Use of Innovation Markets." Antitrust Law Journal, Vol. 63, p. 569.

[6] "Antitrust and the Drive to Innovate," supra, note 3.

[7] See summaries of articles in Gilbert & Sunshine, supra note 5, at 574-76 and Rapp, infra note 8, at 13, note 35.

[8] Richard T. Rapp, The Misapplication of the Innovation Market Approach in Merger Analysis, 64 Antitrust Law Journal 19 (1995).

[9] Even when more traditional analysis is utilized, however, the analytical model for assessing impact on innovation described by Gilbert and Sunshine may be useful.

[10] Gilbert & Sunshine, supra note 5, at 588.

[11] Id.

[12] Gilbert & Sunshine, supra note 5, at 594.

[13] It is worth noting that while the government has brought hundreds of cases involving coordinated prices or output levels, I am aware of only one case challenging coordinating research and development efforts. United States v. Automobile Mfrs. Assn., 1969 Trade Cas. (CCH) para. 72,907 (CD. Cal 1969).

[14] Department of Justice/Federal Trade Commission, 1992 Horizontal Merger Guidelines, Section 2.1.

[15] It is also important to note that if one were to pursue a coordinated effects theory, assigning market shares in order to assess concentration would be very difficult. Gilbert and Sunshine suggest using existing product sales as a proxy. This seems unlikely to accurately reflect ability or likelihood to innovate in many cases. In many industries, smaller, less established competitors have been the engines of innovation and the essence of innovation is to change the products in the marketplace, often resulting in substantially different market shares. Alternatively, Gilbert and Sunshine suggest, as does Commissioner Varney, assigning equal market shares to all competitors who are potential innovators. A simpler approach may be to adopt a safe harbor based on numbers of competitors -- the Guidelines to Intellectual Property Licensing suggest a "safe harbor" if five potential innovators exist in the market. William Baxter suggests concern about coordinated effects on innovation only when there are only three potential innovators in the market, See William F. Baxter, "The Definition and Measurement of Market Power in Industries characterized by Rapidly Developing and Changing Technologies," 53 Antitrust L.J. 717 (1984).

[16] See, e.g., United States v. Baker Hughes, 908 F.2d 981 (D.C. Cir. 1990).

[17] Gilbert & Sunshine, supra note 5, at 597, Varney, supra note 2, at ___ and Yao & DeSanti, supra note 3, at 520. Such efficiencies might include economies of scale and/or scope in research and development, combination of complementary ideas or technology, elimination of duplicative research and development, etc.

[18] Memorandum in Support of Plaintiff's Motion in Limine relating to Efficiencies, United States v. Archer-Daniels-Midland Co., 1991-92 Trade Cas. (CCH) para. 69,647 (S.D. Iowa 1991).