Designing Credit Risk Early-Warning System for Individual and Corporate Customers of The Bank Using Multiple Logit Comparison Model and Survival Function
نویسندگان
1 Department of Accounting, Qom Branch, Islamic Azad University, Qom, Iran
2 Qom branch, Islamic Azad University, Qom, Iran
3 Finance and accounting, Faculty of Humanities, Qom Islamic Azad University, Qom, Iran
4 Associate Professor of Azad University, Tehran markaz branch, Tehran, Iran Islamic Azad University of Central Tehran Branch.(Modern Financial Risk Research Group)
5 faculty of management and accounting, Islamic Azad university, Islam Shahr branch The Islamic Azad University of Eslamshahr.(Modern Financial Risk Research Group)
doi
چکیده
This article aims to estimate the credit risk of individual and corporate customers of Iran's banking system. The estimation of credit risks of banks, financial institutions and insurance companies is not possible without an accurate credit scoring of the customers. Credit scoring or credit rating is a process in which the credit amount of individual and corporate customers of the financial-credit institution and banks is measured using the information provided by the customers. The process makes it possible to obtain a wider knowledge of the people's situation to repay the credit received and, or to measure the loan default probability. The statistical data of 399 individual customers and 780 corporate customers from 2011 to 2019 (7500 data) are used to design credit risk models in this study. Multiple Logit Regression, Survival function and Support Vector Machine (SVM) are used to design credit risk models. The results indicate that the selected factors have a significant impact on the customer default probability and credit risk calculation, based on personality, financial and economic characteristics. The Comparison of the results obtained from the accuracy of the forecast shows a higher explanatory power of the Support Vector Machine model and survival function than the Multiple Logit model for both groups of customers.
