A PSYCHOLOGICAL APPROACH TO MICROFINANCE CREDIT SCORING VIA A CLASSIFICATION AND REGRESSION TREE

• DOI: http://doi.org/10.1002/isaf.1355
• Published date: 01 October 2014
• Date: 01 October 2014
• Author: Ibtissem Baklouti

A PSYCHOLOGICAL APPROACH TO MICROFINANCE CREDIT

SCORING VIA A CLASSIFICATION AND REGRESSION TREE

IBTISSEM BAKLOUTI*

Faculty of Economics and Management of Sfax, ResearchUnit: Corporate Finance and Financial Theory (COFFIT),

Sfax, Tunisia

SUMMARY

Microfinance institutions’(MFIs’) peculiar lending methodology is characterized by an unchallenged decision-

making predominance from the part of loan officers. Indeed, the latter are in charge of providing a great deal of

diagnostic information regarding the entrepreneur’s psychological traits likely to help them run a business. This

paper constitutes an initial attempt towards exploring the role of borrowers’psychologicaltraits in predicting future

default occurrences. It builds on a nonparametric credit scoring model, based on a decision tree, including

borrowers’quantitative behavioural traits as input for the final scoring model. On applying data collected from a

Tunisian microfinance bank, the major depicted result lies in the fact that borrowers’psychological traits constitute

a major information source in predicting their creditworthiness. Actually, the variables deployed have helped

reduce the proportion of bad loans classified as good loans by 3.125%, which leads to a decrease in MFIs’losses

by 4.8%. In addition, the results indicate that the scoring model based on a classification and regression tree

(CART) outperforms the classic techniques. Actually, implementing this CART model might well help MFIs

reduce misclassification costs by 6.8% and 13.5% in comparison with the discriminant analysis and logistic

regression models respectively. Our conceived model, we consider, would be of great practical implication for

microfinance and may provide a means for securing competitive advantage over other MFIs that fail to implement

such a methodology. Copyright © 2014 John Wiley & Sons, Ltd.

Keywords: microfinance institutions; credit scoring; psychological traits; data mining

1. INTRODUCTION

Microfinance institutions (MFIs) focus on providing credits to the poor as well as disadvantaged and

marginalized entrepreneurs who have no access to traditional commercial banks. Such credits have been

oriented for entrepreneurial activity purposes as well as other ends. In this regard, microfinance has been

widely recognized as one of the most promising tools designed to alleviate povertyworldwide, and above

all in the developing countries. In addition to their social outreach objective, MFIs are faced with the

problem of maintaining their self-sustainability. Indeed, further to preserving their self-sufficiency, such

institutions have to cover all their actual costs and make profits on the services they offer.

However, owing to high demand pressure, over-indebtedness and economic crises striking several

microfinance markets, MFIs are under an obligation to pursue their social and financial goals in increas-

ingly constrained environments. Hence, for the purpose of persisting and improving both social out-

reach and financial sustainability, many MFIs have come to conclude that developing a reliably

powerful credit risk management tool has become urgently imposed (Caudill, Gropper, & Hartarska,

2009). Therefore, lenders have begun to consider devising a special mechanism to ensure that loan

* Correspondence to: Ibtissem Baklouti, Road of Airport, Km 4 Sfax, 3018, Tunisia. E-mail: bakloutiibtissem@yahoo.fr

Copyright © 2014 John Wiley & Sons, Ltd.

INTELLIGENT SYSTEMS IN ACCOUNTING, FINANCE AND MANAGEMENT

Intell. Sys. Acc. Fin. Mgmt. 21, 193–208 (2014)

Published online 11 June 2014 in Wiley Online Library (wileyonlinelibrary.com) DOI: 10.1002/isaf.1355

repayments are as optimum as possible to preserve stability and avoid charging higher interest rates

likely to defeat the very purpose of microfinance lending.

In this context, several researchers highlight the necessity of implementing the credit scoring

approach by MFIs, as has been successfully adopted by the traditional financial organizations (Viganò,

1993; Vogelgesang, 2003; Schreiner, 2004; Van Gool, Verbeke, Sercu, & Baesens, 2011). Indeed, on

evaluating the credit-scoring efficiency level, the authors have ended up by concluding that a credit-

scoring model could well stand as an appropriate alternative likely to increase operational efficiency

and reduce costs through a fully automated loan-granting process. Similarly, they have advanced that

credit scoring might certainly reduce loan officers’discrimination and prejudice effects. Such a proce-

dure, they think, would certainly help regain the opportunity cost likely to be missed should eligibly

good credit applicants be rejected and avoid would-be missed interest and principle payments likely

to occur should bad credit applicants be accepted. Table I gives an overview of published statistical

credit-scoring models relevant to MFIs. These different studies have found that credit-scoring models

enjoy enough predictive power to enable them to significantly improve evaluating the risk that

microcredit applicants can incur.

Noteworthy,however,is that some authors believe it is difficultto incorporate and adapt this approachas

an integralpart of the credit decision process inthe context of microfinance, for ignoringqualitative factors

might well lead to inaccurate evaluations.These authors (Schreiner, 2004; Bumacov & Ashta, 2011; Van

Gool et al., 2011;Baklouti & Bouri, 2013) have reached the conclusion thatcredit-scoring techniques can

be incorporated intothe microfinance area only as acomplement, rather than a substitute,to the subjective

judgmental approach. They haveadvanced that the loan officer’sexclusion from the microfinance activity

seems so far an impossible undertaking, since a high share of risk in microcredit is usually linked with

qualitative information pertaining to the entrepreneur’s personality and ability to run a business. In this

respect, a few researchers (Cornée, 2009; Baklouti & Bouri, 2014) have responded by developing a

credit-scoring model using both hard and soft information. Soft information denotes the loan officer’sin-

tuitions, impressions or feelings regarding an entrepreneur’s management capacity. The empirical results

attained by the studies mentioned have shown that the mixed model involving both types of facts has a

higher forecast quality than one including exclusively hard information. Nevertheless, it should be

highlighted that, for this conclusion to be consolidated, other factors need to be studied; for instance,

the cost of training and maintaining experienced and qualified credit evaluators and the cost of a slower

approval process. Indeed, collecting soft information wouldtypically involve a face-to-face meeting with

each applicant, during which the loan officer shapes and constructs their judgment about the applicant’s

creditworthiness after interviewing them. In addition, owing to human bias, a loan officer’ssubjective

judgment doesnot necessarily reach accurate conclusions about the potentialborrower’s creditworthiness.

Table I. Overview of published credit scoring models for MFI

Author Country Sample size Technique(s)

Performance

of model

(PCC

total

) (%)

Viganò (1993) Burkina Faso 100 observations Multivariate discriminant analysis 77

Schreiner (2004) Bolivia 39,956 observations Logistic regression 91

Vogelgesang(2003) Bolivia 8002 observations Bivariate probit model 84

Van Gool et al. (2011) Bosnia 6722 observations Logistic regression 78.48

Blanco, Pino-Meijas, Juan,

and Salvador (2013)

Peru 5500 observations Multilayer perceptron neural

networks

88.33

Baklouti and Bouri (2013) Tunisia 5022 observations Logistic regression 74.6

194 I. BAKLOUTI

Copyright © 2014 John Wiley & Sons, Ltd. Intell. Sys. Acc. Fin. Mgmt., 21, 193–208 (2014)

DOI: 10.1002/isaf