Review of recent empirical literature on screening methods for cartel detection

Abstract

Objective: The objective of this article is to present a review of recent empirical literature regarding the use of screening methods to detect cartels in Brazil and around the world. Such methods have been increasingly used by antitrust authorities as they complement and reinforce traditional reactive methods of cartel detection, such as leniency agreements. There are several studies that discuss its applications, most focusing on the behavioral screening approach. This article aims to complement such work in two ways: (i) evaluate the most recent literature on the topic, which has specialized, for instance, in machine learning methodologies; and (ii) discuss applications of structural approaches, which are more simplified detection methods than behavioral screening, but also used by antitrust authorities and in the literature.

Method: Literature review and qualitative analysis on national and international cartel detection.

Conclusions: Several antitrust authorities have been following the recommendations of international organizations to combine the use of reactive and proactive methods in cartel detection. There are significant initiatives in the use of screenings, especially in the context of public tenders, given the increased availability of data. In empirical literature, machine learning techniques have gained traction, either due to the challenges in identifying econometric models or the rising complexity of collusive agreements. This represents a new trend primarily focused on enhancing predictability and precision in the cartel detection process, rather than drawing conclusions about causality. Given the diverse sectoral and regional dynamics in the economy, its use still requires caution on the part of researchers and antitrust authorities, but it undoubtedly represents a significant and promising advancement in the fight against cartels.