This short piece serves as an introduction to Thibault Schrepel & Teodora Groza’s working paper
entitled “Computing Innovation Competition” available here.
***
1. Different visions of competition
There is little consensus in the antitrust circles. Nonetheless, one of the fundamental points of agreement is that innovation fosters economic growth—hence competition policy should be targeted towards preserving, promoting, and encouraging innovation. Unsurprisingly, much has been written on the topic: a Google Scholar search for the terms “antitrust” and “innovation” gives more than 290,000 results, and they follow an ascending trend. This trend is palpable in courtrooms, too, as adjudicators on both sides of the Atlantic have been factoring innovation concerns in their judgments for decades.
The recognition of innovation as an important parameter of competition is well-established. Recently, however, a more specific concept—innovation competition—has gained traction. To understand the difference between innovation as a parameter of competition and innovation competition, we need to make sense of two different visions of competition. Traditionally, competition was considered to be driven by prices: firms compete to offer the lowest price for similar goods or services. This gave rise to the concept of price competition, under which innovation was a factor which could help firms compete more vigorously.
Today, scholars argue that “innovation has become the dominant mode of competition,” provoking a shift from price competition to innovation competition. This new mode of competition involves companies competing on the basis of novel ideas, technologies, or capabilities that can redefine markets and consumer experiences. Innovation competition thus elevates innovation from a parameter amongst others to the main driver of competition.
This evolving view of competition is slowly infiltrating the work of agencies and courts. In the EU, the earliest mention of the concept dates back to a 2002 CJEU decision, whereas on the other side of the Atlantic the FTC has grappled with the notion since 1990. Much has changed in the meantime, both in terms of decisional practice and market realities. But one trend in particular merits our attention: the increasing adoption of computational tools by antitrust agencies. The evolution of computational antitrust from an insular phenomenon to a global trend has been extensively documented by the annual reports of the Stanford Computational Antitrust Project. But what does this mean for the relationship between antitrust and innovation competition?
At a first glance, we might be tempted to think that innovative enforcement tools go hand in hand with assessing innovation competition. But it’s not so simple. Most of the computational tools developed so far have been centered on processing big data through quantifiable metrics. Can innovation be measured? And, more importantly, can it be computed?
2. The Challenge for Antitrust Agencies
Antitrust agencies face significant challenges in adapting their frameworks to account for innovation competition. The primary issue lies in the difficulty of developing metrics for assessing innovation. Traditional antitrust analysis typically focuses on measurable variables, such as price levels and market shares. These can be easily computed. However, innovation is difficult to measure, and as the OECD points out, assessing innovation requires “mov[ing] beyond aggregate numbers or indices.” Innovation requires know-how, long-term strategic investments, capabilities, incentives—factors that are not easily quantified.
This brings us to the second issue. Economists agree that innovation is the most important driver for economic growth, but they hardly agree on anything else. We have no consensus on whether and how market structures affect innovation, how to balance trade-offs between present benefits and future costs (and vice versa), the impact of M&A on innovation incentives—and the list can go on.
In the absence of an economic theory of innovation to fall back on, antitrust agencies are required to opt for a case-by-case approach. Instead of measuring innovation in the abstract, they rely on several proxies to consider the unique circumstances of each case, such as patent activity, levels of R&D investment, the speed with which new products and processes are introduced, etc. On one hand, it allows for flexibility, enabling agencies to consider the unique circumstances of each case. On the other hand, it is increasingly impractical given the rising volume of cases involving complex innovation-related dynamics. With the growth of digital markets, agencies are inundated with cases that require them to assess large amounts of data and various interconnected factors, challenging the scalability and efficiency of a case-by-case approach.
3. Innovation Competition Meets Computation
Despite the difficulties of measuring innovation, antitrust agencies are not armless. Far from it: computational tools can help with fine-tuning the assessment of innovation at all stages of antitrust enforcement: understanding pre-existing case law; defining relevant markets; detecting and analyzing anti-competitive practices; analyzing mergers; designing remedies; and offering compliance tools.
Innovation concerns permeate agencies’ decisional practice, across all industries and practice areas. Consequently, it is difficult to bring together these separate threads and make sense of how innovation is featured in past cases. To overcome this hurdle, agencies use network analysis to map out trends in case law, with tools to visualize cases by industry and track how innovation considerations are plugged in past decisions, as demonstrated by France’s competition authority.
Defining relevant markets, especially in tech-driven sectors, remains complex. Computational techniques, such as natural language processing (NLP), help identify emerging “innovation markets” as well as “nascent markets,” where traditional data on prices and quantities is limited. Geographic information systems (GIS) and business intelligence platforms are also relied on to assess competitive landscapes by mapping market players and their reach.
For identifying anti-competitive practices, some authorities use neural networks to monitor bid interactions or detect pricing collusion among digital retailers. Additionally, machine learning tools can identify restrictive contract terms or fake reviews, both of which can hinder innovation by misrepresenting product value and/or limiting market entry for competitors.
Merger analysis now involves computational assessments of merging firms’ innovative capacities, such as R&D investments and patent activities. In the future, it is envisage that agencies will use conglomerate charts and network analysis to understand the relationship between merging parties’ innovation pipelines, aiding in decisions to approve or block deals based on potential impacts on competition.
In imposing remedies, tools like automated compliance monitoring ensure that companies adhere to conditions aimed at safeguarding competition. Agencies are also exploring tools to help companies self-assess compliance with antitrust and AI governance standards, as seen with Singapore’s “AI Verify Toolkit.” These computational advances help streamline enforcement, making it possible for antitrust agencies to address the complexities of innovation competition more effectively and with greater predictive power.
4. Conclusion
Innovation is hard to measure, but it need not be hard to compute. As our paper documents, antitrust agencies across the world already rely on computational tools to assess innovation at every stage of antitrust analyses, from market definition to imposing remedies. The question that remains unanswered is whether computational tools will get us closer to developing an economic theory of innovation by enhancing our ability to process big data and to model the behavior of firms and industries. Our hunch is that we will get there.
Teodora Groza
| Citation: Teodora Groza, From Theory to Practice: Computing ‘Innovation Competition’ in Antitrust, Network Law Review, Fall 2024. |
