United States: Merger Policy & Innovation

Do any of you remember the 1951 British film called The Man in the White Suit? Alec Guinness played the role of Sidney Stratton, a laboratory dishwasher in a textile mill. He invents a fabric that never gets dirty and can't be harmed. But his reward for this innovation is an enormous group boycott by all of the suit makers, textile firms, tailors, dry cleaners, soap companies, labor unions, and just about everyone else. In the end, it turns out that the fabric disintegrates over time -- to the delight of all the representatives of the status quo.

I always thought the Guinness character had a pretty good refusal to deal antitrust case. But that was in 1951. Today, as we all know, the Supreme Court would toss his case out for lack of antitrust injury. Any sequel would go direct to video.

The Man in the White Suit recognized that, although innovations unambiguously benefit consumers, they may seriously injure incumbent firms and that incumbents, therefore, may have strong incentives to take steps to retard or prevent innovation. At the very least, that has been the view of the Justice Department and FTC in a number of merger cases brought in the 1990s. The agencies have brought over a dozen cases based at least in part on the concept of "innovation markets" -- that is, markets where the existing competition consists of research and development rather than the production or sale of a commercialized product. In our time together today, I will discuss a few of these recent cases and then analyze the policy questions they raise.

The Cases

In several recent cases, the FTC has challenged drug company mergers involving products in relatively early stages of development, well prior to commercial marketing. In American Home Products/Cyanamid, for example, one of the markets involved was the research and development of vaccines for rotavirus, a potentially serious diarrheal disease affecting children. The Commission required AHP to license the Cyanamid research. More recently, the Commission issued a consent in the Upjohn/Pharmacia matter. This one involved two potential chemotherapy treatments for colon cancer. The consent requires a divestiture of the research and development assets relating to the Pharmacia drug. The market is several years away.

Another recent case -- Sensormatic/Knogo -- involved competing technologies for anti-theft labels that can be attached by the manufacturer, another product that does not yet exist but is in development. Here, the Commission's decree effectively converts Sensormatic's proposed acquisition of the patents into a nonexclusive license.

A few of the new cases have focused more on development and design competition in existing markets than on potential competition in future markets. Probably the best example of this kind of case is the GM/ZF deal in 1993. In that one, GM and ZF dominated the global market for the development and design of automatic transmissions for medium and heavy duty vehicles. Justice sued to block the deal and the transaction was abandoned.

The other matter I want to mention has not resulted in any case. That is the Justice investigation of Microsoft's bundling of the new Microsoft Network into Windows 95. Firms like American Online were arguing that Microsoft was using its operating systems dominance to attempt to monopolize the market for online services. The twist here was the concept of attempted monopolization of a market in which Microsoft, at the time, had no product. In opposing a Microsoft motion to quash one of its subpoenas, Justice endorsed that concept, at least in principle.

New?

Do these cases represent a new policy from the enforcement agencies? Or is it really just old potential competition theory with a new set of buzzwords?

I think much of the policy truly is new. Certainly it is new to think about research, development, and design as relevant markets. We usually think of markets as involving at least some production and some sales. The agencies are now looking at research and development as a type of nonprice competition, and at R&D devoted to a particular type of product as a current, existing market for antitrust purposes. The drug cases and the GM/ZF matter demonstrate that these concepts go beyond pure theory. They are being applied in actual cases. That is unquestionably different from the perspective taken prior to 1988.

Some of the "innovation market" cases have been based less on current R&D than on future effects in markets that do not yet exist. That concept is not new. You can go back as far as 1975 to a business review letter involving the Salk Institute to see that the Justice Department has long been sensitive to potential effects in future markets. But while the concept is not new, the heavy emphasis on future competitive effects is, at the very least, a substantial change in focus.

What is driving this change? We should not kid ourselves and believe that the agencies have developed a new program to encourage full employment for scientists. Despite the new close scrutiny of present R&D and the argument that R&D is an aspect of current market nonprice competition, the real concern underlying the recent cases is not the abstract concept of current competition in research and development. The concern is the protection of competition on price, quality, service, and consumer choice for actual products to be marketed in the future. Absent strong reason to believe that, sooner or later, the merger will impact traditional competitive variables in traditional product markets, none of these cases would have been brought. In that sense, the view that the innovation market approach is really old potential competition theory with a new vocabulary has real support.

The major change effected by the new innovation markets policy is really a matter of degree. The agencies are now projecting analysis of future product market effects beyond the one to two-year time frame we became used to following the 1982 Merger Guidelines. The FTC's Federal Register filing in the Upjohn case makes this very clear. It acknowledges that there is no present market for colon cancer chemotherapy products, but says that this market is expected to exceed the $100 million sales level by the year 2002 -- six years from now.

The Case Law

A good question is whether the current innovation markets approach is consistent with the case law. The best answer is yes and no.

None of the new cases has gotten even near the stage of a judicial or Federal Trade Commission ruling. Most of the cases have resulted in consent decrees. In a few cases, the transactions have been abandoned.

The case most closely on point is SCM v. Xerox, decided by the Second Circuit in 1981, and that case is generally contrary to the new approach being taken in Washington. The SCM decision held that Xerox's acquisition of the controlling xerography patents in 1956, four years prior to the commercial introduction of the Xerox machine, could not be challenged under the antitrust laws. The court said that the policies of the patent laws preclude antitrust liability for a patent acquisition occurring prior to the emergence of a relevant product market.

The SCM case is clearly inconsistent with the proposed complaint in Sensormatic, which specifically involved an acquisition of patents for a market that does not yet exist. SCM is also at least in tension with several of the other recent cases, such as Upjohn and Cyanamid, which involved stock acquisitions or mergers providing access to patentable technologies.

But despite SCM, the agencies' recent cases are not inconsistent with the broader fabric of the case law. A reasonable case can be made that the new policy is the General Dynamics doctrine viewed from a different perspective. General Dynamics establishes that present production statistics may overstate future production and market shares. By the same token, present production statistics may also understate future production and market shares. That is the premise of the agencies' new innovation policy.

Policy

Is this new policy approach a good thing?

The answer, in general, should be yes. But there are real dangers too. And the agencies must be cautious.

The key point is the protection of competition in future markets. No one seriously disputes the importance of that objective. The disagreements, rather, are really matters of degree as to how aggressively the new approach should be pursued and how far out in the future we should go to predict competitive consequences.

1. One of the main arguments for a cautious approach is that there is a lack of empirical evidence supporting a correlation between high R&D concentration and decreased output of innovations. Many forceful arguments have been made that monopoly is more conducive to innovation than competition, as the work of Joseph Schumpeter amply attests. But this point can be argued effectively both ways, as the more recent debate between Gilbert and Rapp in Antitrust Law Journal illustrates. (The arguments are eerily reminiscent of the intellectual debate concerning the wisdom of antitrust legislation in the 1880s.)

Whatever the outcome of the empirical debate, it is hard to expect the Antitrust Division and Federal Trade Commission to base an antitrust enforcement policy on the proposition that less competition is better than more or, put differently, that fewer innovators are better than more. If the agencies are taking the view that a threatened increase in concentration of research and development sources in a particular market warrants scrutiny and possible challenge, we really have to acknowledge that that is their job.

2. Another argument is that identifying an R&D market is extraordinarily difficult. This is a serious objection. Defining the area of research involved and identifying the participants and potential entrants are very difficult tasks. And the closer we get to pure research, the harder that job becomes. Ideas are one type of commodity that can never be monopolized.

Where the R&D is much closer to the "D," it makes a lot more sense to attempt to define a market, identify the participating firms and potential entrants, and to assess probable competitive impact. And any resulting case will make more sense. Where the case is more at the "R" stage, aggressive antitrust intervention is difficult to justify.

This seems to be the approach the agencies are taking. Sensormatic and the drug cases are examples of cases where the products in issue were in the development stage, not the research stage. GM/ZF involved actual and measurable current competition in design. I don't know of any case where the issue was pure research rather than development, and I'd be surprised if any case like that were ever brought.

3. Another criticism of the innovation market cases goes to the relief frequently obtained. In some -- not all -- of the cases, the relief is licensing rather than divestiture. That relief can be counterproductive by diminishing the licensor's incentives to continue the product's development. It's a fair comment that many cases would be better off if not brought at all than settled with a licensing decree. To me, the agencies' frequent imposition of licensing relief is the one aspect of present enforcement policy that most requires further thought.
4. The comment most frequently advanced in criticizing the innovation market approach is that future effects are speculative. Predicting competitive impacts in existing markets is tough enough, and it becomes really hard when the market is changing rapidly or has not yet come into existence.

The Windows 95 investigation provides an excellent example. When the investigation began, it was not unreasonable to fear that automatic access to MSN in Windows 95 would enable Microsoft to monopolize the online services market. A number of commentators in the trade press were predicting exactly that result.

Less than a year later, the world has changed completely. Not only has MSN pretty much turned out to be a dud -- at least so far -- but it's now doubtful that online services even represent a relevant product market. That entire industry is rapidly being overtaken by the Internet generally and the World Wide Web in particular.

But despite these problems, it would be unsound for the agencies to ignore or downplay reasonably foreseeable future competitive effects. Merger analysis is always based on predictions of the future and, when we move into prospective relevant markets or evolving markets, the change in policy is really only a matter of nuance and emphasis.

One way to reduce the potential for error in cases involving these types of future effects is to concentrate on cases where the future market is going from two competitors to one or from three to two. If the case involves a six to five or even a four to three, the agencies should forget about it, at least in the absence of a more traditional showing of impending competitive harm.

If we look at the actual cases that have been brought, the agencies have been basically on target. The Windows 95 investigation involved matters of real speculation, but Justice hasn't brought a case. The cases actually brought have involved markets going largely from two to one, and the products were fairly well along in development. These are really pretty good cases, even if the relief obtained in some is open to doubt. Cases like Upjohn in particular are likely to provide important economic benefits. Instead of one drug to treat colon cancer, we may have two. That should lead to more cures and lower prices. Even the textile workers in The Man in the White Suit would go for that.