Abstract

The dynamics of credit risk assessment have changed so fast that it has surpassed the abilities of the traditional financial metrics. As financial markets continue to move toward awareness digitization, the nature of the behavior of borrowers, including making payment, regularity in account usage and behavior in the digital context, holds some untapped predictive value in determining credit risk. This paper aims to establish the extent to which incorporation of non-financial behavioral markers into credit analysis systems can be used exponentially to improve performance of the default prediction and warning systems of the borrowers. This research aims at coming up with a sound and adaptable hybrid model based on real-time behavioral data, which employs the concepts of machine learning as well, in view of the behavioral finance concepts. Ethical, regulatory, and operational issues associated with the employment of behavioral data also find a spot on the critical section of the paper examination. The results are expected to help not only to advance the academic discourse but also the feasible application in the field of banking, fintech, and regulations, which, eventually, will lead to more responsive, fair, and transparent credit systems.