CREDIT CARD FRAUD AND TRANSACTION COSTS: THE ROLE OF MACHINE LEARNING IN ECONOMIC EFFICIENCY

Abstract

Credit card fraud represents one of the main economic challenges of the contemporary financial system, affecting efficiency, trust, and institutional governance. This study analyzes the economic impact of fraud from the perspective of Transaction Cost Theory, examining how machine learning can serve as a tool for enhancing efficiency and promoting institutional innovation. Based on public financial transaction data and theoretical review, the paper interprets the use of machine learning algorithms as a mechanism for rationalizing monitoring and verification costs, thus reducing losses and improving productivity. The results show that models such as Logistic Regression, Support Vector Machines, and Decision Trees have strong potential to detect anomalous patterns, decrease audit costs, and enhance confidence in the financial system. It is concluded that machine learning represents a Schumpeterian innovation capable of reducing transaction costs and increasing economic efficiency, establishing itself as a strategic element for the modernization and security of financial institutions in Brazil.