Early Prediction Detection of Retail and Corporate Credit Risks Using Machine Learning Algorithms

Abstract

Nowadays, banks operate in a highly dynamic environment where substantial vulnerability to credit risk exposures threatens their performance by affecting the quality of bank portfolios and increasing their vulnerability to insolvency. In this context, the paper reviews the existing literature and finds no studies investigating the determinants of retail and corporate credit risk using machine learning techniques to enhance the predictive performance of bank credit risk exposures. Consequently, the paper aims to utilize machine learning algorithms, regression analysis, and classification models to identify the most effective predictive model that can improve banks' credit risk prediction capabilities. It will cover the period from 2011 to 2023 and analyze a sample of 26 banks operating in Egypt. Additionally, it classifies credit risk into retail and corporate categories to develop more robust predictive models tailored for the retail and corporate sectors of bank credit risk management, thereby underscoring the paper’s novelty. The findings showed that the Random Forest and Kernel SVM can be used to improve the prediction of corporate and retail credit risk by utilizing bank-specific factors like profitability, liquidity, income diversification, capital, asset size, and operating efficiency, as well as macroeconomic factors like external debt, inflation, exchange rate, GDP, interest rate, and foreign direct investment.