CAUSAL INFERENCE-BASED DIGITAL PAYMENT FRAUD DETECTION: FROM FINANCIAL SECURITY TO ECONOMY-WIDE RESILIENCE

Abstract

With the explosive expansion of digital payment systems, financial fraud has now become one of the most serious threats facing economic stability in many sectors. This paper presents a critical analysis of how methods for causal inference could contribute to improvement of fraud detection by revealing underlying patterns rather than correlations. The paper presents a theoretical model that integrates machine learning and causal analysis methods to improve the differentiation between legitimate and fraudulent transactions. Through the detection of interaction networks and behavioural patterns, the methodology attains a higher level of accuracy in identifying sophisticated fraud schemes than a traditional rule-based system. The results propose that causality methods do not just mitigate false positives in financial industries but they also present actionable risk controls for the application domain of e-commerce, healthcare and digital transaction processing. The research serves to enhance financial security efforts by designing a more thorough methodology which is also flexible enough to evolve in response to advances in fraud techniques. Next steps include generalizing causal models to cover new threats in decentralized finance and cross-border payments.