Modeling Digital Credit Risk in Commercial Banks with Regression and Algorithms

The advent and rapid growth of digital credit has made financial access easier. The population, traditionally considered “unbankable” can now access the much desired funds for both household and business needs. Though faster and convenient, digital loans, carry with them unique risks due to uncommon characteristics such as quicker disbursements upon request, proxy borrowing, lack of collateral, improper customer identity verification and unclear purposes for the funds. These factors, along with the limitations of conventional risk models, which are static, necessitate the adoption of dynamic models capable of capturing intricate and nonlinear patterns among variables. This study addressed this gap by developing predictive models using both regression analysis and machine learning algorithms. Logistic and Cox PH regression models were used to measure the direction and magnitude of factors influencing digital credit risk while machine learning algorithms were used to capture complex and nonlinear interactions inherent in digital credit risk. Using 6,000 simulated loan records, necessitated by restricted access to real world data, the study evaluated performance of the four models, among which, random forest emerged as the most robust. Even though random forests model initially achieved an Area Under Curve score of 1.0, under simulated conditions, a 10-fold cross validation model produced a more realistic mean AUC of 0.90. This reflects strong discriminative ability and high predictive accuracy. These findings demonstrate that machine learning algorithms can enhance risk assessment framework in commercial banks thereby minimizing financial losses due to loan provisions. A limitation of this study was the use of simulated data, which may not fully capture the complexities of real world digital loan scenarios. Future research should focus on validating these models using real world borrower data from financial institutions to enhance generalizability and assess operational performance under actual lending conditions.