The 36th Annual Antitrust Institute
San Francisco, California
Commissioner, Federal Trade Commission. The views expressed are those of the Commissioner and do not necessarily reflect the views of the Federal Trade Commission or any other Commissioner or staff.
Evaluating mergers in dynamic and innovative markets presents challenges to traditional antitrust theory. One particular situation where the Commission confronts the changing nature of dynamic transactions involves vertical mergers. I want to discuss the theory of anticompetitive effects from vertical mergers and then give you a real world example of an enforcement action by the FTC in the computer industry.
Vertical integrations are usually mergers of noncompeting companies where one's product is a necessary component or complement of the other's. Such mergers can achieve procompetitive efficiency benefits. Vertical integration can lower transaction costs, lead to synergistic improvements in design, production and distribution of the final output product and thus enhance competition. Consequently, most vertical arrangements raise few competitive concerns.
However, as reflected in the 1984 Merger Guidelines, some vertical acquisitions can be anticompetitive. Vertical mergers can create or raise entry barriers that lead to higher prices or lower quality or innovation for consumers. For example, in industries with extensive networks, many firms already have market power through their ownership of established networks or installed bases involving huge sunk costs. Vertical mergers can, in certain instances, increase those barriers to entry even more, raising costs and reducing innovation and quality for consumers.
How can a vertical merger increase barriers to entry? The first general category of anticompetitive theories posits that, in certain instances, vertical integration can foreclose rivals from access to needed inputs or raise their costs of obtaining them. For example, in a recent article, Professors Riordan and Salop have developed further anticompetitive theories of "raising rivals' costs," where a vertically integrated company may be able to increase the costs of its rivals in either the upstream or downstream market. Such foreclosure effect can raise prices or reduce quality or innovation to consumers downstream. Ultimately, such a foreclosure effect may require that firms seeking to enter one of the markets must enter both markets, significantly increasing the difficulty of entry.
Second, a vertical merger can facilitate collusion in either the upstream or downstream market. Acquisition of a supplier by a purchaser may create opportunities to monitor the upstream supplier's competition. Also, a vertical merger may involve the purchase of a particularly disruptive downstream buyer. By eliminating a buyer who played one upstream firm off of another, such a merger may facilitate collusion in the upstream market.
Most of my discussion this morning will bear on these two main theories of anticompetitive harm -- foreclosure and facilitating collusion -- from vertical mergers. I should note, however, that there is a third theory of anticompetitive harm arising from vertical mergers -- vertical mergers that are designed to evade pricing regulations. For example, when regulation seeks to constrain the market power of a natural monopoly, the monopolist may have incentives to integrate vertically into unregulated markets in order to extract the monopoly rents denied it in the regulated market. This theory was the basis for much of the Modified Final Judgment in the AT&T monopoly case and was most recently utilized by DOJ in itsBritish Telecom/MCI transaction challenge.
One criticism of the first theory of anticompetitive harm -- the foreclosure or "raising rivals' costs" theory -- is that harm to competitors does not always result in harm to competition. Remember that the cornerstone of antitrust law is "the protection of competition, not competitors." Thus, even in a raising rival's cost theory, some showing of likely consumer injury should be required before a vertical merger is challenged -- that is, a likely increase in quality-adjusted price or likely decrease in output.
Although it may not seem intuitively obvious when a vertical merger indeed will lead to such an anticompetitive outcome, rather than simply harm rivals, there are several indicia to help antitrust enforcers in making such a determination. Certainly the greater the market share of the companies that are vertically integrating, the greater the probability that downstream customers will be injured. Furthermore, in conducting investigations, staff contacts a broad array of rivals -- as well as downstream customers -- in order to obtain a general sense of the consequences of a vertical transaction. Concerns of vertical foreclosure and raising rivals' costs obviously become more credible the more often such views are convincingly repeated by industry participants.
Now, there is a great deal of theoretical controversy about the effects of vertical mergers. Theory, however, must be informed by facts in antitrust enforcement, and I believe that the best way to approach vertical merger enforcement is through case-by-case analysis. When there is a factual basis supporting a likely anticompetitive effect from a vertical acquisition, we should act. Some vertical mergers in the past may have had anticompetitive consequences -- I raise as but one possible example integration by airlines into computer reservations systems in the 1970's. While judging the probability of such an outcome is not easy, we, as antitrust enforcers, have the tools available to ensure that only prudent enforcement actions are taken in this area. We simply cannot abdicate our responsibility to foster competition and protect consumers.
As a part of the FTC's case-by-case analysis, antitrust enforcers must take great care when considering the nature and extent of the remedy in vertical merger cases. Since many vertical mergers result in procompetitive efficiencies, we must craft relief narrowly to permit procompetitive efficiencies to come to fruition whenever possible.
Now that I have explained the theory, let me move on to give you a sense of a recent FTC vertical merger enforcement action in the computer industry. Vertical merger enforcement in the computer industry affords us the opportunity to ensure that competition in industries characterized by networks and extensive installed bases remains as robust as possible at all levels -- hardware as well as software -- while permitting parties to achieve the synergistic and innovation-unleashing benefits of vertical integration.
This recent case concerned Silicon Graphics, Inc.'s plan to acquire two of the world's three leading entertainment graphics software firms -- Alias Research Inc. and Wavefront Technologies, Inc. The software at issue is used in producing sophisticated three-dimensional graphics and animation images for the entertainment industry (movies, videos and television). Such software has been used to produce special effects like the dinosaurs in Jurassic Park, Forrest Gump's participation in history-making events in Forrest Gump, and the liquid metallic robot in Terminator 2. The software is also used in the creation of electronic games, interactive programming and other video and graphic media. The complaint alleged that Silicon Graphics has a 90 percent share of the market for the workstations that run such software. These workstations are powerful computers costing $10,000 to $200,000 with extensive memory capable of performing high resolution graphics and complex calculations. Although other companies manufacture workstations with graphic capabilities for other applications like industrial or scientific uses (for everyone from automotive manufacturers to universities), the entertainment graphics software at issue in this case is developed almost exclusively for use on Silicon Graphics workstations. The two software companies to be acquired -- Alias and Wavefront -- along with a third company, SoftImage, Inc. (recently acquired by Microsoft), are industry standards, according to the complaint in this matter. Thus, the Commission alleged that the ability to run the software of these companies is critical for any computer workstation manufacturer to compete successfully in the entertainment graphics workstation market.
The Commission's complaint notes that, prior to the announcement of the acquisitions at issue, Alias negotiated with other workstation manufacturers to enable its entertainment graphics software to run on their computer operating systems. Silicon Graphics also maintained an open software interface for its entertainment graphics workstations, which meant that independent software developers could access specification information about Silicon Graphics workstations to ensure that their products would run on them. It also sponsored independent software developer programs, and shared advance information about its new product with software developers so as to promote competitive development of new entertainment graphics software.
The FTC complaint alleged that Silicon Graphics' acquisition of Alias and Wavefront could have a number of anticompetitive effects. The first concern was that competition in the entertainment graphics workstation market could be harmed: namely, that, as a result of this vertical integration, the combined entity would end discussions over making its acquired software compatible with another workstation. Since Alias' and Wavefront's entertainment graphics software are considered the industry standard, another workstation producer would be left at a disadvantage or entirely foreclosed from competing in this market. The second concern was that competition in the entertainment graphics software market could be harmed: namely, that, as a result of this merger, the combined entity might end Silicon Graphics' open architecture policy, raising the costs of independent developers of entertainment graphics software. The complaint alleged that, as a result of these two foreclosure effects, the merger would raise barriers to entry by new competitors by making it necessary for them to enter at both the workstation and software level as well as make it easier for Silicon Graphics to use its market power in workstations to engage in price discrimination and raise prices to entertainment graphics customers.
Let's examine the theory behind the complaint. One might ask why the combined entity would have an incentive to disadvantage rival software developers -- would not doing so only reduce demand for Silicon Graphics' workstations? It is important to note that the combined entity would not need to bar other software developers completely, but could redirect them away from direct competition, by, for example, encouraging development of products that are complements to, rather than direct substitutes for, Alias and Wavefront software. Consequently, competition in the software market could be harmed without necessarily reducing demand for workstations.
Now, we can flip that question and ask whether, since Silicon Graphics already has market power in the workstation market -- largely as a result of its success and its resulting extensive installed base, what added competitive harm can acquiring software developers cause? Even when a hardware company arguably has market power, independent software companies have strong incentives to prevent the hardware manufacturer from exploiting fully its market power. Independent software companies should be concerned that high hardware prices will drive down overall demand for software and, thus, will seek to work with new hardware entrants whenever possible. Even apart from a concern that prices or quality will be affected, independent software developers, enmeshed in contractual relationships with hardware suppliers, may be wary of the bargaining consequences of dealing with a hardware supplier with market power. Thus, a hardware company with market power may find it difficult to extract fully any supra-competitive rents from ultimate consumers because of the efforts of these independent software developers. Consequently, when a company has market power, vertical integration may enhance that power or make it more effective.
Furthermore, independent software companies that resell hardware may make it difficult for a hardware company to price discriminate successfully against users with inelastic demand -- users who will not or cannot switch -- or against consumers in end-use markets where it has market power. An inelastic consumer can purchase the hardware separately from the software re-seller, thus making it difficult for the hardware company to price discriminate against that consumer. Indeed, the Commission's complaint here alleged that, as a result of gaining control of Alias and Wavefront, SGI enhanced its ability to price discriminate against entertainment graphics end-users.
There's a final question you might ask about this case. In the midst of this vertical acquisition, there is a straightforward horizontal merger -- the combination of Wavefront and Alias. Indeed, the complaint in this case alleges that the horizontal combination of Alias and Wavefront was anticompetitive. Why did the Commission not block this horizontal aspect of the acquisition by seeking a preliminary injunction? The determination whether to seek an injunction in federal court to block a horizontal merger is one fraught with many considerations, including litigation risks. The parties presented evidence that the combination of Wavefront and Alias would afford them significant efficiencies in developing entertainment graphics software and other types of graphics software. Consequently, rather than block the horizontal combination outright, the Commission accepted a settlement that it believed resolved not only the vertical concerns, but at the same time allayed any concerns that the merger would decrease competition in the software market.
Let's turn to the specifics of the settlement with Silicon Graphics. First, because Alias and Wavefront are industry standards with an installed base, other workstation producers would be less likely to enter, unless Alias or Wavefront software was made compatible with their workstations. The proposed consent agreement seeks to solve this problem by requiring that Alias's two major entertainment graphics software programs are made compatible with another workstation producer's product. This provision should increase competition in the entertainment graphics workstation market.
Second, because Silicon Graphics' workstation operating system is an industry standard with an installed base, the agreement would require Silicon Graphics to maintain an open architecture and publish its application programming interfaces so that software developers other than Alias and Wavefront could develop entertainment graphics software for use on Silicon Graphics' workstations. In addition, the agreement would require Silicon Graphics to offer independent entertainment graphics software companies participation in its software development programs on terms no less favorable than it offers other types of software companies.
It is hoped that these provisions will prevent the combined entity from discriminating against independent software producers. Trying to prevent self-favoritism is, of course, not an easy thing to do. There are difficulties in constructing restrictions that are easily monitored. Furthermore, there is concern that such restrictions not impede any likely efficiencies from collaboration between hardware and software development personnel within a vertically integrated firm. In this case, we were fortunate to have a benchmark to judge how these companies would treat rivals if the market were more competitive. There are other applications for which SGI workstations are used such as industrial and scientific applications. These applications are more competitive in both the hardware and software markets. Consequently, the order was drafted to require that Silicon Graphics treat independent entertainment graphics software developers in the same way it treats independent software developers in those other more competitive markets. The order thus has the twin virtues of being easy to monitor and restoring competition as best as possible short of requiring a divestiture.
A third aspect of the settlement -- a firewall provision to prevent the transfer of competitively sensitive information -- deserves mention because it relates to a competitive concern in vertical acquisitions that has been encountered before. Let me provide some background first. Because, by definition, vertical acquisitions involve companies making products one of which is a necessary complement to the other's, vertical acquisitions can give the combined entity the ability to obtain competitively sensitive information about competitors in either market. The information could involve nonpublic pricing information difficult to obtain elsewhere, in which case the competitive concern would be that collusion in one of the markets could be facilitated as a result of the merger. For example, in Lilly's acquisition of PCS, a pharmacy benefit management company, the Commission's proposed settlement requires that the merged entity construct a firewall to prevent Lilly from obtaining other drug manufacturers' bids, proposals, contracts, prices, rebates, discounts, or other terms and conditions of sale.
Pricing information, however, is not the only competitively sensitive information that can be passed on. Particularly in high tech industries where upstream and downstream companies must work closely, providing sensitive information about design specifications to the other, a vertically integrated entity's division in one market could pass on sensitive information about the design or innovation of a competitor in the other market to its division in that other market. The competitive concern is that, as a result, the merged entity could "free ride" off its competitors' hard work, potentially inhibiting the competitors' incentives to innovate. Commentators agree that such free riding may chill innovation by reducing companies' incentives to innovate and, ultimately, hurt consumers.
In this matter, the settlement with Silicon Graphics prohibits the release of non-public information from the workstation producer, who is making the Alias software compatible with its workstation, to those Silicon Graphics or Alias employees not participating in this compatibility process. The particular concern here is that the merged entity could find out sensitive, nonpublic information about its hardware competitor via this process. Consequently, the agreement builds a "firewall" preventing information from passing beyond those employees actually engaged in process.
The Commission in several recent vertical cases has required firewalls to prevent the passage of competitively-sensitive information. Of course, it remains to be seen whether such firewalls in fact prevent the dissemination of competitively- sensitive information within a firm. Anecdotal evidence, however, suggests that firewalls in the defense industry have been successful.
Of course, independent workstation and software producers may ultimately fail because of Silicon Graphics' and Alias/Wavefront's success and their extensive installed base. Thus, it could be argued, this settlement does not ensure increased competition. The antitrust laws, however, do not seek to guarantee the success of particular commercial endeavors; rather, they strive to break down anticompetitive barriers or bottlenecks to entry. To the extent that these companies obtained their market power through legitimate means -- they made better workstations and software and, consequently, were able to develop an installed base of entertainment graphics customers -- it would be inappropriate for antitrust enforcers to seek to deprive them of the fruits of their hard work. Consequently, by lowering barriers to entry and thus providing dissatisfied customers with the option to switch to a new hardware or software entrant, this settlement seeks to prevent only the exercise of market power stemming from the unlawful and anticompetitive acquisitions here.
The Commission's action with respect to Silicon Graphics demonstrates the care with which the Commission proceeds when considering vertical mergers. It also shows that antitrust can play a vital role in preventing bottlenecks in industries characterized by networks and extensive installed bases. Finally, it demonstrates that the best way to develop a sensible vertical merger enforcement policy is to rely on the factual evidence presented and to act on a case-by-case basis when the facts support a plausible theory of anticompetitive harm.