Using Predictive Analytics to Prevent Electronic Invoice Fraud: How the Brazilian State of Paraña Curbs Fake Companies

In Brazil, electronic invoicing is mandatory for all VAT taxpayers, with transactions requiring real-time authorization from state tax administrations. However, this system has been exploited by fraudulent companies issuing fake invoices to generate illegitimate tax credits for buyers, causing significant financial losses. To combat this, Paraña's state tax administration developed SiMFE, an automated system powered by Teradata's analytics platform. SiMFE processes over 10 million electronic invoices weekly, applying business rules to identify fake companies engaged in tax fraud. Fraudulent company registration numbers are flagged and sent to the transactional system, blocking their access and preventing further fraudulent activity. The platform also provides tax auditors with detailed BI reports on blocked transactions, including identifying buyers who benefited from the fraud. Since its launch three years ago, SiMFE has achieved an 80% accuracy rate, blocking over 2,100 fake companies and preventing an estimated $3.1 billion USD in fraudulent tax credits. This success highlights the importance of detailed transaction data and a unified data and analytics infrastructure in curbing tax fraud and ensuring transparency.