Práticas anticoncorrenciais no Brasil: verificação analítica, empírica e sistêmica da integridade de aquisições públicas por pregões eletrônicos

Abstract

In September 2015, world leaders met at the United Nations (UN) headquarters to decide on a plan of action called Agenda 2030 to eradicate poverty, protect the planet, and ensure that people achieve peace and prosperity. The document contains a set of 17 Sustainable Development Goals (SDGs), which in sub-clause 16.5, provides for the "substantial reduction of corruption and bribery in all its forms." Using the foundations of Public Choice Theory and its ramified Economics of Corruption, this research in the format of "three articles" analytically and empirically verifies the integrity of public procurements through electronic auctions within the Public Procurement System of Brazil, also considering the dynamics of this systemic process, and defends the central thesis that there is a set of variables at micro and macro scale, proposed and tested, that impact the probability of occurrence of fraud or anti-competitive practices in the public tender Electronic Auction. An analysis is carried out based on the service bids that occurred in the Brazilian Federal Government Purchasing Portal, in the period from 2014 to 2018, aiming to enable the suggestion of actions to mitigate the problem. The first study uses the Newcomb-Benford Law (Law-NB), for exploratory empirical analysis of the values of electronic biddings, aiming to analyze their conformity to the Law-NB, and aiming to verify anomalies that represent indications of fraud. It can be stated that there are statistically significant anomalies in this analysis, and the study is pioneering for the Brazilian database. The main limitation of this study is the non-application of the Z-test, in order to provide greater robustness, and the possibility that the results obtained, instead of revealing indications of fraud, indicate a possible occurrence of persistent errors. The second study aims to investigate what factors are associated with the occurrence of anti-competitive practices in electronic biddings extracted from the same database. The analysis considered all services contracted between 2014 and 2017 and uses the logit regression model, with a dependent binary variable y, as a function of a vector x, of explanatory variables composed of the characteristics of the Process, and the Supplier, thus quantifying the relationship between the variables and the probability of fraud occurrence. It can be stated that the factors firm size and the Apparent Over-billing Index (AOI) are associated with the occurrence of fraud and anti-competitive practices in electronic auctions. The main limitation of this second study is that the proxy for "occurrence of fraud" is "having been punished", but not all fraudsters always commit fraud, and the efficiency of the "punishment system" is not taken into account. Another limitation was the low R2 of the regression. In the last step of the research, a small and highly stylized System Dynamics model of corruption and its potential impacts on fraud in public procurement in Brazil is presented. Focused on only two feedback loops, the model was calibrated based on the Brazilian Corruption Perception Index (CPI). A scenario analysis is also performed and based on these results policy implications are discussed. The main limitation of this third study is having a low internal validity. The results indicate that the level of corruption depends on the sensitivity of the actors and the delays in the system. Overall and considering the magnitude of the financial resources involved in public procurement processes, this research is an important tool for the learning process of public administrators.