Computational antitrust promises not only to help antitrust agencies preside over increasingly complex and dynamic markets, but also to provide companies with the tools to assess and enforce compliance with antitrust laws. The Stanford Computational Antitrust project is pushing new boundaries and exploring new territories in the space.
Computational antitrust for agencies
Current research dedicated to computational antitrust has been primarily dedicated to supporting antitrust agencies. These agencies face new challenges that only computational tools can resolve: they need to analyze billions of data points to detect practices, they need to automate the analysis of the documentation they receive from companies, they also need to rely on new techniques to streamline internal processes, etc.
On the ground, we see that agencies are proactively adopting these tools while researchers are pushing new ideas (see here). Some academics have shown how agencies can code their decisions and identify new patterns from network analysis. Researchers have also explained how computational tools can help gather external data using web scraping or natural language processing. Others have explained how agencies could use computational solutions to better analyze practices and mergers after they have collected data, and how agent-based modeling could simulate the effects of decisions and policies.
Computational antitrust for companies
“The Making of An Antitrust API: Proof of Concept” fills the gap in the literature by offering an innovative solution for companies. Specifically, this article serves as a proof of concept whose aim is to guide antitrust agencies in creating a decision-trees-based antitrust compliance API intended for market players.
The antitrust API this article introduces has been voluntarily developed using a “no-code” solution. The goal is to offer a method that all 100+ competition authorities can use. The method requires little to no computer science expertise, a small team (one or two employees), no dedicated budget, and only a few weeks of work.
The companion article offers access to the open-access prototype that automates compliance with Article 102 TFEU. It discusses its limitations and lessons to be learned. In short, the article introduces a new computational antitrust that is fundamentally non-confrontational and whose goal is to help companies willing to comply with antitrust rules. The link to the API (free for everyone to use) is right here.