Ten-year evolution on credit risk research: a systematic literature review approach and discussion

Fernanda Medeiros Assef,Maria Teresinha Arns Steiner

doi:10.15446/ing.investig.v40n2.78649

Fernanda Medeiros Assef, Maria Teresinha Arns Steiner

Open Access

https://doi.org/10.15446/ing.investig.v40n2.78649

Copy DOI

Abstract

Given its importance in financial risk management, credit risk analysis, since its introduction in 1950, has been a major influence both in academic research and in practical situations. In this work, a systematic literature review is proposed which considers both “Credit Risk” and “Credit risk” as search parameters to answer two main research questions: are machine learning techniques being effectively applied in research about credit risk evaluation? Furthermore, which of these quantitative techniques have been mostly applied over the last ten years of research? Different steps were followed to select the papers for the analysis, as well as the exclusion criteria, in order to verify only papers with Machine Learning approaches. Among the results, it was found that machine learning is being extensively applied in Credit Risk Assessment, where applications of Artificial Intelligence (AI) were mostly found, more specifically Artificial Neural Networks (ANN). After the explanation of each answer, a discussion of the results is presented.

Highlights

Credit risk analysis is an active research area in financial risk management, and credit scoring is one of the key analytical techniques in credit risk evaluation (Yu, Wang, and Lai, 2009; Steiner, Nievola, Soma, Shimizu, and Steiner Neto, 2007)
Having built several non-parametric credit risk models based on Multilayer Perceptron (MLP) and benchmarks of their performance against other models which employ the traditional Linear Discriminant Analysis (LDA), Quadratic Discriminant Analysis (QDA) and Logistic Regression (LR) techniques, based on a sample of almost 5500 borrowers from a Peruvian microfinance institution, the results presented in Blanco et al (2013) showed that NN (Neural Networks) models outperform the other three classic techniques both in terms of area under the receiver-operating characteristic curve (AUC) and as misclassification costs
This paper extended studies in two main ways: firstly, it proposed a method involving machine learning to solve the reject inference problem; secondly, the Semi-Supervised Support Vector Machines (SSVM) model was found to improve the performance of scoring models compared to the industrial benchmark of LR

Summary

Introduction

Credit risk analysis is an active research area in financial risk management, and credit scoring is one of the key analytical techniques in credit risk evaluation (Yu, Wang, and Lai, 2009; Steiner, Nievola, Soma, Shimizu, and Steiner Neto, 2007). Credit risk evaluation is a data mining research problem, both challenging and important in the field of financial analysis. This assessment is used in predicting whether or not there is a possibility for credit concession. According to the work of Zhang, Gao, and Shi (2014), there is a wide range of methodologies for solving credit risk classification problems. These methods include mainly logistic regression, probit regression, nearest neighbor analysis, Bayesian networks, Artificial Neural Networks (ANN), decision trees, genetic algorithms (GA), multiple criteria decision making (MCDM), support vector machines (SVM), How to cite: Assef, F.

Objectives

Methods

Findings

Conclusion