Exploring Personal Digital Footprints and Credit Risk Assessment

Traditional credit evaluation for a long time depends on financial statements and credit reports; it does not include people who do not have an existing credit history. Digital footprint is a form of unstructured data. The form includes the social and e-commerce behavior, the transaction with the finance and the geolocation data, which could be analyzed to determine the repayment capacity and willingness. In this paper, I review the types and characteristics of personal digital footprints, and study how they are applied in credit risk models, and explain their effectiveness with behavioral economics and information asymmetry theory. A survey of national studies and cases, followed by national and international studies and cases, is conducted. Evidence from the FDIC and others is cited showing that even small digital footprint variables can produce AUC predictions as good as traditional credit scores. In short, the future developments of standardization technologies on federated learning and online learning systems as well as multimodal data fusion, can be anticipated. The studies in this paper can provide theoretical and practical help to the scientific application of digital footprint on credit risk assessment.