Application of bayesian additive regression trees in the development of credit scoring models in Brazil

Daniel Alves De Brito Filho,Rinaldo Artes

doi:10.1590/0103-6513.20170110

Daniel Alves De Brito Filho, Rinaldo Artes

Open Access

https://doi.org/10.1590/0103-6513.20170110

Copy DOI

Journal: Produção	Publication Date: Jul 2, 2018
Citations: 3	License type: cc-by

Affiliation: Insper

Abstract

Abstract Paper aims This paper presents a comparison of the performances of the Bayesian additive regression trees (BART), Random Forest (RF) and the logistic regression model (LRM) for the development of credit scoring models. Originality It is not usual the use of BART methodology for the analysis of credit scoring data. The database was provided by Serasa-Experian with information regarding direct retail consumer credit operations. The use of credit bureau variables is not usual in academic papers. Research method Several models were adjusted and their performances were compared by using regular methods. Main findings The analysis confirms the superiority of the BART model over the LRM for the analyzed data. RF was superior to LRM only for the balanced sample. The best-adjusted BART model was superior to RF. Implications for theory and practice The paper suggests that the use of BART or RF may bring better results for credit scoring modelling.

Highlights

Credit analysis is a key activity for retail banks
Based on the results of the hypothesis tests obtained for the balanced models, we can say that the Bayesian additive regression trees (BART) model was superior to the logistic regression model ( p < 0.01)
This paper empirically evaluated the performance of two machine learning models, BART and random forest, applied to credit scoring for predicting a “good” or “bad” payer

Summary

Introduction

Credit analysis is a key activity for retail banks. Credit scoring models have become an important tool in credit analysis due to the need for standardization and agility in decision making, and there are situations in which credit approval or refusal is fully automated. Since 2004, with the Basel II agreement (Bank for International Settlements, 2004, 2006), banks have been encouraged to improve their internal credit risk models to obtain the authorization to use them as a basis for capital allocation adjusted to that risk. In order to obtain the approval for the use of advanced credit scoring models, known as the Advanced Internal Rating Approach (A-IRB), banks need to demonstrate their ability to accurately assess their risk. Banks with an A-IRB certification have competitive advantages over other banks because they are authorized by regulators to allocate less capital to credit risk

Objectives

Methods

Findings

Conclusion