Machine learning-driven credit risk: a systemic review

Abstract

Credit risk assessment is at the core of modern economies. Traditionally, it is measured by statistical methods and manual auditing. Recent advances in financial artificial intelligence stemmed from a new wave of machine learning (ML)-driven credit risk models that gained tremendous attention from both industry and academia. In this paper, we systematically review a series of major research contributions (76 papers) over the past eight years using statistical, machine learning and deep learning techniques to address the problems of credit risk. Specifically, we propose a novel classification methodology for ML-driven credit risk algorithms and their performance ranking using public datasets. We further discuss the challenges including data imbalance, dataset inconsistency, model transparency, and inadequate utilization of deep learning models. The results of our review show that: 1) most deep learning models outperform classic machine learning and statistical algorithms in credit risk estimation, and 2) ensemble methods provide higher accuracy compared with single models. Finally, we present summary tables in terms of datasets and proposed models.

Credit Risk Management of a Commercial Bank: Problems and Solutions (On the Example of JSC “Halyk Bank”)

“…For example, the American Bank Wells Fargo uses machine learning to determine the risk of fraud, and the Chinese bank "Ping An" uses neural networks to predict credit risk (Lei, 2021). However, it is necessary to take into account some risks associated with the use of AI, such as the need for high data accuracy, system reliability and data security (Shi et al, 2022). In addition, it is necessary to ensure a balance between the use of AI and the human factor in order to maintain a humanitarian approach to credit risk management.…”

Predicting customer deposits with machine learning algorithms: evidence from Tunisia

“…A large group of literature focusing on applying machine learning in the banking industry. Bank insolvencies (Le and Viviani, 2018;Petropoulos et al, 2020;Sen and Figueiredo, 2021;Kristof and Virag, 2022), bank performance (Abu Bakar and Tahir, 2009;Kablay and Gumbo, 2021;Ainan and Nur-E-Arefin, 2022) and bank credit worthiness (Turkson et al, 2016;Kumar et al, 2021;Shi et al, 2022;Sigrist and Leuenberger, 2023) are the most important issues that have been treated in this context. A few studies have dealt with customer deposits prediction, and Pangrahi and Patnaik (2020) tried to forecast customer behavior from a bank direct marketing survey using neural network techniques.…”

Evaluating Various Machine Learning Techniques in Credit Risk Area

“…With the impressive evolution of economy and computer science, data related to financial and credit risks are increasingly collected and stored in digital. Many financial institutions are trying to use different machine learning algorithms to gradually establish and improve their own credit risk prediction system [1]. There is no doubt that a sound credit prediction system can help society, enterprises, and individuals develop more rapidly and safely.…”