Description:
Using big data from the Brazilian public procurement system, this research aims to investigate what factors are associated with the occurrence of anticompetitive practices in electronic bidding. Our analysis considers all services contracted between 2014 and 2017. The study uses the logit regression model with a dependent binary variable to model the probability of bidding fraud as a function of a vector of explanatory variables composed of the characteristics of the bidding process and of the supplier, in terms of firm size, thus quantifying the relationship between the explanatory variables and the probability of fraud occurrence. We find that firm size and the Apparent Overbilling Index which we propose are associated with the occurrence of fraud and anticompetitive practices in electronic auctions. This study contributes to the accumulation of empirical and theoretical knowledge, expanding safer bases for governments and society in the fight against corruption.