Benford's law for collusion detection: An experimental validation

Benford's Law has been widely used to detect anomalies in numerical data, including applications aimed at uncovering price manipulation and collusion, yet most applications rely on observational data in which market competitiveness is not directly controlled. This paper provides an experimental validation of Benford-based tests by exploiting exogenous variation in competition generated by a field entry experiment in rural Kenyan agricultural markets. Using transaction-level price data from Bergquist and Dinerstein (2020), we examine whether deviations from Benford's Law respond systematically to experimentally induced changes in market structure. We find that prices deviate substantially from Benford's Law in low-competition environments, and that conformity improves as competition increases. Importantly, this improvement is driven by entry of traders without prior connections to incumbents, while entry per se produces only modest changes.