Historically Brazil faces the social problem of corruption and fraud in its public administration. Gangs have specialized in circumventing the system in order to subtract values from the public treasury, either by fraudulent public bidding or by invading government systems and their benefit programs paid to citizens. A pioneering initiative of the Ministry of Labor is reversing this equation that has always favored fraudsters: with the use of Big Data Analytics and Business Intelligence, with the crossing of several governmental databases (federal revenue, official banks, social security, federal police, electoral court and several others) in Teradata and Hadoop environments, some petabytes are searched from the perspective of the business and anti-fraud areas in search of points of convergence and possible investigation. Only one of the benefits analyzed has led to the blocking of more than 40,000 applications with only 22 false-positive and savings of over 600 million reais (approximately 200 million dollars). This research intends to answer the following question: what is the potential of savings with the use of similar tool for analysis of other official benefits? From a correlation analysis and the respective construction of scenarios in function of the values regularly paid and potential fraud is expected a realistic projection and consequent possible cost-benefit analysis of investment in information technology. The literature on the subject is not yet extensive, but works such as Baesens (2015) entitled "Fraud Analytics using descriptive, predictive and social network techiques" provide a robust theoretical basis for the challenge presented. Another source for theoretical reference is reserved manuals of the federal police of the country that for years has been fighting fraud in its analog form but recently also began to focus on electronic fraud.
Preliminary results point to a potential savings of more than 3 billion reais (approximately US $ 1 billion) per year, with a small increase in investments made in the construction of the anti-fraud model, currently only analyzing the unemployment insurance paid to those workers registered in the market that may lose their jobs, up to 1.500 reais for up to five months. So far, the total budget for the project is in the order of 80 million reais (approximately US $ 20 million), with transfer of technology and knowledge by the company contracted to technicians in the service of the government.
The creativity of the fraudsters seems unlimited. They create false individuals, take advantage of data from dead people or just take advantage of the vast national territory and the lack of alignment and synchrony between the public databases to use data from a real worker in several other locations. The country has more than 70 million registered workers and more than 1 million operating companies. Crossing the volume of data from these individuals is a task that requires the latest technology and use of the state of the art of Business Intelligence, as recommended in Taurion's works (2014), entitled "Big Data" or Reed (2015) Data Analytics ".
