The Misapplication Of The Innovation Market Approach To Merger Analysis

Richard T. Rapp[*]

I. INTRODUCTION

The Department of Justice and the Federal Trade Commission have recently assumed the responsibility for protecting competition in research and development via the application of a new policy towards merger review under Section 7 of the Clayton Act[1] and Section 5 of the Federal Trade Commission Act.[2] The new policy creates a separate procedure for the analysis of R&D effects; the procedure parallels the price-effects-oriented procedure of the 1992 Horizontal Merger Guidelines. T[3]he analysis of relevant innovation markets under this new policy is already well integrated into both agencies' merger review procedure, and is evident in complaints, consent decrees, speeches and writings of agency officials.[4]

Although the definition of an innovation market as a relevant market for antitrust purposes has an antecedent in the "R&D market" concept that appeared in the 1984 National Cooperative Research Act[5] and in the 1988 International Guidelines,[6] the "innovation market approach" to merger analysis is largely novel in its aim and approach. The aim of the new policy is to introduce dynamic efficiency considerations into merger enforcement, to recognize the importance of innovation as a means of nonprice competition and a source of welfare gains, and to prevent mergers that would reduce competition in innovation. Since innovation is hard to measure, the innovation market approach focuses on R&D, the primary innovation-creating activity of modern firms.

In most applications, the innovation market approach is merely superfluous - a new way of talking about potential competition. In some instances, however, the innovation market approach represents a leap into the unknown with a potential for harm to economic welfare as great as any potential benefit. Neither economic theory nor any factual analysis of the connections between market structure, R&D and innovation provides a persuasive basis for the innovation market approach, the likes of which now appears to be accepted practice at DOJ and certainly at the FTC. Increases in R&D concentration may or may not lead to less R&D and cutbacks in R&D may or may not affect future prices and output quantities in product markets, which are (or should be) the ultimate concern of antitrust. When the enforcement agencies require merging firms with overlapping research projects to license or divest, they cannot know whether their actions will help or hinder innovation.

This article begins with a brief description of the sources and early applications of the innovation market approach, as well as an overview of the proposed sequence of steps - which parallels the 1992 Horizontal Merger Guidelines - for using innovation markets to assess the implications of a merger for R&D (Part II). For the latter, I rely upon Gilbert and Sunshine's article in this Journal,[7] which also lays out the thinking behind the approach and discusses the underlying economic literature with candor and balance. My summary of the supporting argument and my critique of the approach appear as Part III.

II. "THE INNOVATION MARKET APPROACH"

A. Origins and Early Applications

The earliest directive that relevant antitrust markets be defined around research and development activities can be found in The National Cooperative Research Act of 1984, which mandated that cooperative research be subject to an evaluation of competitive effects "in properly defined, relevant research, development, product, process and service markets."[8] The 1988 International Guidelines followed this approach in Case #6 on R&D joint ventures, according to which research joint ventures were to be evaluated by DOJ in R&D markets as well as technology or downstream product markets. The 1988 International Guidelines defined a relevant R&D market to include all firms that "...appeared to have the incentive and ability, alone or cooperatively, to undertake R&D comparable to the R&D proposed to be undertaken by the joint venture..."[9] In this discussion, the 1988 International Guidelines recognized the inherent difficulty and qualitative nature of R&D market analysis, noting that in some instances a successful R&D effort might only be accomplished by cooperation amongst a majority (or even all) of the R&D market participants and that concentration in the downstream product market does not necessarily imply the same for the R&D market.[10]

R&D markets were defined and employed by the FTC in Roche-Genentech (1990).[11] In that case, the FTC opposed Roche's acquisition of a controlling interest in Genentech, alleging a lessening of competition in the research, development, production and marketing of (1) vitamin C, (2) human growth hormone, and (3) CD4-based AIDS and HIV treatments. At the time of the challenge: (1) Roche manufactured vitamin C while Genentech had developed a new patented process for producing vitamin C but had not yet done so; (2) Genentech had a human growth hormone on the market and Roche had a competing product in clinical trials; and (3) both firms were researching CD-4 based AIDS/HIV treatments but neither had a product approved for sale. The eventual consent decree required divestiture of Genentech's vitamin C interests and Roche's human growth hormone businesses including both existing technology and R&D assets.

The first application of the current innovation market approach occurred in 1993 when the DOJ opposed the merger of the Allison Transmission Division of General Motors and ZF Friedrichshafen, AG,[12] the world's two largest manufacturers and innovators of medium and heavy automatic transmissions for trucks, buses and other commercial and military vehicles. The complaint alleged that the GM-ZF combination would diminish competition not only in several relevant product and geographic markets but also in a worldwide innovation market encompassing the technological design, development and production of such automatic transmissions. The production facilities of the merging firms were characterized as specialized assets - the only sites for testing innovations in this market.

In 1994, the DOJ opposed the merger of Flow International and Ingersoll-Rand's (IR) Waterjet Cutting Systems Division on similar grounds.[13] The merging firms were the two dominant manufacturers in the United States - the relevant product market included a small competitive fringe - of ultra-high pressure waterjet intensifier pumps, a key component of the waterjet systems used in industrial cutting and cleaning applications. The complaint alleged that competition between the two firms extended into R&D and technological innovation, both of which the DOJ contended would be significantly reduced by the merger.

In both GM-ZF and Flow International, product market considerations - e.g., unfavorable entry conditions, a lack of substitute products, high market shares for the merging firms, significant increases in the post-acquisition HHIs - gave the DOJ sufficient reason to oppose the mergers without resorting to innovation market considerations. After the filing of the complaints, both mergers were quickly abandoned by the parties.

Since the release of the 1995 Intellectual Property Guidelines in draft in August 1994 and in final form in April 1995,[14] the FTC has leveled complaints against several additional mergers on the grounds that innovation markets would be harmed and has obtained relief either in divestiture or compulsory licensing. These include:

1. American Home Products-American Cyanamid;[15]
2. Glaxo-Wellcome; [16]
3. Sensormatic;[17]
4. Wright Medical Technology;[18]
5. Boston Scientific;[19] and
6. Royal Dutch Petroleum-Montedison.[20]

Thus, the application of the innovation market approach in merger review had a fairly lengthy gestation period. Its current status in agency procedure is the result of progressive adoption. It isn't a hasty experiment.

B. Innovation Market Analysis

The 1995 Intellectual Property Guidelines govern agency review of licensing practices, not mergers. Judging by Gilbert and Sunshine, and complaints brought by the agencies, the definition of "innovation market" spelled out formally in the 1995 IP Guidelines is being used in the merger application as well. According to the 1995 IP Guidelines, innovation markets consist of:

...the research and development directed to particular new or improved goods or processes, and the close substitutes for that research and development. The close substitutes are research and development efforts, technologies, and goods that significantly constrain the exercise of market power with respect to the relevant research and development, for example by limiting the ability and incentive of a hypothetical monopolist to retard the pace of research and development.[21]

In other words, by analogy to the 1992 Horizontal Merger Guidelines' reference to the power to increase price as the manifestation of market power in goods or technology markets,[22] the 1995 IP Guidelines' definition of "innovation market" and the "innovation market approach" proposed by Gilbert and Sunshine treat cutbacks in R&D expenditure as the expression of market power in innovation markets.[23]

Gilbert and Sunshine propose a procedure that an enforcement agency would follow to determine whether a merger should be challenged on grounds that it would enhance market power in an innovation market. The procedure they describe consists of the following five steps:

1. Identify the Overlapping R&D Activities of the Merging Firms

Define the relevant innovation market by identifying the economically relevant overlapping R&D activities (defined as those activities which may lead to improved products or processes of the firms in question.)[24] Only those endeavors which are likely to have an impact in one or more downstream product markets will be included in a relevant innovation market.

2. Identify the Alternative Sources of R&D

Identify other market participants - firms which are sources of substitute R&D activities - including both firms with current R&D capabilities and firms which might enter the R&D race within a short period of time.[25] Only R&D projects for which specialized assets (such as production or testing facilities, or technical knowledge) are required in the R&D process are to be subject to the innovation market approach.[26]

3. Evaluate Actual and Potential Competition from Downstream Products

Downstream product market competition is a check on the willingness of merging firms to reduce R&D. If the effects of lost competitive advantage downstream exceed the savings from reduced R&D expenditures, then merging firms have no incentive to downsize and there is no consequent danger in the merger.

4. Assess the Increase in Concentration in R&D and Competitive Effects on Investment in R&D

Measure the merging firms' share of the relevant market using actual R&D expenditures, production or assets of the individual firms, or equal shares when all competitors in the relevant innovation market are equally likely to have effective R&D programs.

5. Assess R&D Efficiencies

Evaluate whether combining the merging firms' R&D programs is likely to create efficiencies in R&D, such as eliminating redundant R&D effort, gaining complimentarities or economies of scale, or otherwise enhancing the prospect for successful innovation.

III. THE CRITIQUE OF MERGER ENFORCEMENT IN INNOVATION MARKETS

Adding a new type of market power analysis to merger review increases the cost of enforcement both to the agencies (whose investigations must be more extensive) and to merging firms in research-intensive industries (whose investigational responses must be more extensive). There should be good reasons for taking such a step. Gilbert and Sunshine outline the reasoning behind the use of the innovation market approach in merger review, which can be summarized as follows.[27]

First, innovation is the main engine of modern economic growth. U.S. economic growth in the 20th century has been due less to increases in labor and capital inputs than to shifts in the aggregate production function, i.e., the economy's ability to extract more and more from the same inputs. The source of welfare gains is technological change, and technological change comes from innovation. These points are fundamental and uncontroversial.[28]

Second, innovation happens in the marketplace.[29] The desire to commercialize and capture the gains from invention motivates firms to invest time, capital and entrepreneurial effort in innovation. Innovative new products or product characteristics are as effective and welfare-improving a means for winning customers in competition as cutting prices.[30] Therefore, many firms compete by inventing new products and product characteristics, as well as cost-reducing processes. Innovation benefits consumers and producers, and more innovation is better than less innovation.

Third, if the capacity to innovate in a relevant market were monopolized, or if steps were taken in that direction, competition in that market would be lessened substantially and consumers would be harmed. A merger that reduced the level of competition in innovation would harm consumers because reduced competition in innovation affects prices, product characteristics and the rate of new product introductions in downstream product markets. Cost-reducing production processes might not be invented or put into practice. New products and improved products might be late to market or never emerge at all.

Fourth, not only is it possible to lessen the capacity to innovate in a market by merger, but firms may have an incentive to do so.[31]

Fifth, market power in innovation markets may mean cutting back R&D to a sub-competitive level for the sake of increasing the present value of profits. This assertion implies that innovative activity can be equated with R&D, thereby justifying intervention against mergers whose possible effect may be to cause a "small but significant non-transitory reduction in R&D."

Sixth, a new approach is needed to deal with mergers that harm innovation, since conventional product (or technology) market analysis under the 1992 Horizontal Merger Guidelines is inadequate in the face of these problems. To begin with, the innovation market approach facilitates analysis of the effects of a merger or acquisition on potential competition. According to Gilbert and Sunshine, "a merger or acquisition that adversely affects innovation... may reduce the probability of entry into and the intensity of competition in markets where the merging firms do not compete prior to the merger."[32] More significantly, Gilbert and Sunshine argue that these effects may arise in product markets in which the merging firms do not compete before the merger and may not even be potential competitors.[33] Therefore, they hold that the innovation market approach covers an important gap in merger analysis.

In sum, the innovation market approach is motivated by the desire to account for the importance of nonprice, technological competition in merger reviews, thereby protecting the dynamic efficiency of the economy, especially in high technology industries. Regrettably, the innovation market approach does not live up to these worthy intentions.

A. The Lack of Connection Between Concentration and R&D or Innovation

The utility of the innovation market approach, as outlined in the sources discussed above, depends upon the validity of the following statements concerning a relevant market of substitute R&D projects:

1. An increase in R&D concentration is likely to reduce the amount of R&D undertaken.

2. Reducing the amount of R&D is likely to diminish innovation.

There is little basis in fact or theory for these statements. The connection between market structure and innovation has been debated by economists for decades without resolution.[34] Competition is a powerful incentive to innovate, the better to gain cost saving or new product advantages over rivals. On the other hand the price competition associated with unconcentrated, highly competitive markets tend to drive innovative activity to suboptimal levels.

Innovation is intangible, uncertain, unmeasurable and often even unobservable, except in retrospect. There are no market transactions in innovation, only in the inputs - the labor of scientists, engineers and entrepreneurs and the capital to build research facilities and to fund R&D in advance of returns - and the outputs - technology and products. Thus, R&D expenditure is an input to the process of innovation.[35] When we use the term "innovation market," but measure market shares in terms of R&D expenditures or R&D capacity,[36] we are making either an error or a leap of faith. The error would be to suppose that innovation and R&D are the same thing. The leap of faith is to believe that there is a positive functional relationship between the rate of R&D expenditure (or the amount of R&D capacity) and the quantum of innovation produced by a firm.

Gilbert and Sunshine make the leap in the following way:

Innovation generates new products that consumers can enjoy, and consumers will buy the new products only if they provide net positive surplus. Thus, consumers are strictly better off with more product innovation, provided that there is no reduction in the supply or increase in the price of other products or services. Consumers are also better off, or at least no worse off, when innovation leads to less expensive products or cheaper means of manufacture.

It is conceivable that firms might squander resources in costly innovation races that lead to increased costs. Wasteful innovation competition could translate into higher prices that consumers must ultimately pay to subsidize costly research budgets. The available empirical evidence, however, confirms the conventional wisdom that society is better off with greater investment in research and development.[37]

Gilbert and Sunshine acknowledge the leap from preserving a rate of innovation to preserving R&D competition to preventing a merger of R&D capacities, but regard it as only a "complicating factor," not a fatal flaw in the chain of reasoning:

The lack of a deterministic relationship between R&D expenditure and innovation makes it more difficult to link market structure and the pace of technological innovation. While R&D spending may be necessary for innovation, it is anything but sufficient. This is especially true of more fundamental innovations.[38]

1. Findings on Concentration

Numerous research projects have been organized to test variations on the "Schumpeterian hypothesis" - that monopoly is more conducive to innovation than competition. These studies have sought to find statistical relationships between firm size or market concentration and various measures relating to R&D and innovation, including R&D expenditure, R&D productivity, patent counts and counts of significant innovations. In his survey of these studies, Scherer concludes that the effects of firm size and concentration on innovation appear not to be important.[39] In other words, if one observes a decrease in the number of firms engaging in related R&D projects, one cannot predict whether total R&D activity in the market or innovation in the market will increase or decrease as a result. In some cases, concentration reduces inefficiency and permits a larger scale of effort without diminishing the incentive to bring a successful project to market. In other settings where R&D competition is an important propulsive force, reducing the number of players may slow down the pace of the game. There are no principles for predicting the outcome from economic data about the market in question.

Advocates of the use of innovation markets in merger enforcement may argue that the inconclusiveness in the literature about a causal relationship between concentration and innovative activity is no more disquieting than the inconclusiveness of an older literature on the connection between market structure (mainly concentration) and market performance (profitability).[40] Yet the 1992 Horizontal Merger Guidelines, which are currently accepted by economists and antitrust practitioners as offering the best approach for defining markets and detecting the potential for market-power enhancing mergers, do not depend upon regression studies of HHIs on price-cost margins for their economic integrity. Instead, they depend upon a plausible behavioral theory of oligopoly, especially as it concerns information and the maintenance of collusive arrangements not to compete: more pricing voices in a relevant market comprising only sellers of good substitutes makes it harder for those sellers to raise price by agreement.[41] This theoretical underpinning is most apparent in Section 2 of the Guidelines, "The Potential Adverse Competitive Effects of Mergers," which discusses inter alia how coordination may be achieved and policed. Application of the theory of oligopoly coordination explains why certain merger-induced increases in concentration threaten to restrict output and harm consumers. No such causal explanation connects mergers which concentrate R&D expenditures to reductions in R&D output. The incentive to collude to reduce R&D output is not presumed to influence innovating firms as broadly as the incentive to fix prices influences most sellers. The secrecy of R&D, the improbability of effective coordination and greater certainty of gains from chiseling, all tend to destabilize R&D restricting agreements.[42]

Neither does theory offer any guidance for distinguishing between the rationalizing R&D cutbacks two merging firms might make and those intended to reduce R&D to subcompetitive levels. There is, therefore, no basis for controlling R&D concentration in the course of merger review under either single-firm monopoly theory or Stiglerian oligopoly theory - the current basis for horizontal merger enforcement policy.

2. The Treatment of Research Joint Ventures

If firms have the incentive and ability to exercise market power in a concentrated innovation market by combining to reduce competition in commonly-pursued R&D projects, then a research joint venture in a such a market is a cartel, pure and simple. Yet rather than being the subject of suspicion, research joint ventures are seen as worthy of encouragement. For example:

Joint research ventures, like production joint ventures, often provide procompetitive benefits. These benefits include sharing the substantial economic risks involved in research and development (R&D); increasing economies of scale beyond what individual firms could realize; pooling important information or complementary skills; and overcoming the free rider disincentive to invest in R&D by including likely end users of the R&D in undertaking the research efforts and sharing the costs. For this reason, the antitrust enforcement authorities have long viewed joint research ventures more favorably than many other types of joint ventures. Congress codified this more lenient approach in the National Cooperative Research Act of 1984 (NCRA).[43]

Two years before the passage of the NCRA, Ordover and Willig wrote an influential article advocating special, favorable treatment of research joint ventures. Starting from the premise that the relationship between concentration and R&D investment and technological change is unknown, they reasoned that "...significant losses in dynamic efficiency may be caused by the application of antitrust laws, traditionally aimed at preventing harmful concentration of static market power, to R&D-intensive business combinations such as mergers in high-technology sectors or research joint ventures."[44] They concluded that research joint ventures without ancillary restraints on product market competition almost certainly increase innovation and product market competition, as long as the primary R&D competition comes from other firms, rather than from the collaborating firms. In addition, a research joint venture would harm R&D competition only if post-development merger and withdrawal of one product would yield large profit gains.[45] The extension of their analysis to mergers leads Ordover and Willig to conclude that mergers tend to foster the incentives that encourage R&D and provide efficiencies unavailable from research joint ventures.[46] They argue that a merger should be allowed if:

(1) but for innovation by the candidates, the merger would not be challenged under static merger analysis within a relatively short span of time; and
(2) at the time the candidates would otherwise be expected to bring their innovations to market, a merger between them would go unchallenged.[47]

3. Cooperation Outside Research Joint Ventures

In a series of articles, Jorde and Teece argue that because of the special nature of innovation in high technology industries, antitrust strictures against horizontal collaboration impede progress.[48] In certain high technology settings, cooperation not only in research - including formal joint ventures, strategic alliances and other combinations designed to bring together R&D and other activities of rivals - but also in downstream manufacturing and distribution may actually be beneficial to consumers. According to Jorde and Teece, successful innovation including both invention and commercialization - often requires that firms cooperate to compete successfully. The reasons include: the scale and scope of investment, talents and assets may be too much for one firm to handle; the financial risk may be too great for one firm to bear; one firm's access to complementary technologies may be incomplete; and in industries where patents (or other intellectual property) are ineffective, sharing the risks and rewards with rivals may be the best way to capture gains from innovation.

Given these factors, Jorde and Teece argue that U.S. antitrust laws inhibit cooperation that would aid innovation and the diffusion of new technology. The problem, as they see it, is that the rule of reason standard by which a cooperative agreement would be judged is blurry and therefore dangerous for firms engaging in procompetitive cooperative ventures. The threat of a treble damage judgment weighs against the uncertain gains that might arise from procompetitive cooperation.[49]

While the policy prescriptions of the Jorde and Teece papers do not diametrically conflict with the innovation market approach, the underlying point of view is fundamentally at war with it. In the innovation market approach, borrowing from the 1992 Horizontal Merger Guidelines, it is the small but significant nontransitory reduction in R&D arising from a merger that results in the lessening of competition that harms consumers.[50] Market power in innovation markets is exercised by R&D cutbacks in the same way that output restriction in product markets leads to consumer harm. Cooperative arrangements - except those which arise between firms with small shares in relevant innovation markets - almost by definition reduce the sort of R&D competition that the innovation market approach intends to foster. If Jorde and Teece are right that cooperative R&D arrangements by firms of consequential size in their relevant innovation markets are likely, on balance, to be efficient, then the foundations of the innovation market approach must necessarily be shaky.[51]

In fact, the problem with the thinking favoring permissiveness toward R&D cooperation is fundamentally the same as the problem with the innovation market approach - neither has any real theoretical underpinnings, therefore neither one has any real explanatory or predictive power. Both proceed from the proposition that a static view of rapidly changing R&D-intensive industries may lead to error. This observation comes trippingly off the tongue and it may be right some of the time. But what is the dynamic view? That in innovative activities there are important scale effects, complementarities and recursive processes that call for cooperation or warrant higher concentration than the static view would allow? Or that such consolidation results in reduced incentives to innovate? Neither of these holds any water as a useful generalization although each may be apt in some specific setting. There is no general rule to cover the relationship between market structure and innovation. What may be a remedy in one innovation market may be the undoing of another.

B. Lack of Connection Between R&D and Innovation

Although R&D is a primary innovative activity, more R&D is not necessarily better than less R&D. In fact, there is no functional relationship between the level of R&D expenditure and the level of innovation at the market level.[52] Market factors, including diminishing returns to R&D investment, the opportunity cost of funds and expectations about the appropriability of the gains, determine how much firms in a market will spend on R&D. Reductions in the demand for R&D and increases in its cost will cause firms to cut back R&D without harm to competition. It is therefore wrong to stop mergers which reduce R&D using simple rules such as those which derive (by analogy) from the appropriately simple rules used to stop price-increasing mergers.

On average, the social returns to R&D exceed the private returns. This tells us that society gains, on average, when individuals invest in R&D. It is also true that, in theory, more inputs imply more output so that a firm which puts in more R&D will get more innovation out, ceteris paribus. These insights are an inadequate basis for innovation-oriented merger policy, however, because they abstract from quality considerations, duplication of effort, waste and bad decisions. Industry-level or cross-sectional studies fail to yield evidence of a functional relationship between R&D expenditure and innovation precisely because of these factors. The simple theoretical link between R&D expenditure and innovation gives way to the reality of poor ex post returns to projects that looked good ex ante. In merger enforcement, even if one could predict with confidence that a merger would lead to a reduction in R&D expenditure, a principled basis for saying whether this is good or bad for competition and innovation does not exist.[53]

The evidence from financial markets is that firms often invest in projects - including R&D projects - which have negative discounted cash flows, i.e., "losers." One reason is imperfect foresight: managers cannot always tell beforehand which investments will pan out and which will not. Miscalculation and dashed hopes, however, are not the only reason for wasteful R&D investment. Managers have incentives to invest in growth beyond the point where the firm's investors benefit, for the sake of increasing the resources under their control. This agency cost of the separation of ownership and control is disciplined by competition in product markets. "However, product and factor market disciplinary forces are often weaker in new activities and activities that involve substantial economic rents or quasi rents."[54]

The scale of wasteful investments can be massive. In his measurements of the productivity of R&D and capital spending by large U.S. firms during the 1980s, Jensen found that many billions of dollars were misspent. For example:

It is clear that GM's R&D and investment program produced massive losses. The company spent a total of $67.2 billion in excess of depreciation in the period and produced a firm with total ending value of equity (including the E and H shares) of $26.2 billion. Ironically, its expenditures were more than enough to pay for the entire equity value of Toyota and Honda, which in 1985 totaled $21.5 billion.[55]

Undoubtedly virtuous-sounding R&D programs were part of the story at GM and a long list of other firms that would have benefitted shareholders by increasing dividends instead of overspending on unproductive research.[56]

Imperfections in innovation and technology markets also create situations where the level of R&D may be excessive. Because the first to win a patent has the right to exclude others from making, using or selling the patented technology, the system creates "winner take all" patent races. This leads to overinvestment relative to the social optimum because "...competing researchers care about who wins a patent race, whereas society is only interested in having some unit succeed."[57] In addition, resources devoted to R&D may be wasted in competitive innovation markets simply because of excessive duplication and haste in research taking place in small markets with numerous competitors.[58]

Market factors determine how much firms spend on R&D. For example, an increase in the cost of capital or in the attractiveness of alternative investments for any firm would reduce its incentive to maintain R&D spending levels. Moreover, as Gilbert and Sunshine recognize, the appropriability of returns varies widely from industry to industry and investment in R&D is affected by the appropriability regime.[59] When appropriability is imperfect, R&D competition reduces the gains to any firm from innovation and renders projects unprofitable (not worth undertaking) that might be worthwhile if protection were more secure.[60]

Although economists sometimes like to assume that nearly perfect capital markets make funds available to support any project whose expected return exceeds its cost of capital, we observe that firms prefer to finance investment out of cash flows.[61] A decline in a firm's cash flows or operating profits signals a reduced supply of investment funds and, possibly, a diminished need to invest (if the present is a reflection of what the future holds). This is a common enough setting in which adversity, including increased competition, may lead to reduced R&D expenditure in a manner we would deem entirely consistent with competition.

Indeed, reduced R&D expenditure originating this way may be caused by heightened competition. For example, over the past several years, price pressures on the pharmaceutical industry in the U.S. have intensified owing to the increase in downstream competition among health care institutions - especially among HMOs and other managed care entities whose success is tied partly to success in cost reduction. Their bargaining power, owing to the control they exert over the prescribing practices of affiliated or employee physicians, has cut substantially into the margins of most of the drug manufacturers. In this environment, some research projects that would have been profitable no longer will be.[62] Rational drugmakers in competition can be expected either to pare these projects from their portfolios or to continue the project as a shared costs/shared gains joint venture. In addition, given that the portfolios of research projects of firms overlap,[63] merger is one of the devices by which efficient research cutbacks can be made. Therefore, as a result of competition and as part of a process with no anticompetitive causes or consequences, R&D expenditure gets cut back even in highly research-dependent industries like pharmaceuticals.

There is no basis either in the innovation market approach or in formal economics for segregating such decisions by firms to reduce R&D funding into procompetitive and anticompetitive classes. Since there is no principled way to distinguish good cutbacks from bad ones, an innovation market approach, even performed on a studious case-by-case basis would be a mistake. A far worse mistake would be to proceed on this basis of "We will know it when we see it." This is the worst sort of enforcement policy because it both leads to bad decisions by the agencies and prevents firms from making good judgments ex ante about how the agencies will react to any specific case.

C. The Capacity to Innovate is Hard to Monopolize

There is little in the history of modern technology to suggest that firms are able to monopolize innovative capacity. Although patented technology can be monopolized, the piece-parts of modern R&D - research scientists, engineers, software developers, laboratories, computer centers, etc. - are continuously on the market for sale. If the main inputs to innovation are continually 'in play', there is no opportunity to corner the market for innovation.

Although the 1995 IP Guidelines do not explicitly discuss the fact that, in general, innovative capacity would be hard to monopolize, they implicitly address the issue in the section which limits innovation market inquiries to cases where the merging firms own unique facilities:

The Agencies will delineate an innovation market only when the capabilities to engage in the relevant research and development can be associated with specialized assets or characteristics of specific firms.[64]

Gilbert and Sunshine also advocate this approach:

The delineation of innovation markets should be limited to markets in which R&D directed toward particular new products or processes requires specific assets that are possessed by identified firms. If innovation directed to particular products or processes does not require specific assets, entry into R&D would be easy and the innovation market would be competitive.[65]

Therefore, such "specialized assets or characteristics" are distinguished partly by the fact that they are not up for grabs to competitors.

No doubt R&D sometimes depends upon unique facilities, without which technological progress could not be made. For example, the GM-ZF complaint plausibly contends that to compete in research on medium and heavy automatic transmissions requires a factory in which experiments can be put into practice, thereby limiting the innovation market to the two merging firms.[66] The hard question is: What constitutes a "specific asset"? Can the enforcement agencies argue that experience, for example, is a specific asset, implying that the research establishment of any firm can be said to have unique advantages if it is first or furthest down a particular research path? More importantly, how do the enforcement agencies intend to identify cases where specific assets in the possession of the merging firms constrain competition? And absent any clear policy, how does a firm anticipate the agencies' judgment about this fairly crucial issue?

The worry is not that the agencies will be capricious without formal guidelines, but rather that the "specialized assets or characteristics" prerequisite will simply be ignored. In fact, most of the complaints in FTC innovation market cases do not identify the specialized assets which triggered the innovation market challenge.

D. Product and Technology Markets are Sufficient

The innovation market approach would represent a useful improvement upon merger analysis only if the existing procedures - as codified in the case law and the 1992 Horizontal Merger Guidelines - were inadequate. Gilbert and Sunshine, in fact, argue that existing merger analysis seriously needs the supplement of the innovation market approach. The main weakness they identify in conventional merger analysis concerns cases in which a merger leads to anticompetitive effects in product markets which are remote from the existing product markets of the merging firms. In addition, they contend that in certain settings a conventional analysis might fail to pick up harm to competition in product markets which are directly downstream from the merging R&D.[67]

Gilbert and Sunshine offer the example of a duopoly of R&D-performing vertically integrated sellers. Each firm makes an upstream product (ingot), as well as one downstream product which is sold in competition (cable) and another downstream product which is sold by each as a monopoly (lawn chairs) in geographically separate home markets. Conventional product market analysis of a merger between the two firms reveals increased concentration in the ingot and cable markets but not in the lawn chair markets. No harm to competition can be inferred in those markets. As Gilbert and Sunshine explain, R&D competition in ingot-making may yield cost reductions for all downstream aluminum products, thereby resulting in reductions in the monopoly price of lawn chairs in addition to competitive gains in the cable market. Conventional product market analysis would fail to capture the threat to those gains from concentration.

Note, however, that in this example and cases like it, the improvement in the analysis from the definition of a separate innovation market arises because concentration in R&D capacity differs from the concentration in the product market. What makes the example work is the artificial isolation of the two lawn chair markets. With downstream competition in lawn chairs, we would have to ask: "If the lawn chair makers were numerous and competitive, how is it possible that only two of them have most of the R&D capacity?" If, instead, lawn chair-making were concentrated, then conventional product market analysis will reveal the problems inherent in the merger thereby eliminating the need to resort to an innovation market analysis.[68]

Presumably, one important reason for preventing mergers between firms that are engaged in overlapping R&D projects is that either or both firms threaten to enter the relevant product markets corresponding to the overlapping research area. In other words, R&D makes one or both firms either potential competitors or entrants in conventional product markets. Gilbert and Sunshine contend that the innovation market approach "...facilitates analysis of the effects of a merger or acquisition on potential competition."[69] However, they argue that "[t]hese effects follow from changes in the current state of actual (not potential) competition in innovation markets."[70]

The use of potential competition in merger analysis antedates the 1984 Merger Guidelines.[71] The doctrine of potential competition has traditionally been invoked by agencies and the courts when a merger reduces the number of entrants who could discipline competition in markets in which they do not currently compete. The idea behind the doctrine is that either (1) the acquiring (or acquired) firm's presence and plausible potential for entry into the market has a procompetitive impact on the incumbent's behavior (i.e., a perceived potential competitor); or (2) the firm would likely enter the market at some future time, thereby increasing competition in the market (i.e., an actual potential competitor). Despite differing court opinions regarding application of the theory,[72] two common themes prevail. First, the analysis of potential competition applies only in concentrated markets with dominant participants which have the ability to control output and prices.[73] Second, and related to this, potential competition analysis is relevant only when entry into the market is difficult or potential entrants are not numerous. In cases where these two criteria have not been met, potential competition challenges have failed.[74] These conditions are appropriately applied to the settings in which innovation markets are used. It makes little sense to prevent the acquisition of any potential entrant if there is no shortage of other potential entrants into the product market. Viewing R&D competitors in this light leads to the same conclusion.

Along with the doctrine of potential competition, the analysis of supply response and entry as set forth in the 1992 Horizontal Merger Guidelines is a means by which dynamic effects enter merger review. "Uncommitted entry" covers the case of a firm not yet serving buyers in a product market but which could do so within a year without the expenditure of significant sunk costs; "committed entry" refers to supply responses that take more time or cause greater sunk costs.[75] Although entry analysis most often deals with the prospect of firms that are not parties to the merger, the concept has utility when applied to merging firms too. In cases where both firms are developing products that would compete in an existing market (such as new drugs destined to compete in an old therapeutic category), conventional merger analysis would see them as uncommitted entrants and evaluate the impact of the merger on the HHIs in the product market. Importantly, their capacity to supply the product market, not their R&D expenditure is what gets counted in the measurement of concentration.

A review of some of the complaints and consent decrees incorporating the innovation market approach indicates that the new innovation market approach has not in fact improved upon the analysis in the 1992 Merger Guidelines or traditional potential competition doctrine. We can divide the cases in which the innovation market approach is used into three categories: (1) cases where there is a clear competitive overlap between the merging firms at the product market level and the innovation market analysis is essentially a repetition of the product market analysis; (2) potential competition cases, where one or the other firm is a potential entrant into the product market of the other (with both entrant and incumbent engaging in R&D) and traditional potential competition analysis will generally suffice; and (3) R&D-only cases, where no products have emerged and the only competition subject to analysis is R&D competition.

The GM-ZF and Flow International complaints fall into the first category of cases in which the enforcement agencies use innovation market contentions as gratuitous backstops for product market contentions, rather than as independently useful elements of merger analysis. According to the GM-ZF complaint, "ZF and Allison are each other's main competitors in sales of automatic transmissions for medium and heavy trucks, buses, and other commercial and military vehicles."[76] The merger would leave a combined firm with 78 percent of the product market for transit bus automatic transmissions[77] and 100 percent of the markets for heavy refuse truck automatic transmissions.[78] Moreover, the complaint explicitly states: "...market shares in the Innovation Market can be approximated by the number of units produced worldwide by each manufacturer of medium and heavy automatic transmissions for commercial and military vehicles."[79] Thus, for the purpose of gauging the lessening of competition from this merger, product markets and product market shares are entirely sufficient.[80] According to the Flow International complaint, the merging firms' combined share of sales in the relevant product market equaled around 90 percent.[81] Here, too, the decision to challenge on the basis of the product market alone - without any consideration of the innovation market - was an easy one.

The FTC's complaints in Boston Scientific, Wright Medical Technology and Glaxo-Wellcome fall into the second category of potential competition cases in which the innovation market approach is unnecessary because the harmful effect that is of concern to the agencies is discernable and subject to analysis in the product market itself. The challenged merger between Boston Scientific and SCIMED is an excellent example of a standard potential competition case promoted as innovation market analysis. In a product market for intravascular ultrasound catheters (IVUS), Boston Scientific's and CVIS' combined share approaches 90 percent.[82] SCIMED, which conducted research on IVUS, is merely a potential competitor, as the FTC complaint alleges:

But for its acquisition by Boston Scientific, SCIMED, which has the capacity, incentives and economic interest for entry, is likely to enter the U.S. IVUS catheter market within two to three years. No other firm has an entry advantage similar to SCIMED. SCIMED was perceived by Boston Scientific and others to be a potential competitor in the manufacture and sale of IVUS catheters in the United States.[83]

The relatively short time horizon for SCIMED's expected entry - two to three years - leaves little space for concentrating on something as long term, uncertain and remote as R&D competition.

The challenged acquisition of Orthomet by Wright is another case in which ordinary potential competition analysis would suffice. The FTC alleged a relevant market of orthopaedic hand implants. Wright competed in this market, but Orthomet did not. As the complaint states:

Orthomet is a potential competitor of WMTI [Wright] in the market for orthopaedic implants used or intended for use in the human hand approved by the FDA. WMTI and Orthomet are actual competitors in the market for the research and development of orthopaedic implants used or intended for use in the human hand.[84]

In this case, the acquisition could be barred on potential competition grounds; the innovation market analysis is superfluous.

The same is true of the FTC's complaint concerning Glaxo's acquisition of Wellcome, in which the alleged relevant market concerned the R&D of noninjectable 5HT[1D] agonists, noninjectable anti-migraine drugs. Both firms were engaged in R&D in this field, but Glaxo's Sumatriptan[85] was already in the market and profitable.[86] The acquisition poses a straightforward potential competition problem.

Two FTC cases do not fit neatly into the conventional product competition or potential competition categories. In its complaint threatening to halt American Home Product's 1994 acquisition of American Cyanamid, the FTC alleged that competition in the market for the R&D of a Rotavirus vaccine would be harmed. Yet neither of the merging firms (nor any other) manufactured a Rotavirus vaccine at the time of the merger. Since the product market is nonexistent, no potential competition could be claimed. Similarly, the FTC based its complaint in Sensormatic entirely upon concerns about R&D concentration.

This third category of cases in which no product yet exists represents the avant-garde application of merger enforcement under Section 7 of the Clayton Act and the only setting where the innovation market approach is arguably necessary - conventional merger analysis does not cover such situations. The combination of R&D programs in a merger may have consequences for downstream product markets, however, because there are as yet no actual products, the assessment of the competitive effects must be based upon a surmise about whether the combination will advance or retard technical progress. This application of the innovation market approach is also the most potentially dangerous application, precisely for the reasons discussed in Sections A and B above. Economists and lawyers do not have the means to predict whether the combination of projects will advance or retard technical progress.

We need look no further than the cases already brought by the FTC to understand the likely risks. However much it may look like "preserving R&D competition," there is no reasonable prediction that can be made on the basis of available theoretical or factual knowledge that requiring American Home Products to license American Cyanamid's vaccine research will improve the odds or move forward the day that a vaccine will be introduced. Indeed, to the extent that granting a compulsory license might weaken American Home Products' resolve to pursue its research, the effect could be the opposite of what is intended. More likely, however, since the newly merged firm has not been asked to hold separate and divest American Cyanamid's intellectual property or to transfer assets and employees to another firm, the license will be of little consequence.

Even in the less complex technology of merchandise security labeling, the proposition that divestiture or compulsory licensing improves the odds and speed of innovation is nothing more than speculation. In Sensormatic, the complaint alleges that the acquisition, inter alia, decreases the number of R&D "tracks" on which R&D is taking place to develop source labeling systems, with the inference being that this would impair innovation.[87] In fact, there is no basis for supposing that keeping the assets associated with one developmental technology (in this case, Knogo's patented SuperStrip technology) outside the merger will improve the pace of technical change. The reasoning behind the consent agreement is to preserve the continued incentives of both Sensormatic and Knogo's independent successor to work on source labeling technology. However, even if source labeling is a separate relevant market, each company has an incentive quickly to produce security labeling products that will supplant the current generation of security tag systems. Whether innovation will be improved by preserving the number of tracks on which R&D is taking place is pure speculation.

The problem, simply put, is that R&D competition is more complicated than price competition, and the incentives, path of progress and outcomes are much harder to predict. A confident-sounding avowal in a complaint or consent order that R&D competition will be enhanced by the preservation of an extra actor in an innovation market simply does not carry over to a useful conclusion about the effect on innovation and welfare in the same way that preserving pricing voices does in conventional product markets. Invoking the "case-by-case" approach[88] is no help in this regard since there is no principled way to evaluate any individual case. The case-by-case approach ought to distinguish any antitrust inquiry; the problem here is a lack of means for judging whether innovation is harmed or served by a merger when there are as yet no products to evaluate. In the absence of theoretical underpinnings, it does not matter how meticulously a "case-by-case" approach is applied because the analyst cannot tell whether the remedy is likely to have the desired outcome. The more removed the R&D projects in question are from issuing commercially viable products or processes, the less predictable and more fragile is their state. Just where the innovation market approach is most apt, it is most dangerous.

Given these objections, how is it that merging firms have been willing to enter into consent decrees which involve relief (divestiture or compulsory licensing) consistent with the innovation market approach? Confronted with the prospect of stalling a merger over a research project with an uncertain future payoff, the temptation to knuckle under on an innovation market issue will be great. One of the chief dangers of the innovation market approach is the ease with which the agencies can impose divestiture or compulsory licensing precisely because a single overlapping R&D project is likely to be a small part of any merger and one that is likely to be discounted for the time and risk associated with future earning streams.

In theory, this would be an easy problem to solve by means of "surgical challenges." When it is clear that the only objection to a merger is an R&D project, then "unscrambling eggs" should not be as difficult as it is in cases concerning product markets. The merger might be allowed to proceed without delay and the relief sought would be limited to licensing or divestiture of the project subject to challenge. This represents an abdication of the holdup power of the agencies, but produces a procedure that is more consistent with the intentions of the innovation market approach. Otherwise, agency victories will reflect nothing so much as the small expected value of an uncertain research project relative to the larger gains from a merger. But the better cure would be to abandon the innovation market approach entirely.

IV. CONCLUSION

The Federal Trade Commission and the Department of Justice should review their commitment to the innovation market approach, cautiously resisting the urge to be fashionably high-tech when the benefits to consumer welfare are at best unclear. A reason for circumspection is that the agencies are the custodians of the integrity of the 1992 Horizontal Merger Guidelines approach. Although the courts have not been generous in their recognition of the Guidelines as the most authoritative, principled approach to market power analysis, that view is, I believe, very widely held among modern industrial economists and practitioners of antitrust law. The integrity of the Guidelines stems in part from its theoretical underpinnings, which were strong at the outset in 1984 and strengthened further in the 1992 revisions. Incorporating the innovation market approach into merger review under the Guidelines diminishes this solid theoretical foundation.

More important than this indirect weakening of the foundation of current merger enforcement is the direct risk of bad decisions that the innovation market approach creates. Although the innovation market approach is an analogy of the 1992 Horizontal Merger Guidelines' product market analysis, the analogy is false. While in product markets, the optimal product price is based on marginal cost and upward departures of price from cost are unequivocally bad for consumers, in innovation markets the optimal amount of R&D is unknown. Therefore, the relationship of actual amounts to the optimal amount is unknown. Moreover, we cannot explain or predict the effect of a change in the structure of a relevant innovation market on the amount of R&D effort. The hope that a divestiture or compulsory licensing will increase rather than decrease investment in R&D is nothing more than that - a hope.

An additional risk is the ease with which the innovation market approach can be abused. Merger enforcement in recent years has become largely a regulatory process which rarely enters the courts. Mergers are time-sensitive transactions and the potential duration and uncertainty of a court battle is often enough to dissuade all but the most ardent firms from completing a deal. The reality these days is that merger candidates must confront the enforcement agencies to negotiate permission to merge. By opening a second front on R&D activities, the agencies have tilted the balance of bargaining strength away from merging firms in research-intensive industries. The innovation market approach viewed in this context is a second weapon that can be used to strengthen draft complaints thereby enhancing the threat of an agency challenge. Alternatively, the approach can be used to force divestiture or require compulsory licensing of active R&D projects - radical remedies whose connection to any reliably predictable consumer benefit is speculative at best.

An R&D project can turn into any one of the following four things:

1. A paradigm-shifting blockbuster innovation that transforms whole technologies and markets;

2. A curve-shifting innovation that saves substantial costs or creates important new products;

3. A minor improvement over current technology; or 4. A waste of money.

The reader may easily guess which of these is the rarest and which the most frequent, but from the vantage point of a box of second request documents, no one can predict how a given research project will end up. The underlying absence of cause and effect connections between concentration, R&D expenditure, innovation and competition make the innovation market approach a theoretically weak adjunct to the well-groomed 1992 Horizontal Merger Guidelines. To try to manage R&D competition using the blunt instruments of Clayton Act enforcement is a mistake.

Endnotes:

[*] President, National Economic Research Associates, Inc. (NERA). The author is grateful to Sumanth Addanki, Victor Goldberg, Michael Salinger and Richard Schmalensee for valuable discussions and suggestions, and to David Backer, Elisabeth Forest and Una Grewal for research assitance. The author's opinions presented here do not necessarily reflect those of other economists at NERA.

[1] 15 U.S.C. Section 8.

[2] 15 U.S.C. Section 45.

[3] U.S. Department of Justice and Federal Trade Commission, Horizontal Merger Guidelines (April 2, 1992), reprinted in 4 Trade Regulation Reporter (CCH) (1992) para. 13,104.

[4] Examples are cited below in discussion of specific aspects of the policy.

[5] 15 U.S.C. 4302 (1984). With the enactment of the 1984 NCRA, Congress sought to encourage certain cooperative research endeavors and dispel uncertainties for market competitors by defining protected joint R&D activities and acceptable conduct, standards of review and limitations on antitrust civil remedies (in particular, actual versus treble damages) if the venture is properly disclosed but later determined to violate the law.

[6] U.S. Department of Justice, Antitrust Enforcement Guidelines for International Operations, (November 10, 1988), reprinted in 4 Trade Regulation Reporter (CCH) (1992) para. 13,109. The 1988 International Guidelines explicated the Department of Justice's antitrust enforcement policies and procedures and was designed to provide firms engaged in international operations with insight into the potential evaluation of competitive effects of business conduct, particularly with respect to joint ventures and intellectual property licensing agreements, among other things.

[7] Richard J. Gilbert and Steven C. Sunshine, "Incorporating Dynamic Efficiency Concerns in Merger Analysis: The Use of Innovation Markets," Antitrust Law Journal; 63(2), Winter 1995, 569-601. Although I have used this as a source document for the innovation market approach and how I believe it is presently employed by the DOJ and the FTC, the authors state that the views expressed in their article do not represent the policies of the DOJ for which both worked at the time of writing. Following Gilbert and Sunshine I will use the phrase "innovation market approach" to mean the procedure described in Gilbert and Sunshine at 594-597.

[8] 15 U.S.C. 4302 (1984).

[9] Antitrust Enforcement Guidelines for International Operations, supra note 4 at 20,624.

[10] Id. at 20,625.

[11] Roche Holdings Ltd., 113 F.T.C. 1086 (1990).

[12] United States v. General Motors Corp., et al., Civ. No. 93-530 (D. Del filed November 16, 1993).

[13] United States v. Flow International Corp. and Ingersoll-Rand Co., Civ. No. 94-71320 (E.D. Mich. filed April 14, 1994).

[14] U. S. Department of Justice and Federal Trade Commission, Antitrust Guidelines for the Licensing of Intellectual Property (April 6, 1995), reprinted in 4 Trade Regulation Reporter (CCH) (1995) para. 13,132.

[15] American Home Products Corporation, C-3557 (February 14, 1995). The FTC's complaint analyzed the effects of the merger in terms of several relevant product markets for certain vaccines, in addition to the market for R&D of a Rotavirus vaccine. AHP and Cyanamid were two of three competitors currently undertaking R&D for a Rotavirus vaccine and Cyanamid's effort was believed to have the potential for yielding a superior vaccine. In the consent agreement, AHP agreed within one year to license on a non-exclusive basis Cyanamid's research in Rotavirus vaccines and to provide technical support to a FTC-approved third party.

[16] Glaxo plc., File No. 951 0054 (March 16, 1995). The FTC opposed Glaxo's acquisition of Wellcome on the basis of "[t]he relevant line of commerce...[into]...the research and development of non-injectable 5HT1D agonists" in the U.S., a specific class of drugs for migraine headaches. The complaint further addresses the merger's effect not only on actual competition in the relevant market but also on the number of R&D "tracks" and the likelihood of a reduction in R&D "both in the near future and in the long term" (Id. at 3.) In its consent agreement with the FTC, Glaxo agreed to divest Wellcome's worldwide R&D assets for the non-injectable 5HT1D agonists to a FTC-approved third party and to provide technical assistance to ease the purchaser into the manufacture and sale of migraine drugs.

[17] Sensormatic Electronic Corp., File No. 941 0126 (January 4, 1995). The FTC opposed Sensormatic's acquisition of Knogo's non-North American assets and of certain patents and royalty-free cross-licensing and grantback agreements to become effective subsequent to the acquisition between Sensormatic and a new, independent successor to Knogo in North America. The FTC's opposition was based on a line of commerce defined as "...the research and development of disposable labels developed or used for an electronic article surveillance system. The type of labels in question are part of a new "source" labeling system in which the manufacturer, rather than the retailer, applies the label to the merchandise. No such system is currently in use. Both Sensormatic and Knogo currently produce competing, incompatible labeling systems and have source labeling products in development. The FTC alleged that the merger would allow firms in this relevant market to "...restrict output of research and development..." both in the near and the long term. Under the terms of the consent agreement, Knogo's successor would retain its patented products and technology. Sensormatic is prohibited from entering into anything but nonexclusive licenses for the use of these patented products and technologies in the U.S. and Canada or from any grantback arrangements involving the products or technology.

[18] Wright Medical Technology, Inc., C-3564 (March 23, 1995). At the time of the proposed merger between Wright Medical Technology and Orthomet, the companies did not compete in any product market - in fact, Wright allegedly controlled 95 percent of the market for human hand implants, while Orthomet did not have an FDA-approved product on the market. Nonetheless, the complaint contended that they were competitors in the market for R&D of next-generation finger implants.

[19] Boston Scientific Corp., File No. 951 0002 (February 24, 1995). In this case, the FTC reviewed Boston Scientific's proposed acquisition of Cardiovascular Imaging Systems (CVIS) and SCIMED Life Systems. Boston Scientific and CVIS were current competitors in the intravascular ultrasound (IVUS) imaging catheter product market with an aggregate post-acquisition market share of 90 percent. A single additional competitor accounted for the remainder of the market. SCIMED, although not a current competitor in the product market, had conducted significant R&D, leading to a prototype product that the FTC alleged would allow SCIMED to enter the market within two to three years with a next generation product. Furthermore, the FTC alleged that the merger of the extensive patent portfolios of the three companies would significantly deter any further entry. Although the consent agreement allowed for the merger of the three firms, Boston Scientific agreed to grant to any interested entrant a royalty-free, nonexclusive license for any of the three companies' patent portfolios in IVUS imaging catheter products and for the trade secrets, non-patented technology and know-how of either CVIS or SCIMED.

[20] Shell-Montedison, File No. 941 0043 (January 11, 1995). The FTC alleged that a joint venture between Royal Dutch Petroleum and Montedison would result in a combination of the two companies' worldwide production, R&D and licensing of polypropylene technology, even though the venture excluded the assets of Royal Dutch's U.S. subsidiary, Shell, and its technology licensing business. The FTC contended that under the joint venture Royal Dutch would control the world's two main competitors in polypropylene innovation and licensing which would jeopardize innovation and licensing competition between Shell and Montedison in the U.S. Although the companies proposed excluding their own polypropylene innovation and technology licensing businesses, the FTC found this remedy insufficient since it contended effective competition could only originate with a firm that also had production capabilities. The consent order, therefore, required Royal Dutch to divest an integrated polypropylene company that would continue to undertake R&D and technology licensing in polypropylene. If fact, the innovation market claim was an inconsequential addition to the complaint since the merging firms controlled the technology market even if they had never innovated again.

[21] Antitrust Guidelines for the Licensing of Intellectual Property, supra note 14 at 20,738.

[22 ] The distinction between innovation markets' and technology markets' is spelled out clearly in the 1995 IP Guidelines. A technology market' is a market for licensed intellectual property, including all close substitutes for it. For example a relevant technology market in electronic calculator technology might include all the patents a calculator manufacturer would need to practice in order to make and sell the devices without infringing, as well as design-around technologies. Technology transfers such as licensing are the transactions that happen in technology markets. By contrast, the corresponding innovation market would consist of the R&D expenditures or capacities of firms devoting themselves to the development of calculator technology. Although innovation markets may have transactions, such as acquisitions of R&D facilities, groups or institutions, the focus of the analysis is simply the aggregate level of R&D activity in the market.

[23] See, e.g., Antitrust Guidelines for the Licensing of Intellectual Property, supra note 14 at 20,738 and Gilbert and Sunshine, supra note 7 at 594-597.

[24] Under the innovation market approach, the geographic dimension is assumed to be worldwide in the absence of trade or regulatory barriers to the contrary.

[25] The two-year time horizon for competitive entry discussed in the 1992 Horizontal Merger Guidelines would also apply to the innovation market.

[26] Gilbert and Sunshine state that in other cases, entry will be easy and the innovation market competitive.

[27] I rely upon but depart from Gilbert and Sunshine's discussion in outlining each of their underlying arguments.

[28] That technical change is the primary engine of U.S. economic growth was Robert Solow's finding in "Technical Change and the Aggregate Production Function," The Review of Economics and Statistics; 39(3), August 1957, 312-320. The sources of output growth and competitive advantage are not necessarily the same. On the importance, for example, of resource endowment, see Richard Nelson and Gavin Wright, "The Rise and Fall of American Leadership: The Post-War Era in Historical Perspective," Journal of Economic Literature; 30(4), December 1992, 1931-1964.

[29] See Zorina B. Khan and Kenneth L. Sokoloff, "Schemes of Practical Utility: Entrepreneurship and Innovation Among "Great Inventors" in the United States, 1790-1865," Journal of Economic History; 53(2), June 1993, 289-307. Khan and Sokoloff examine the careers of 160 inventors in order to track the growth of inventive activity during early American industrialization. Their comprehensive analysis of biographical information and patent histories reveals that inventive activity was aimed at satisfying emerging commercial needs and was disproportionately concentrated in areas conducive to diffusion and its commercialization. Based on their findings, the authors conclude that these inventors were entrepreneurs who responded systematically to market demand. For further discussion on market structure and innovation, see F.M. Scherer, "Schumpeter and Plausible Capitalism," Journal of Economic Literature; 30(3), September 1992, 1416-1433. Scherer addresses the "Schumpeterian question": Which market structure is most favorable to technological change and hence to economic growth? After reviewing the economic literature devoted to this question, he concludes that (1) most of the research supports the conclusion that Schumpeter overstated the importance of monopolistic firms as leaders in the technological race; and (2) the empirical and theoretical analysis has failed to establish any definitive relationship between market structure, innovation and economic welfare.

[30] For example, the latest in a long line of research on the welfare-improving effects of new product (and product characteristic) introductions, as reflected in consumer price indices is Jerry Hausman, "Valuation of New Goods Under Perfect and Imperfect Competition,"

National Bureau of Economic Research,

Working Paper No. 4970 , December 1994. Hausman (and others) demonstrated that the CPI is not an accurate indicator of consumer prices since it fails to incorporate the effects of the introduction of new or improved goods. New or improved goods have positive welfare consequences which are similar to those from price reductions.

[31] Gilbert and Sunshine, supra note 7, at 590-593.

[32] Id. at 570.

[33] Gilbert and Sunshine cite to the GM-ZF complaint as containing an example of competition in innovation affecting markets where the merging firms are not likely potential competitors. They offer the FTC's complaint in Roche-Genentech as an example of competition in innovation reducing the likelihood of entry into markets where the merging firms do not compete prior to the merger but are potential competitors.

[34] See, e.g., Scherer, supra note 29 and the sources discussed in note 38 below.

[35] This point is emphasized in Scherer, supra note 29 at 1423.

[36] "The Agencies may base the market shares of participants in an innovation market on their shares of identifiable assets or characteristics upon which innovation depends, on shares of research and development expenditures, or on shares of a related product" (Antitrust Guidelines for the Licensing of Intellectual Property, supra note 14 at 20,738).

[37] Gilbert and Sunshine, supra note 7 at 573-574 (italics added).

[38] Id. at 579.

[39] See Scherer, supra note 29 at 1416-1433. Among the notable empirical studies are Wesley M. Cohen, Richard C. Levin and David C. Mowery, "Firm Size and R&D Intensity: A Re-Examination," The Journal of Industrial Economics; 35(4), June 1987, 543-565; Zoltan J. Acs and David B. Audretsch, "Innovation in Large and Small Firms: An Empirical Analysis," American Economic Review; 78(4), September 1988, 678-690; and P.A. Geroski, "Innovation, Technological Opportunity, and Market Structure," Oxford Economic Papers; 42(3), July 1990, 586-602.

Cohen et al. investigate Schumpeter's hypothesis that large firms generate a disproportionately large share of society's technological growth. After a thorough analysis of the FTC Line of Business data, they conclude that, among the business units that conduct R&D, there is no significant relationship between the size and R&D intensity, although the size of the business unit is positively correlated with the probability of engaging in R&D. Further, once outliers are removed from the data, a simple regression of R&D intensity on firm size reveals a statistically insignificant association. Firm size and business unit size together explain less than one percent of the variation in R&D intensity. Cohen et al. also use survey data collected by Levin, Klevorick, Nelson and Winter [1984] to explore industry effects on R&D intensity. Their inquiry suggests that industry effects have a significant impact on R&D and explain nearly half the variation in R&D intensity.

Acs and Audretsch present a new and direct measure of innovative activity and use this measure to examine the degree to which innovative activity is affected by industry characteristics. Their empirical analysis reveals that innovative activity increases, at a decreasing rate, with increases in industry R&D expenditure. In other words, industry innovation tends to decrease as the level of concentration increases. Their analysis also provides evidence for the fact that patterns of innovative activity in large and small firms differ considerably.

Geroski uses data on major innovations introduced in the United Kingdom during 1970-1979 to conduct an empirical analysis of Schumpeter's hypothesis that actual monopoly power stimulates innovative activity. The author concludes that there is absolutely no support in the data for Schumpeter's assertions that monopolists have greater incentives to carry out innovations. Instead, Geroski's results suggest that competition promotes innovation, despite the possibility of lower post-innovation returns due to the presence of firm rivalry.

For a critique of the research on Schumpeter, see Franklin M. Fisher and Peter Temin, "Returns to Scale in Research and Development: What Does the Schumpeterian Hypothesis Imply?" Journal of Political Economy; 81(1), January/February 1973, 56-70. Fisher and Temin find fault with the narrow focus of the empirical tests designed to evaluate Schumpeter's hypothesis that the returns to R&D increase as smaller firms merge to form a "large-scale establishment or unit of control." The authors conclude that the empirical tests are inappropriate because they do not accurately reflect the "richness of Schumpeter's hypothesis."

[40] For a summary, see Richard Schmalensee, "Inter-Industry Studies of Structure and Performance," Chapter 16 in Richard Schmalensee and Robert D. Willig, eds., Handbook of Industrial Organization, Volume 2 (Amsterdam: North-Holland, 1989).

[41] It is an overstatement to say that the 1992 Horizontal Merger Guidelines are simply an application of George Stigler's theory of oligopoly. This assertion ignores refinements concerning single-firm market power and the influence of conditions of entry. Nevertheless, the theoretical nexus that ties the HHIs in relevant markets defined by substitution criteria to the preservation of price competition is Stigler's well-accepted idea that more pricing voices make it harder to enforce cartel discipline.

[42] Cartel prices (and chiselers' prices) must be disclosed to buyers in transactions. R&D need not involve any disclosure to outsiders which makes policing agreements hard. When chiseling on a pricing agreement is discovered, cartel prices will drop to punish the chiseler or the cartel may dissolve. In either case the gains to the chiseler are fleeting. Chiseling on an R&D-restricting agreement, by contrast, can yield first-mover advantage, a preemptive patent or at least lead time. This is not to say that such agreements can never arise, just that they necessarily must be very rare. For example, see Dennis Yao and Susan De Santi, "Innovation Issues Under the 1992 Merger Guidelines," Antitrust Law Journal; 61(2), Winter 1993, 505-521, esp. 513-517. Gilbert and Sunshine acknowledge that "[c]ollusion in R&D is difficult..." (Gilbert and Sunshine, supra note 7 at 597).

[43] Section of Antitrust Law, American Bar Association, Antitrust Law Developments, Volume I (3d ed. 1992), 388 (italics added).

[44] Janusz A. Ordover and Robert D. Willig, "Antitrust for High-Technology Industries: Assessing Research Joint Ventures and Mergers, Journal of Law and Economics; 28(2), May 1985, 312.

[45] Id. at 325.

[46] Id. at 327.

[47] Id. at 332-333.

[48] See, e.g., Thomas M. Jorde and David J. Teece, "Innovation and Cooperation: Implications for Competition and Antitrust," Journal of Economic Perspectives; 4(3), Summer 1990, 75-96; and Thomas M. Jorde and David J. Teece, "Innovation, Cooperation and Antitrust," High Technology Law Journal; 4(1), Spring 1989, 1-112.

[49] The U.S. is said to compare unfavorably in this regard with Japan, which encourages cooperation through its Ministry of International Trade and Industry in consultation with the Japan Fair Trade Commission, and the European Union, which grants a block exemption for firms with low market shares.

[50] Gilbert and Sunshine, supra note 7 at 594. The small but significant nontransitory reduction in R&D is the basis for defining a market. A market, by definition, is susceptible to cartelization (in any aggregation smaller than a market, close-substitute outsiders would compete to the disadvantage of the insiders who tried to cut back.) The avoidance of these cutbacks by prevention of R&D concentration in relevant markets defined by them is the essence of the innovation market approach.

[51] Even if cooperation in the Jorde and Teece vocabulary simply means information sharing, and the only effect of cooperation is saved resources and quicker progress, the point stands. To an unomniscient observer, these look like expenditure, input or capacity cutbacks (relative to the noncooperative alternative) that are indistinguishable from a small but significant nontransitory reduction in R&D.

[52] For a summary of the relevant literature, see Jennifer F. Reinganum, "The Timing of Innovation: Research, Development and Diffusion," Chapter 14 in Richard Schmalensee and Robert D. Willig, eds., Handbook of Industrial Organization, Volume 1 (Amsterdam: North-Holland, 1989); and F. M. Scherer and David Ross, Industrial Market Structure and Economic Performance (Boston: Houghton Mifflin Company, 1990), 630-660.

[53] That the social gains from R&D exceed the private gains does not argue for a policy of discouraging reductions in R&D either. It is merely the obverse of the fact that only a fraction of the gains from successful innovation is captured by the innovator, even in taut intellectual property regimes. See the discussion in Jonathan Baker, "Fringe Firms and Incentives to Innovate," Antitrust Law Journal; 63(2), Winter 1995, 621-641, esp. 621-622, notes 5 and 6.

[54] Michael C. Jensen, "Agency Costs of Free Cash Flow, Corporate Finance and Takeovers," American Economic Review; 76(2), May 1986, 323.

[55] Michael C. Jensen, "The Modern Industrial Revolution, Exit, and the Failure of Internal Control Systems," The Journal of Finance; 48(3), July 1993, 858.

[56] Id. at 860.

[57] Sudipto Bhattacharya and Dilip Mookerjee, "Portfolio Choice in Research and Development," Rand Journal of Economics; 17(4) Winter 1986, 594, citing others. Counter tendencies may arise from the payoff structure on innovation. Long lead times with no payoff until the patent race is won leads to underinvestment in R&D. See Gene M. Grossman and Carl Shapiro, "Optimal Dynamic R&D Programs," Rand Journal of Economics; 17(4), Winter 1986, 581-593.

[58] See Partha Dasgupta and Joseph E. Stiglitz, "Industrial Structure and the Nature of Innovative Activity," Economic Journal; 90(358), June 1980, 266-293.

[59] See Richard C. Levin, et al., "Appropriating the Returns from Industrial Research and Development," Brookings Papers on Economic Activity; 3(0), 1987, 783-820.

[60] Gilbert and Sunshine, supra note 7 at 577.

[61] See Kenneth A. Froot, David S. Scharfstein and Jeremy C. Stein, "Risk Management: Coordinating Corporate Investment and Financing Polices," The Journal of Finance; 48(5), December 1993, 1629-1658.

[62] Downward pressure on R&D budgets due to competition is a general trend in the 1990s shared by many research-intensive industries. See Business Week, "Blue-Sky Research Comes Down to Earth," July 3, 1995, 78-81.

[63] For example, 98 companies are currently testing 215 different drugs for 20 different types of cancer, double the number of firms from just two years ago. For breast cancer alone, Biomira, Bristol-Myers Squibb, Ciba, Du Pont Merck, Genentech, Janssen, Liposome and Pfizer all have drugs in Phase III trials. See Pharmaceutical Research and Manufacturers of America, New Medicines in Development for Cancer, 1995 Survey, May 1995.

[64] Antitrust Guidelines for the Licensing of Intellectual Property, supra note 14 at 20,738

[65] Gilbert and Sunshine, supra note 7 at 596; see also Id. at 588.

[66] United States v. General Motors Corp., et al., supra note 12 at 12-13.

[67] Gilbert and Sunshine, supra note 7 581-586.

[68] While not theoretically out of bounds, pairings of concentrated R&D capacity with unconcentrated product markets must be rare. Typically, individual firms' shares of aggregate R&D expenditure and sales revenues in a given industry are fairly consistent. Across industries, the same relationship holds: Scherer and Ross found that the shares of aggregate R&D outlays and sales revenue for a sample of 915 corporations were roughly proportional for each of a series of groups chosen based on sales rankings (Scherer and Ross, supra note 51 at 654-655). These findings imply a lack of cases where a few firms are able to lock up the innovative capacity associated with an unconcentrated product market. Therefore, in most instances, the conventional analysis of product and technology markets will suffice.

[69] Gilbert and Sunshine, supra note 7 at 570.

[70] Id.

[71] U.S. Department of Justice, Merger Guidelines, (June 14, 1984), reprinted 4 Trade Regulation Reporter (CCH) (1984) para. 13,103.

[72] For a detailed discussion of the two distinct doctrines of potential competition involving actual and perceived competition, and differing interpretations of each by the courts and FTC in Section 7 actions, see Phillip E. Areeda and Herbert Hovenkamp, Antitrust Law: An Analysis of Antitrust Principles and Their Application, 1994 Supplement (Boston: Little, Brown and Company, 1994), 749-753.

[73] Marine Bancorporation, 418 U.S. 602, 630-631 (1974).

[74] See, e.g., United States v. Black & Decker Manufacturing, Co., 430 F. Supp. 729, 771-772 (D. Md. 1976); and Champion Spark Plug Co., 103 F.T.C. 546, 631 (1984).

[75] Horizontal Merger Guidelines, supra note 3 at Sections 1.32 and 3. "Supply substitution" is a synonym for uncommitted entry. Under the Merger Guidelines approach, a firm doing R&D on an extant product but not actually producing it would be counted in the market if entry were less than a year off and sunk costs were not an issue. If entry took longer or were more costly but was still timely and likely, the firm would be treated as a committed entrant.

[76] United States v. General Motors Corp., et al., supra note 12 at 1.

[77] Id. at 8.

[78] Id. at 10.

[79] Id. at 13.

[80] Gilbert and Sunshine contend that other automatic transmission markets supplied by one or another of the firms would have been adversely affected by the merger (Gilbert and Sunshine, supra note 7 at 587). If only unconcentrated product markets were adversely affected by the merger's R&D-concentrating impact, then the innovation market approach could be said to have succeeded where the conventional approach failed. This, however, is not the case.

[81] United States v. Flow International Corp. and Ingersoll-Rand, supra note 13 at 7.

[82] Boston Scientific Corp., supra note 19 at 2.

[83] Id. at 3.

[84] Wright Medical Technology, Inc., supra note 18 at 3.

[85] Glaxo plc., (Agreement Containing Consent Order), supra note 16 at 4.

[86] Prepared remarks of Susan S. DeSanti, "Innovation and Merger Enforcement: The View from the FTC," before Symposium on Innovation, American Bar Association Spring Meeting, Washington, D.C., April 6, 1995, 9.

[87] Sensormatic Electronic Corp., supra note 17 at 3. No product markets are alleged, only a narrow R&D market for "source labeling." Commissioner Azcuenaga dissented from that narrow market definition.

[88] See, e.g., Baker, supra note 52 at 621-641; and remarks by Christine A. Varney, "Why Innovation Market Analysis Makes Sense," before the Antitrust 1995 Conference, Washington, D.C., March 15, 1995, 2-3.