A Machine Learning Model for Accurate Credit Risk Forecasting in Banking Systems: An Empirical Investigation

A Machine Learning Model for Accurate Credit Risk Forecasting in Banking Systems: An Empirical Investigation

Credit risk consists is the expectation of losses stemming from the inability of a borrower to repay a loan. For the purpose of accurate control of credit risks, banking systems seek developing financial information portfolios upon their customers using sophisticated models which are not only restri...

Full description

Saved in:
Bibliographic Details
Published in:International journal of supply and operations management Vol. 11; no. 2; pp. 168 - 187
Main Authors: Hasni, Marwa, Aguir, Mohamed Salah, Babai, Mohamed Zied, Lemai, Zied
Format: Journal Article
Language:English
Published: Tehran Kharazmi University 01-05-2024
Subjects:
Banking
Credit risk
Customers
Deep learning
Forecasting
Machine learning
Neural networks
Risk management
Statistical analysis
Time series
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Credit risk consists is the expectation of losses stemming from the inability of a borrower to repay a loan. For the purpose of accurate control of credit risks, banking systems seek developing financial information portfolios upon their customers using sophisticated models which are not only restricted to collecting information on borrower's characteristics, but also, provide visibility on their respective default risk. This paper introduces a novel deep learning model to forecast the credit risk of company customers in banking systems. In particular, we develop a hybrid SVM-LSTM based neural network that predicts the total turnover of a company given the historical data records of its economic and financial features within specific periods. Through an empirical investigation based on data of 13 Tunisian manufacturing and service companies, we show that our proposed model results in more accurate statistical performances compared to the standard LSTM and to the linear regression that is commonly used in the area of credit risk management.
ISSN:2383-1359
2383-2525
DOI:10.22034/IJSOM.2023.109898.2722