How to Use Data Science in Economics: A Classroom Game Based on Cartel Detection

The intersection of economic theory and data science serves as a critical frontier for modern regulatory enforcement, particularly regarding the detection of illicit market behavior. By simulating a procurement market, participants engage in a multi-stage exercise that transitions from the strategic decision-making of firms to the analytical rigor of antitrust authorities. This pedagogical framework bridges the gap between abstract market dynamics and the practical application of machine learning, providing a comprehensive understanding of how collusive bidding patterns are identified and prosecuted in contemporary economies.

The simulation operates across three distinct phases, requiring participants to first adopt the role of firms alternating between competitive and collusive strategies before shifting to the perspective of regulators. In the final stage, students employ statistical screens and machine-learning models to classify tenders as either suspicious or legitimate. This process emphasizes the inherent trade-offs in antitrust enforcement, specifically the necessity of balancing high detection rates against the risk of false positives, which carry significant real-world legal and economic consequences.

Designed for groups ranging from 6 to 32 participants, the exercise offers a scalable approach to teaching complex regulatory concepts. Technical implementation is supported by formal mathematical formulas for bid-rigging screens and an integrated R code walkthrough. Utilizing a random forest model, the simulation demonstrates a high degree of efficacy, achieving an 83.4% accuracy rate in identifying collusive behavior. By combining these quantitative tools with a nuanced understanding of firm incentives, the curriculum prepares students to navigate the evolving landscape of digital market oversight and evidence-based policy intervention.