DISSENTING STATEMENT OF COMMISSIONER ROSCOE B. STAREK, III
In the Matter of
Cadence Design Systems, Inc. and Cooper & Chyan Technology, Inc.
File No. 971 0033
I respectfully dissent from the Commission's decision to accept a consent agreement with Cadence Design Systems, Inc. ("Cadence"), a supplier of software for the design of integrated circuits ("ICs"). The proposed complaint alleges that the merger of Cadence and Cooper & Chyan Technology, Inc. ("CCT") -- a producer of software complementary to Cadence's -- is likely substantially to lessen competition in violation of Section 7 of the Clayton Act, 15 U.S.C. § 18, and Section 5 of the Federal Trade Commission Act, 15 U.S.C. § 45. To justify the proposed complaint and order, the Commission once again invokes the specter of anticompetitive "foreclosure" as a direct consequence of the transaction. As I have made clear on previous occasions,(1) foreclosure theories are generally unconvincing as a rationale for antitrust enforcement. The current case provides scant basis for revising this conclusion.
The theory of harm presented here is the same as -- and thus shares all of the defects of -- that offered in Silicon Graphics, Inc. ("SGI").(2) In SGI, the Commission alleged that the merger of a computer hardware manufacturer (SGI) and two software vendors (Alias and Wavefront) would result in the post-acquisition "foreclosure" of other independent software suppliers, leading to monopoly prices for graphics software. The Commission claimed that because the acquisition would give SGI its own in-house software producers, SGI no longer would allow unaffiliated software vendors access to its hardware platform.
In the current incarnation of this theory, Cadence is cast in the role of SGI and CCT in the role of the software vendors. The Commission alleges that Cadence no longer will allow independent suppliers of "routing" software -- the type of software sold by CCT -- to write programs that can interface with other IC layout programs in the Cadence suite. To mitigate these supposed anticompetitive incentives, the proposed order would require Cadence to provide independent vendors of routing software access to its "Independent Software Interface Programs" (e.g., to its "Connections Program") on terms "no less favorable" than the terms offered to other independent software vendors.(3)
The logic of the proposed complaint is fundamentally flawed. Even if we assume arguendo -- as the proposed complaint in this case does -- that Cadence is "dominant" in the supply of software components complementary to the router,(4) the fact remains that it has no incentive to restrict the supply of routers. I noted in SGI that "SGI ha[d] strong incentives to induce expanded supply of SGI-compatible software: increasing the supply of compatible software (or of any complementary product) increases the demand for SGI's workstations."(5) The same is true here: the introduction of a lower-priced or higher-quality routing program increases the value of Cadence's "dominant" position in the sale of software complementary to the router, because it increases the demand for Cadence design software, thereby allowing Cadence to increase the price and/or the output of these programs. Despite the majority's assertions to the contrary,(6) this is true whether or not Cadence has vertically integrated into the sale of routing software, for efficient entry into the production of routing software increases the joint profits of the entrant and Cadence. If the Commission is correct that Cadence is "dominant" in the supply of software components complementary to routers, then of course Cadence may be in a position to expropriate -- e.g., via royalties paid to Cadence by the entrant for the right to "connect" to Cadence's software -- some or all of the "efficiency rents" that otherwise would accrue to an efficient entrant. This, however, would constitute harm to a competitor, not to competition, and Cadence would have no incentive to set any such rates so high as to preclude entry.
The theory of harm and the remedy proposed here also share many of the flaws that I pointed out in Time Warner.(7) In that case the Commission's action was based to a significant degree on the argument that increased vertical integration into cable programming on the part of Time Warner and Tele-Communications, Inc. would increase those firms' incentives to reduce the supply of independently produced television programming. Carried to its logical conclusion, this theory of harm constitutes a basis for challenging any vertical integration by large cable operators or large programmers -- even vertical integration occurring via de novo entry by a cable operator into the programming market or de novo entry by a programmer into distribution.
Now apply this train of thought to the current matter. Contrary to the analysis presented above, suppose that somehow Cadence could profit anticompetitively from denying interconnection rights to independent router vendors. If that were so, then it would not be sufficient merely to prevent Cadence from acquiring producers of complementary software. Rather, the Commission would have to take the further step of preventing Cadence from developing its own routers; for under the anticompetitive theory advanced in the complaint, any vertical integration by Cadence into routers, whether accomplished by acquisition or through internal expansion, would engender equivalent post-integration incentives to "foreclose" independent vendors of routing software.(8) Of course, as I noted in Time Warner, there is likely to be little enthusiasm for such a policy because there is a general predisposition to regard internal capacity expansion as procompetitive.(9)
Not only am I unpersuaded that Cadence's acquisition of CCT is likely to reduce competition in any relevant market, but -- as in SGI and Time Warner -- I would find the proposed order unacceptable even were I convinced as to liability. As in Time Warner, the Commission seeks to impose a "most favored nations" clause that would require Cadence to allow all independent router developers to participate in its software interface programs on terms that are "no less favorable than the terms applicable to any other participants in" those interface programs. Even apart from the usual problems with "most favored nations" clauses in consent orders,(10) this order -- as in both SGI and Time Warner -- will require that the Commission continuously regulate the prices and other conditions of access.
Indeed, compared to the proposed order in the present case, the order in Time Warner was a model of clarity and enforceability. What does it mean to mandate treatment "no less favorable than" that granted to others, when Cadence's current Connections Program -- with well over 100 participants -- allows access prices to differ substantially across participants and imposes substantial restrictions on the breadth and scope of the permitted connection rights?(11) Does it mean that router vendors pay a connection fee no higher than the highest fee paid by an existing participant? Or would they pay a fee no higher than the current lowest fee? Or does it mean something else? Router vendors surely will argue for the second interpretation -- a view also apparently shared by the Commission majority(12) -- yet there is no obvious reason why router vendors should be entitled to such a Commission-mandated preferential pricing arrangement, and neither the majority nor the Analysis to Aid Public Comment has offered one.
Similarly, does the "no less favorable" requirement mandate that the vendors of routing software obtain access rights as broad as the broadest rights now granted, or simply no worse than the narrowest now granted? And since the current Connections contracts are terminable at will by either party with 30 days' notice, does "no less favorable" mean only that router vendors must be given the same termination terms as other software vendors, or does it mean something else (e.g., termination only for cause, where the "reasonableness" of the termination is subject to ex post evaluation by the Commission)?(13) The former interpretation of the order seems the most straightforward; however, it is also one that essentially would nullify the protection of independent router vendors and thus would render the order meaningless.(14)
The preceding suggests strongly that the real (albeit unstated) goal of the order is not to nullify any actual anticompetitive effects from the proposed transaction, but rather to invalidate the principal aspects of Cadence's "Connections Program" (i.e., the ability to charge different connection fees and to terminate vendors at will) without demonstrating that the program's provisions violate the law. There is little reason to believe that this program is harmful to competition, and there are strong efficiency reasons for allowing Cadence to set different fees for different vendors. Moreover, setting a uniform fee would result in price increases to at least some vendors.
Because I do not accept the majority's theory of liability in this case, and because I find the proposed remedy at best unenforceable and at worst competitively harmful, I dissent.
1. See Dissenting Statement of Commissioner Roscoe B. Starek, III, in Time Warner Inc., et al., Docket No. C-3709 (consent order, Feb. 3, 1997); Dissenting Statement of Commissioner Roscoe B. Starek, III, in Waterous Company, Inc. and Hale Products, Inc., Docket Nos. C-3693 & C-3694 (consent orders, Nov. 22, 1996); Dissenting Statement of Commissioner Roscoe B. Starek, III, in Silicon Graphics, Inc. (Alias Research, Inc., and Wavefront Technologies, Inc.), Docket No. C-3626 (consent order, Nov. 14, 1995); Remarks of Commissioner Roscoe B. Starek, III, "Reinventing Antitrust Enforcement? Antitrust at the FTC in 1995 and Beyond," remarks before a conference on "A New Age of Antitrust Enforcement: Antitrust in 1995" (Marina del Rey, California, Feb. 24, 1995).
2. Supra note 1.
3. Proposed order, ¶ II.A.
4. The anticompetitive theory requires Cadence to have substantial monopoly power: if there were numerous good alternatives to Cadence's suite, other independent vendors of routing software could affiliate with them and there would be no "foreclosure."
5. Dissenting Statement in SGI, supra note 1, at 2. Moreover, as was also true in SGI, the description of the premerger state of competition set forth in the complaint itself tends to exclude the possibility of substantial postmerger foreclosure. In SGI, the complaint alleged that software producers other than Alias and Wavefront were competitively insignificant prior to the merger, and that premerger entry barriers were high. Similarly, the current complaint (¶ 11) alleges that there are substantial premerger barriers to entry into the market for the kind of "router" software that CCT produces. But one cannot find both that the premerger supply elasticity of substitutable software is virtually zero and that the merger would result in the substantial postmerger foreclosure of independent software producers. If entry into constraint-driven, shape-based IC router software is effectively blocked premerger, as the complaint contends, it cannot also be the case that the merger would cause a substantial incremental reduction in entry opportunities.
6. The majority asserts that "Cadence clearly also has an incentive to prevent loss of sales in its competing products." (Majority Statement at 4; emphasis in original.) Similarly, the Analysis of Proposed Consent Order to Aid Public Comment simply asserts (at 5) that "Cadence does not . . . have incentives to provide access to its integrated circuit layout environments to suppliers of integrated circuit layout tools that compete with Cadence products." Because neither the majority statement nor the Analysis to Aid Public Comment describes how this conclusion was reached, it is difficult to identify precisely the source of the erroneous reasoning. Chiefly, however, it seems to reflect a manifestation of the "sunk cost fallacy," whereby it is argued that because Cadence has now sunk a large sum of money into acquiring CCT, this in and of itself would provide Cadence with an incentive not to deal with independent vendors of complements. This reasoning, of course, is fallacious: the cost incurred by Cadence in acquiring CCT -- whether a large or a small sum -- is irrelevant to profit-maximizing behavior once incurred, for bygones are forever bygones. The introduction of a superior new router, even if by an independent vendor, will increase the joint profits of Cadence and this vendor (irrespective of the amount spent in acquiring CCT), and both parties will have a profit incentive to facilitate its introduction.
Moreover, the majority also imputes a sinister motive to Cadence's reluctance to deal with certain competitors, while failing to acknowledge that this reluctance almost surely represents a legitimate and well-founded interest in protecting its intellectual property. As the Analysis to Aid Public Comment notes (at 4): "Cadence's leading competitor in the supply of integrated circuit layout environments, Avant! Corporation, has been charged criminally with conspiracy and theft of trade secrets from Cadence, and several top Avant! executives have been charged criminally as well."
7. See my Dissenting Statement in Time Warner Inc., et al., supra note 1.
8. Thus, it is unclear how the Commission should respond, under the logic of its complaint, were Cadence to introduce an internally developed software program (now provided by one or more independent vendors) that is complementary to its "dominant" suite of programs. Obviously Cadence would be in a position (similar to that alleged in the Commission's complaint) to block access to the Cadence design software if it wanted to. Even if Cadence did not terminate the independent vendors, consistent application of the economic logic of the present complaint seemingly would require the Commission to seek a prophylactic "open access" order against Cadence similar to the order sought here. This enforcement policy would of course have a number of adverse competitive consequences, including deterrence of Cadence from efficiently entering complementary software lines through internal expansion.
The observation in note 3 of the majority statement that antitrust law has treated vertical integration by merger differently from internal vertical integration "for more than one hundred years" suggests that I do not recognize that the law provides for differential treatment of mergers and internal expansion. I simply intended to point out the illogicality of finding vertical integration with identical economic consequences to be illegal under the Commission's standards of merger review, when that integration would be of no concern (and might even be applauded) if it resulted from simple internal expansion.
9. In the present case, as in Time Warner, the Commission has alleged the existence of substantial pre-acquisition market power in both vertically related markets (routing software and the rest of the IC layout "suite" here, see complaint ¶¶ 9-11, and cable television programming and distribution in Time Warner). Under these circumstances, there is a straightforward reason why vertical integration is both profitable and procompetitive (i.e., likely to result in lower prices to consumers): vertical integration would yield only one monopoly markup by the integrated firm, rather than separate markups (as in the pre-integration situation) by Cadence and CCT.
10. As I noted in Time Warner, these clauses have the capacity to cause all prices to rise rather than to fall. Dissenting Statement, supra note 1, at 20. The majority (at 5) seems comfortable with this outcome, provided that all vendors pay the same price.
11. For example, CCT had been permitted to participate in the Connections Program with its printed circuit board router but not with its IC router.
12. See Majority Statement at note 9.
13. Moreover, does the terminability of the Connections contract on 30 days' notice mean that the "no less favorable" requirement might need to be reviewed every 30 days?
14. The majority implies (Majority Statement at note 9) that the exercise of this right would indeed constitute a violation of the order.