Author: Nicolas Petit
This is a guest post from noted antitrust and competition law scholar Nicolas Petit. Petit is a Professor of Competition Law in the Department of Law at the European University Institute. He is also the co-founder of the Dynamic Competition Initiative with the University of California – Berkeley. Longtime blog readers will remember that I have reviewed Petit’s work before, both here and here. As I discussed in another recent post, Petit joined me on April 3 in Washington, D.C. at the ABA Antitrust Spring Meeting for the Chair’s Showcase Panel “Have We Been Doing It All Wrong” to discuss dynamic competition and related topics. Petit graciously agreed to share this lightly edited version of his opening remarks from that panel with readers of The Antitrust Attorney Blog. Enjoy! – Steve Cernak
I’m the only non-American on this panel, and as it happens, I’m French. As former President Bush once said, the trouble with the French is that they don’t have a word for “entrepreneur.” This should make me the least qualified today to talk about innovation. Now, while President Bush was incorrect, his joke highlights an important truth: understanding the entrepreneur is crucial for fostering innovation.
Today, I want to discuss how antitrust law can better support innovation. A common term used by antitrust people like me who care about innovation is “dynamic competition”, or competition through technological change.
My main point is this: antitrust law needs smart glasses to effectively support dynamic competition. The goal of the “dynamic competition” view is to supply them. We want to enhance antitrust law’s ability to identify restraints of competition that effectively harm dynamic competition, and those that do not.
The dynamic competition view makes three assumptions:
First, technological change is a vector of competition that can be better seen by antitrust law. Second, perceiving the augmented reality of dynamic competition through technological change can be done at reasonable cost. Third, an agenda of augmentation of the reality of antitrust is more acceptable than the jump in virtual reality proposed by Neobrandeisians or the weird world of Trumpian antitrust.
I want to elaborate on the first point. In some industries, technology is a vector of invisible competitive pressure that may complement, and perhaps dominate, product rivalry. In 1946, Joseph Schumpeter had an interesting line about “the businessman [who] feels himself to be in a competitive situation even if he is alone in his field”. He warned against government experts who, failing to see any product rivalry with other firms conclude that “his competitive sorrows is all make believe”.
Today, the Schumpeter aphorism resonates vividly. The R&D investments of the Magnificent 7 represent effort levels consistent with cutthroat competition. Yet, agency and court cases give them short shrift, reaching one finding of unlawful monopolization in core platform segments after the other.
It is challenging for antitrust to deal with the coexistence of large R&D budgets and monopoly shares. The puzzle is the following: Is dynamic competition seen in R&D expenditure conditional on the maintenance of some base of monopoly power in a bottleneck market? Or is dynamic competition the outcome of collapsing monopoly power in that market?
These “correlations” between static and dynamic competition, as Richard Gilbert calls them, can be positive or negative depending on the industry.[1] Different innovation-minded policies may be required depending on how the correlations work.[2] If an industry works on a negative correlation between static and dynamic competition, antitrust forbearance will be the optimal policy to protect innovation. A generalization of that negative correlation was used by the Supreme Court in Trinko to exonerate nearly every refusal to deal from Section 2 liability.[3] Now, contrast this with an industry where a positive correlation exists between static and dynamic competition. In that scenario, strong antitrust enforcement can support dynamic competition through the commodification of aging firms’ monopoly rents.[4] This may be the implicit, perhaps unconscious, logic underpinning the liability theories in United States v. Google and FTC v. Facebook.
Antitrust law’s selection of one correlation model or the other cannot be made abstractly. A concrete evaluation of (i) whether the industry has dynamic competition potential, and (ii) whether it is monopoly power, or static competition, that supports it, is needed. And note that if an industry has no dynamic competition potential – in clear, technological change is slow, because of a fixed production possibility frontier – then traditional antitrust enforcement should make markets work well.
A related way to look at this is to ask whether antitrust law should adopt a holistic lens on competition. Let us consider for a minute that innovation, technology, or R&D activity are forms of “broad spectrum competition”, “hypercompetition”, or “superimposed competition”.[5] Now, today’s partial equilibrium logic prevents a twin evaluation of that competitive layer and of rivalry within a relevant market. We do not examine what I have called the “moligopoly” dimension in the picture. The point is this: a business organization is a whole. It is not one thing when it competes with a field of peer firms in the technological environment, and an entirely other thing when it competes with rivals in the product space.
Now, to be fair, antitrust law has attempted to adopt a broader lens. The US Guidelines for the licensing of IP refer to R&D markets. The EU market definition notice of 2024 takes note of “structural market transitions”.[6] The EU horizontal merger guidelines contain an explicit reference to dynamic competition. In merger cases, agencies have experimented with new concepts. AMAT/TEL or Nielsen/Arbitron focused on future R&D or prospective product markets. Dow/Dupont introduced a focus on innovation “spaces” in R&D pipeline industries. Booking/eTraveli used ecosystem language to justify prohibition.
But talking about dynamic competition does not mean seeing dynamic competition. A proposition like “Google competes with Amazon” will generally attract laughter on planet antitrust. The idea that OpenAI had Chinese competitors – and perhaps some European ones – was delusion until the DeepSeek release. And a conventional application of merger tools led the UK CMA and CAT to hold that GIPHY had the ability and incentives to outcompete Facebook.
The issue, in my view, is there is no well-articulated economic and legal framework for dynamic competition, innovation, or technological change, and related analytical tools. In practice, innovation-related concepts pop up randomly in cases, sometimes to support antitrust or merger prohibitions, less frequently to credit pro-competitive justifications.[7]
A more thoughtful framework for dynamic competition requires deepening antitrust law engagement with four high-level issues. The first one is ‘out-of-market’ competition. In some industries, market power is not just constrained by direct entry threats within the relevant market. Indirect entry that moves surplus to adjacent or lateral markets is equally as important.
The second one is the discipline of markets other than goods markets for inter-firm competition. Financial, managerial and entrepreneurial markets where firms purchase inputs to technological change are equally as important as the technology markets mentioned in the Guidelines for the licensing of IP.
The third one is change. Firms generally change by developing complements, consciously or unconsciously. Antitrust or merger restrictions on vertical or lateral integration may slow down change.
The last one is the role of firms as internal markets for the coordination of innovation. Merger law’s touchpoints with innovation focus on the merger’s impact on “innovation competition,” not on “innovation potential.” A clear insight of a field known as the ‘capabilities literature’ is, however, that innovation can go up following a merger, even if there is a reduction in rivalry. Horizontal mergers enable beneficial capability combinations through the reallocation of (non-sunk) duplicative resources to invention or to innovation adoption. Non-horizontal mergers enable beneficial capability combinations through the alignment of related capability sets. Post-merger gains in innovation potential deserve to be recognized and sanctioned.
_________________________________________________________
[1] Or its maturity.
[2] If, of course, dynamic competition matters more than static competition as some may believe.
[3] A more empirical consideration of output growth supported the Supreme Court holding in Amex, that restraints of competition like antisteering clauses were compatible with dynamic competition.
[4] The idea is that the antitrust enforcement shock will limit the payoffs of monopoly maintenance, and increase the monopoly firm incentives to exit, diversify, and change towards non monopoly markets.
[5] In economics, one would call this expected competition with net present value.
[6] EU market definition notice, 2024, paras 21 and 55.
[7] For example, the US guidelines for the licensing of technology refer to R&D markets but the US merger guidelines do not. Besides, the definition focuses on substitutes, while dynamic competition often arises from complementarities.
Image by Michelle Pitzel from Pixabay