Developing a CIB-based credit scoring model for retail loan risk assessment: an experimental study on City Bank PLC

Abstract

This internship report examines the creation and testing of a CIB-based credit scoring model designed to enhance risk assessments for retail loans at City Bank PLC. Conducted under the Retail Underwriting Unit, the study was based on over 200 anonymized loan files of new-to-bank applicants, using variables such as Credit Exposure, NPI status, Loan Utilization, and Product Mix from CIB reports. Through score mapping, correlation analysis, and iterative refinement, the experimental model aimed to predict repayment behavior and filter high-risk borrowers efficiently. The findings showed weak correlation between the initial score design and loan default rates, highlighting the need for further data calibration. This report emphasizes the strategic role of automated scoring in enhancing underwriting efficiency and minimizing default risks, while offering insights into operational challenges and future integration opportunities for the model within the bank’s retail credit processes.