Online payment scenarios : a study on user payment behavior interventions to combat fraud risks

Abstract

Online scams have become a global social issue, with their complexity and concealment posing significant challenges for early detection and post-incident enforcements. Consequently, intervention to users in the middle of the scam is particularly crucial. However, many users tend to overlook intervention warnings during a scam, leading to ineffective prevention efforts. This study focuses on online payment scenarios and conducts an empirical analysis based on a major online payment platform in China. By leveraging historical observational data and randomized controlled experiments, the study explores the characteristics of scam-prone populations and evaluates the impact of different intervention measures on user payment behavior, providing both theoretical support and practical guidance for anti-scam intervention strategies. This study makes several contributions. It is the first to propose an intrinsic economic model of scams from the perspective of supply and demand. Additionally, it develops a scam intervention research framework based on the dual-system decision-making model in behavioral economics. The study identifies key demographic characteristics of scam victims, such as gender and age, as well as the distribution patterns of scam types and financial losses within online payment scenarios. These findings offer valuable insights for identifying scam-prone populations and designing effective user behavior interventions. Moreover, through randomized controlled experiments, the study demonstrates the feasibility and theoretical foundation of interventions in altering user payment behavior and reducing scam risks. The findings have significant practical implications for financial institutions struggling with low scam detection accuracy, helping them design intervention strategies that balance user payment experience with risk mitigation. The study has certain limitations. Due to incomplete access to victim data, it may underestimate the actual scale of scams and cannot precisely quantify the effectiveness of scam reduction. Additionally, the study is not able to systematically assess the efficiency loss to usual payment users due to interventions.