Collusion Detection in Public Procurement with Limited Information

Abstract Public procurement data sets usually lack detailed data that are needed to implement existing methods of collusion detection. We design a method to identify and test for bid rigging in procurement auctions using limited information. The method can be applied to limited data sets using standard machine learning tools and statistical software. We implement the methodology to a unique data set of entire Turkish public procurement auctions in years 2004-2010, numbering 470,419. We uncover the structure of collusive behavior in Turkish public procurement auctions. We find that collusion significantly increases procurement costs and decreases cost-effectiveness