Review of Business and Economics Studies, 2016, том 4, № 4

Review of 
Business and
Economics 
Studies

EDITOR-IN-CHIEF
Prof. Alexander Ilyinsky
Dean, International Finance Faculty, 
Financial University, Moscow, Russia
ailyinsky@fa.ru 

EXECUTIVE EDITOR
Dr. Zbigniew Mierzwa

EDITORIAL BOARD

Dr. Mark Aleksanyan
Adam Smith Business School, 
The Business School, University 
of Glasgow, UK

Prof. Edoardo Croci
Research Director, IEFE Centre for 
Research on Energy and Environmental 
Economics and Policy, Università 
Bocconi, Italy

Prof. Moorad Choudhry
Dept.of Mathematical Sciences, Brunel 
University, UK

Prof. David Dickinson 
Department of Economics, Birmingham 
Business School, University of 
Birmingham, UK

Prof. Chien-Te Fan
Institute of Law for Science and 
Technology, National Tsing Hua 
University, Taiwan

Prof. Wing M. Fok
Director, Asia Business Studies, College 
of Business, Loyola University New 
Orleans, USA

Prof. Konstantin P. Gluschenko
Faculty of Economics, Novosibirsk State 
University, Russia

Prof. George E. Halkos
Associate Editor in Environment and 
Development Economics, Cambridge 
University Press; Director of Operations 
Research Laboratory, University of 
Thessaly, Greece

Dr. Christopher A. Hartwell
President, CASE — Center for Social and 
Economic Research, Warsaw, Poland

Prof. S. Jaimungal
Associate Chair of Graduate 
Studies, Dept. Statistical Sciences 
& Mathematical Finance Program, 
University of Toronto, Canada

Prof. Bartlomiej Kaminski
University of Maryland, USA; 

Rzeszow University of Information 
Technology and Management, 
Poland

Prof. Vladimir Kvint 
Chair of Financial Strategy, Moscow 
School of Economics, Moscow State 
University, Russia

Prof. Alexander Melnikov 
Department of Mathematical and 
Statistical Sciences, University of 
Alberta, Canada

Prof. George Kleiner
Deputy Director, Central Economics and 
Mathematics Institute, Russian Academy 
of Sciences, Russia

Prof. Kwok Kwong
Director, Asian Pacifi c Business 
Institute, California State University, 
Los Angeles, USA

Prof. Dimitrios Mavrakis
Director, Energy Policy and 
Development Centre, National and 
Kapodistrian University of Athens, 
Greece

Prof. Steve McGuire
Director, Entrepreneurship Institute, 
California State University, 
Los Angeles, USA

Prof. Rustem Nureev
Head of the Department of Economic 
Theory, Financial University, 
Russia

Dr. Oleg V. Pavlov
Associate Professor of Economics and 
System Dynamics, Department of Social 
Science and Policy Studies, Worcester 
Polytechnic Institute, USA

Prof. Boris Porfi riev
Deputy Director, Institute of Economic 
Forecasting, Russian Academy of 
Sciences, Russia

Prof. Svetlozar T. Rachev
Professor of Finance, College of 
Business, Stony Brook University, USA

Prof. Boris Rubtsov
Deputy chairman of Department 
of fi nancial markets and banks for R&D, 
Financial University, Russia

Dr. Minghao Shen
Dean, Center for Cantonese Merchants 
Research, Guangdong University of 
Foreign Studies, China

Prof. Dmitry Sorokin
Chairman for Research, Financial 
University, Russia

Prof. Robert L. Tang
Vice Chancellor for Academic, De La 
Salle College of Saint Benilde, Manila, 
The Philippines

Dr. Dimitrios Tsomocos 
Saïd Business School, Fellow in 
Management, University of Oxford; 
Senior Research Associate, Financial 
Markets Group, London School 
of Economics, UK

Prof. Sun Xiaoqin
Dean, Graduate School of Business, 
Guangdong University of Foreign 
Studies, China

REVIEW OF BUSINESS 
AND ECONOMICS STUDIES 
(ROBES) is the quarterly peerreviewed scholarly journal published 
by the Financial University under 
the Government of Russian 
Federation, Moscow. Journal’s 
mission is to provide scientifi c 
perspective on wide range of topical 
economic and business subjects.

CONTACT INFORMATION
Financial University
Leningradsky prospekt, 53, 
offi ce 5.4
123995 Moscow
Russian Federation
Telephone: +7 (499) 943-94-53
Website: www.robes.fa.ru

AUTHOR INQUIRIES
Inquiries relating to the 
submission of articles can be sent 
by electronic mail to robes@fa.ru.

COPYRIGHT AND PHOTOCOPYING 
© 2016 Review of Business and 
Economics Studies. All rights 
reserved. No part of this publication 
may be reproduced, stored 
or transmitted in any form or by any 
means without the prior permission 
in writing from the copyright holder. 
Single photocopies of articles may 
be made for personal use as allowed 
by national copyright laws. 
ISSN 2308-944X

Вестник
исследований
бизнеса 
и экономики

ГЛАВНЫЙ РЕДАКТОР
А.И. Ильинский, профессор, декан 
Международного финансо вого факультета Финансового университета 

ВЫПУСКАЮЩИЙ РЕДАКТОР
Збигнев Межва, д-р экон. наук

РЕДАКЦИОННЫЙ СОВЕТ

М.М. Алексанян, профессор Бизнесшколы им. Адама Смита, Университет 
Глазго (Великобритания)

К. Вонг, профессор, директор Института азиатско-тихоокеанского бизнеса 
Университета штата Калифорния, 
Лос-Анджелес (США)

К.П. Глущенко, профессор экономического факультета Новосибирского 
госуниверситета

С. Джеимангал, профессор Департамента статистики и математических финансов Университета Торонто 
(Канада)

Д. Дикинсон, профессор Департамента экономики Бирмингемской бизнесшколы, Бирмингемский университет 
(Великобритания)

Б. Каминский, профессор, 
Мэрилендский университет (США); 
Университет информационных 
технологий и менеджмента в Жешуве 
(Польша)

В.Л. Квинт, заведующий кафедрой 
финансовой стратегии Московской 
школы экономики МГУ, профессор 
Школы бизнеса Лассальского университета (США)

Г. Б. Клейнер, профессор, член-корреспондент РАН, заместитель директора Центрального экономико-математического института РАН

Э. Крочи, профессор, директор по 
научной работе Центра исследований 
в области энергетики и экономики 
окружающей среды Университета 
Боккони (Италия)

Д. Мавракис, профессор, 
директор Центра политики 
и развития энергетики 
Национального университета 
Афин (Греция)

С. Макгвайр, профессор, директор Института предпринимательства 
Университета штата Калифорния, 
Лос-Анджелес (США)

А. Мельников, профессор 
Депар та мента математических 
и ста тистических исследований 
Университета провинции Альберта 
(Канада)

Р.М. Нуреев, профессор, руководитель Департамента экономической 
теории Финансового университета

О.В. Павлов, профессор 
Депар та мента по литологии 
и полити ческих исследований 
Ворчестерского политехнического 
института (США) 

Б.Н. Порфирьев, профессор, 
член-корреспондент РАН, заместитель директора Института 
народнохозяйственного прогнозирования РАН

С. Рачев, профессор Бизнес-колледжа Университета Стони Брук 
(США) 

Б.Б. Рубцов, профессор, заместитель 
руководителя Департамента финансовых рынков и банков по НИР 
Финансового университета

Д.Е. Сорокин, профессор, членкорреспондент РАН, научный 
руководитель Финансового 
университета

Р. Тан, профессор, проректор 
Колледжа Де Ла Саль Св. Бенильды 
(Филиппины) 

Д. Тсомокос, Оксфордский университет, старший научный сотрудник 
Лондонской школы экономики 
(Великобритания)

Ч.Т. Фан, профессор, Институт 
права в области науки и технологии, 
национальный университет Цин Хуа 
(Тайвань)

В. Фок, профессор, директор по 
исследованиям азиатского бизнеса Бизнес-колледжа Университета 
Лойола (США)

Д.Е. Халкос, профессор, Университет 
Фессалии (Греция)

К.А. Хартвелл, президент Центра 
социальных и экономических исследований CASE (Польша)

М. Чудри, профессор, Университет 
Брунеля (Великобритания)

Сун Цяокин, профессор, декан Высшей школы бизнеса Гуандунского 
университета зарубежных исследований (КНР)

М. Шен, декан Центра кантонских 
рыночных исследований Гуандунского университета (КНР)

Издательство Финансового 
университета
123995, Москва, ГСП-5, 
Ленинградский пр-т, 53, 
комн. 5.4
Тел. 8 (499) 943-94-53.
Интернет: www.robes.fa.ru.

Журнал “Review of Business and 
Economics Studies” («Вест ник 
исследований бизнеса и экономики») зарегистрирован в Федеральной службе по надзору в 
сфере связи, информационных 
технологий и массовых коммуникаций 15 сентября 2016 г. 
Свидетельство о регистрации 
ПИ № ФС77-67072. 

Подписано в печать: 29.12.2016.
Формат 60 × 84 1/8.
Заказ № 1430 от 28.12.2016. 
Отпечатано в ООП Издательства 
Финуниверситета 
(Ленинградский проспект, д. 49).
16+

CONTENTS

Applying Discriminant Model to Manage Credit Risk 

for Consumer Loans in Vietnamese Commercial Bank

Nguyen Thuy Duong, Do Thi Thu Ha, Nguyen Bich Ngoc . . . . . . . . . . . . . . . . . . . . . . . . .5

Lean Construction and BIM: Complementing 

Each Other for Better Project Management

Eroshkin S.Y., Kallaur G.Y., Papikian L.M.  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .17

Managing the Demographic Risk of Pension Systems

Kowalczyk-Rólczyńska Patrycja . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .23

Restructuring the Banking System: The Case of Vietnam

To Thuy Duong . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .32

Review of 
Business and 
Economics 
Studies

Volume 4, Number 4, 2016

Вестник
исследований
бизнеса 
и экономики

№ 4, 2016

CОДЕРЖАНИЕ

Применение дискриминационной модели 

в управлении риском потребительских кредитов

в коммерческом банке Вьетнама

Нгуен Тху Дуонг, До Тхи Тху Ха, Нгуен Бих Нгок  . . . . . . . . . . . . . . . . . . . . . . . . . . . . .5

Бережливое строительство и BIM-модель: 

дополняя друг друга для более 

эффективного управления проектами

Ерошкин С.Ю., Каллаур Г.Ю., Папикян Л.М. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .17

Управление демографическим риском в пенсионных системах

Ковальчик-Рульчинская Патриция . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .23

Реструктуризация банковской системы: пример Вьетнама

To Тху Дуонг  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .32

Review of Business and Economics Studies  
 
Volume 4, Number 4, 2016

Applying Discriminant Model 
to Manage Credit Risk for Consumer Loans 
in Vietnamese Commercial Bank

Nguyen Thuy Duong,
PhD, Banking Faculty, Banking Academy of Vietnam
ngocnb@hvnh.edu.vn

Do Thi Thu Ha,
MA, Banking Faculty, Banking Academy of Vietnam
ngocnb@hvnh.edu.vn

Nguyen Bich Ngoc,
MA, Banking Faculty, Banking Academy of Vietnam
ngocnb@hvnh.edu.vn

Abstract. This study estimates a two-group discriminant function to determine the expected 
fi nancial health of the consumer credit customers’ of a bank of Vietnam by using fi ve demographic, 
socio-economic, and loan characteristics of the sample borrowers. The estimated function is 
signifi cant at one per cent level of signifi cance and the model estimates fi nancial health/group 
membership with average seventy-three per cent accuracy. Like developed countries, it is expected 
that use of the estimated discriminant function in the consumer credit decision making will 
decrease bad debts, will help to set risk based credit pricing for the clients and will make the credit 
granting faster and more accurate.

Keywords: consumer credit; fi nancial distress; prediction; demographic and socio-economic 
characteristics; two-group discriminant analysis.

Применение дискриминационной модели 
в управлении риском потребительских кредитов 
в коммерческом банке Вьетнама

Нгуен Тху Дуонг,
д-р экон. наук, Банковский факультет, Банковская академия Вьетнама, Ханой, Вьетнам
ngocnb@hvnh.edu.vn

До Тхи Тху Ха,
магистр, Банковский факультет, Банковская академия Вьетнама, Ханой, Вьетнам
ngocnb@hvnh.edu.vn

Нгуен Бих Нгок,
магистр, Банковский факультет, Банковская академия Вьетнама, Ханой, Вьетнам
ngocnb@hvnh.edu.vn

Review of Business and Economics Studies  
 
Volume 4, Number 4, 2016

1. INTRODUCTION
The idea of consumer credit is extensive. In 
general, consumer credit is the term stands for 
the express loan facilities to the common people that have to repay with interest by equal 
monthly installment and the credit is not used 
for any commercial purpose. The need of consumer credit today is at its highest, but at the 
same time the default rates have risen and from 
the banks’ perspective the riskiness of these 
loans is usually higher than granted loans they 
analyzed defaulted. For the lending institution 
such a default rate affects to its fi nancial performance significantly. So, it is substantially 
better to use discriminant analysis to determine the expected position or a score for the 
borrower to make the credit grant decision. In 
other words, a quantitative effort is made to 
forecast the expected position of the consumer 
credit applicant via the discriminant analysis. 
In the current paper, we use the discriminant 
analysis to develop predictive models allowing 
distinguishing between “good” and “bad” borrowers. The data have been collected from commercial Vietnamese banks over a 3-year period, 
from 2014 to 2016.
The discriminant analysis is look like the 
regression analysis in terms of the number of 
dependent variables (one for both), the number of independent variables (multiple for both) 
and the nature of independent variables (metric for both). But, the discriminant analysis and 
the regression analysis are different in terms 
of the nature of dependent variables. In the 
regression analysis, the dependent variable is 

a metric variable whereas in the discriminant 
analysis, the dependent variable is a categorical/binary variable. Besides, the nature of the 
dependent variable in the binary logit model 
and the two-group discriminant analysis is the 
same. The linear discriminant analysis model 
involves linear combinations of the equation 1 
form:

     Z =  + 1X1 + 2X2 + 3X3 +… + kXk. (1)

In the model, Z = discriminant score,  = constant, ’s = discriminant coeffi cient or weight, 
X’s = predictor or independent variable. The 
coeffi cients of the independent variables are estimated such that the scores differ for the two 
groups substantially. This happens when the 
ratio- between-group sum of squares to withingroup sum of squares is at maximum point. For 
any other combination, the ratio will be smaller. 
The Figure 1 shows the pictorial presentation 
of the data collected on the two variables: X1 
and X2 for the cases of the two-group G1 and 
G2. The X1 axis represents X1 variable and the 
X2 axis represents X2 variable. The discriminant analysis tries to separate the two groups by 
drawing a line as under. If the data is collected 
on more than two variables, then it is not possible to draw a scatter diagram as under as we 
have fi xed two axes in a graph. But regardless of 
the number of variables, the discriminant analysis can generate positive and negative Z scores 
for the cases of the groups and possible to draw 
a diagram as a lower part of the Figure 1. The 
lower part represents the group membership by 

В данной работе с помощью бинарной дискриминационной функции проведена оценка 
ожидаемого финансового «здоровья» пользователей потребительских кредитов, предоставляемых 
банком Вьетнама, используя пять демографических, социально-экономических видов займов 
характеристик пробы заемщиков. Оцениваемая дискриминационная функция оказалась 
достоверной при 1 %-ном уровне значимости и применении модели оценки финансового 
«здоровья» потребителей выбранной группы потребителей, что дало результат с 73 %-ной 
достоверностью. В развитых странах предполагается, что применение оценки с помощью 
дискриминационной функции при принятии решения в области потребительского кредита будет 
способствовать снижению числа плохих долгов, а также даст возможность устанавливать оценку 
платежеспособности с учетом риска. Это поможет ускорить оформление кредита и поднять уровень 
его обеспеченности.

Ключевые слова: потребительский кредит; финансовое неблагополучие; демографические 
и социально-экономические характеристики; бинарный дискриминационный анализ.

Review of Business and Economics Studies  
 
Volume 4, Number 4, 2016

using the estimated discriminant scores (Z) of 
the groups cases. The shaded proportion represents the misclassification of the group membership. The smaller the shaded proportion, 
the bigger the estimation accuracy is assumed 
(Malhotra & Das, 2011; Boyd, Westfall, & Stasch, 
2005)
The objectives are divided into two-broad 
objective and specific objectives. The broad 
objective of the study is to determine the consumer credit customers’ insolvency by using 
demographic & socio-economic characteristics 
and two-group discriminant analysis. In consistent with the broad objective, the specific 
objectives are as follows: (i) To develop discriminant function or linear combinations of the 
predictor, or independent variables, which will 
best discriminate between the categories of the 
criterion or dependent variable. (ii) To examine 
whether signifi cant differences exist among the 
groups ‘in term of the predictor variables’. (ii) 
To determine which predictor variables contribute to most of the inter group differences. 
(iii) To classify cases to one of the groups based 
on the values of the predictor variables. (iv) To 
evaluate the accuracy of the classifi cation. The 
fi rst section of this research report is about introduction to the study which comprises prologue, objectives and methodology of the study. 
The second section contains literature review 
and the variables selection for the study. Empirical study in Vietnam’s commercial banks, 
fi ndings and their analysis are in the third section of the report.

2. LITERATURE REVIEW

2.1. Statistical methods for credit risk 
prediction

In the past, many researchers have developed a 
variety of traditional statistical methods for corporate credit risk prediction, with utilization of 
Linear discriminant analysis (LDA) and Logistic 
regression analysis (LRA) being the two most 
commonly used statistical methods in building 
corporate credit risk prediction models. Possibly the earliest use of applying LDA to corporate credit risk prediction is the work by Durand 
(1941). However, Karels and Prakash (1987) and 
Reichert et al. (1983) pointed that the application of LDA has often been challenged owing to 
its assumption of the categorical nature of the 
corporate credit data and the fact that the covariance matrices of the credit risk and non-risk 
classes are unlikely to be equal. In addition to 
the LDA approach, LRA is another commonly 
used alternative to conduct corporate credit 
risk prediction tasks. Thomas (2000) and West 
(2000) indicated that both LDA and LRA are intended for the case when the under-lying relationship between variables are linear and hence 
are reported to be lacking in sufficient prediction accuracy. Besides above two statistical 
methods, Friedman (1991) reported that Multivariate adaptive regression splines (MARS) is 
another commonly corporate credit risk prediction method. However, the problem with applying these statistical methods to corporate credit 
risk prediction is that some assumptions, such 

Figure 1. Discriminant Analysis

Review of Business and Economics Studies  
 
Volume 4, Number 4, 2016

the multivariate normality assumptions for independent variables, are frequently violated in 
reality, which makes these methods theoretically 
invalid for finite samples.
Although these methods are relatively simple 
and explainable, the ability to discriminate credit non-risk customers from credit risk ones is 
still an argumentative problem. In recent years, 
many studies have demonstrated that Artificial 
intelligence (AI) methods, such as Artificial neural network (ANN) (West, 2000), Decision tree 
(DT) (Jiang, 2009), case based reasoning (CBR) 
(Shin & Han, 2001) and Support vector machine 
(SVM) (Schebesch & Stecking, 2005) can be used 
as alternative methods for corporate credit risk 
prediction. In contrast with statistical methods, 
AI methods do not assume certain data distributions. These methods automatically extract 
knowledge from training samples. According to 
previous studies, AI methods are superior to statistical methods in dealing with corporate credit 
risk prediction problems, especially for nonlinear pattern classification (Huang et al., 2004; 
West, 2000).

2.2. Discriminant Analysis 
for consumer credit

Wiginton (1980) conducted a discriminant 
analysis to model the consumer credit behavior 
by using demographic and economic variables. 
The demographic variables used are: number 
of dependents, living status, moved during last 
year, business use of vehicle and pleasure use 
of vehicle. The economic variables include industry class of employment, class of occupation 
and years in present employment. The right 
prediction power of the model estimated by the 
researcher is not encouraging and predicting 
group membership by using logit model provided better forecasting accuracy. It is concluded 
that years in present employment, living status 
and occupation type are significantly related 
to the credit risk rating. Grablowsky (1975) 
conducted a two-group stepwise discriminant 
analysis in order to model risk in the consumer 
credit by using behavioral, fi nancial, and demographic variables. The behavioral data is collected from the two hundred borrowers through 
a questionnaire of summated ratings scale and 
the fi nancial and demographic data are collected from the loan application forms of the same 

two hundred borrowers. The researcher has 
started analysis with thirty six variables and after a comprehensive sensitivity analysis, found 
that thirteen variables are enough to model 
the consumer credit risk. Although the both 
set of data- analysis sample and holdout sample violated the equal variance-covariance assumptions, the estimated model classifi ed the 
validation sample 94 per cent correctly. Awh & 
Waters (1974) conducted a study to determine 
the bank’s active and inactive credit card holders by using two types of variables-quantitative 
(economic and demographic) and attitudinal. 
The quantitative variables used are: (a) income, 
(b) age, (c) education, and (d) socio-economic 
standing. The socio-economic index is based on 
the respondents’ particular position suggested 
by Reiss (1961). The attitudinal variables used 
are: (a) use or non-use of other credit cards, (b) 
attitude toward credit, and (c) attitude toward 
bank charge-cards. The data for the quantitative and attitudinal variables on the same respondent is collected from the loan application 
forms and by the questionnaires respectively. 
The discriminant function estimated by them 
is significant at 0.01 level and forecasted the 
group membership with 78 per cent accuracy. 
Hand & Henley (1997) reviewed available credit 
scoring techniques in their article titled — “Statistical Classification Methods in Consumer 
Credit Scoring: A Review.” In addition to the 
judgmental method, the available quantitative 
methods are logistic regression, mathematical 
programming, discriminant analysis, regression, recursive partitioning, expert systems, 
neural networks, smoothing nonparametric 
methods, and time varying models. They have 
concluded that there is no best method. What is 
the best method depends on the structure and 
characteristics of the data. For a data set, one 
method may be better than the other method 
but for another data set, the other method may 
be better.
In addition, Davis, Edelman & Gammerman 
(1992) conducted a comparative study of various methods and concluded that all the methods are performed at the same accuracy level 
but the neural network algorithms take much 
longer time to train. According to Hand & Henley (1997), characteristics typical to differentiate the problematic and regular customer are: 

Review of Business and Economics Studies  
 
Volume 4, Number 4, 2016

time at present address, home status, post code, 
telephone, applicant’s annual income, credit 
card, types of bank account, age, country code 
judgment, types of occupation, purpose of loan, 
marital status, time with bank and time with 
employers, etc. The partial list of characteristics those may be useful to determine the group 
membership given by Capon (1982) includes the 
variables-telephone at home, own/rent living, 
age, time at home address, industry in which 
employed, time with employer, time with previous employer, type of employment, number of 
dependents, types of credit reference, income, 
savings and loan references, trade union membership, age difference between man and wife, 
telephone at work, length of product being purchased, age of automobiles, geographical location, debt to income ratio, monthly installment 
etc. Dinh & Kleimeier (2007) conducted a study 
for the Vietnam’s retail banking market by using logistic regression analysis method. The 
variables they have used are age, education, 
occupation, total time in employment, time in 
current job, residential status, number of dependents, applicants annual income, family income, short-term performance history with the 
bank, long-term performance history with the 
bank, total outstanding loan amount, other services used, cash in hand and at bank, etc. They 
have argued that by using quantitative credit 
scoring, the default rate can be minimized from 
3.3 per cent to 2.0 per cent. They also argued 
that by quantifying the credit risk, it is possible 
to set up risk-based pricing in the retail banking market. Consequently, the bank can become 
more efficient and competitive in the market. 
The most important predictors they found are 
time with bank, followed by gender, number of 
loans, and loan duration. Based on the above literature review, experience of the researcher and 
availability of the data, thirteen demographic 
and socio-economic variables are selected for 
this study. The variables are the loan amount, 
number of dependents, years of experiences at 
present job, salary per month, living status, savings per month, cash in hand and at bank, Net 
worth, ACT, N-EMI, EMI, interest rate (%), and 
Guar. The data is collected on the variables from 
the application forms of the consumer credit 
customers by fi lling up the pre-designed questionnaire.

3. RESEARCH METHODOLOGY

3.1. Research design

To be considered as one of the most broadly techniques used to discriminate between two groups 
(Abdou & Pointon, 2011), discriminant analysis 
has long been used by researchers and bank’s 
managers for building credit scoring models to 
distinguish between customers as good credit 
and bad credit (Abdou & Pointon, 2009; Sarlija et al, 2004; Caouette et al, 1998; Hand et al, 
1998; Hand & Henley, 1997 and Desai et al, 1996). 
Therefore, in this article, discriminant model will 
also be used to distinguish between two loan borrower classifi cation groups: repayment and nonrepayment, in which good borrower is coded as 1 
and bad borrower is coded as 0. This use of two 
groups of customers which are either good or bad 
ones is also considered as one approach for classification purposes in credit scoring models by 
many researchers such as Kim & Sohn, 2004; Lee 
et al, 2002; Banasik et al, 2001; Boyes et al,1989 
and Orgler, 1971. These two possible states are 
defi ned by a number of factors which simultaneously infl uence on borrower’s ability to pay and 
willingness to pay. In case of this study, information related to age, salary, years at present career, 
loan amount and number of independents will be 
used to calculate discriminant score Z for a given 
customer as follows:

Zi =  + *X + *X + *X + 
 
+ *X + *X + . 
(2)

Where:
Z is the discriminant score that maximizes 
the distinction between the two groups:
0: constant.
1-5: slopes of independent variables.
X1: Age
X2: Dependents
X3: YAPJ
X4: Salary
X5: Loan amount
: random error.
As can be seen from the model, there are two 
types of variables in this model, which are dependent and independent variables. The only 
dependent variable is status of borrower that is 
a categorical variable. If a borrower’s position is 
default then he is denoted by 0 and if the bor

Review of Business and Economics Studies  
 
Volume 4, Number 4, 2016

rower’s position is regular, then he is denoted by 
1. By contrast, there are two types of the predictor variables are used in this study. Particularly, 
some variables are related with the loan and the 
others are related with the demographic and 
socio-economic conditions of the borrower. The 
variables related with the demographic and socio-economic conditions of the borrower are as 
follows. Age: How old borrower is; Dependents: 
Dependents mean the number of persons who 
are dependent on the borrower; YAPJ stands for 
years at present job; Salary: how much money 
earned by the borrower per month. The independent variable related with the loan is loan 
amount which indicates how much money borrowed by the borrower.
Secondary data will be used in this study instead of primary data. To explain for this choice, 
advantages of using secondary data will be analyzed. Firstly, using secondary data, which already been available in commercial banks, might 
enables me to save time and money (Ghauri & 
Grn̜haug, 2006). Moreover, Stewart and Kamins 
(1993) indicate when comparing between secondary data and own collected data, the quality 
of former is higher than latter. Finally, secondary data has also been used in many researches 
on credit scoring conducted by researchers not 
only in Vietnam (Duong, Tran & Ho, 2015) but 
also in other countries like Wiginton (1980); 
Elena Bartolozzi, Matthew Cornford, Leticia 
García-Ergüín, Cristina Pascual Deocón, Oscar 
Iván Vasquez & Fransico Javier Plaza (2008) and 
Hörkkö (2010). As a result of that, secondary 
data collected from commercial banks in Vietnam will be used.
Besides, related to sample size, it is said that 
the larger the sample size, the better the scoring 
model’s accuracy. However, it is also worth noting that “a sample size of at least twenty observations in the smallest group is usually adequate 
to ensure robustness of any inferential tests that 
may be made” (Hintze, 1998). Therefore, in case 
of this model in which the number of independent variables is fi ve, there should be at least 100 
cases in smallest group to produce right discriminant function.
According to the World Bank, the proportion 
of non-performing loans to total gross loans in 
Vietnam is about 2.94 % or in other words the 
number of non-default borrowers is relatively 

higher than their counterparts, leading to the 
number of good and bad borrowers taken from 
banks in this study is not the same. Therefore, 
like the way other researchers such as Lee et 
al (2002); Desai et al (1996); Boritz & Kennedy 
(1995) and Dutta et al (1994) did, this study also 
choose the proportion of good borrowers to bad 
ones used was seven to three. Particularly, in 
case data of 500 customers will be used in this 
study, the number of good borrowers will be 
350 while their counterpart ones was 150. Moreover, information on 500 customers then will 
randomly be divided into two different groups 
named analysis sample and hold out sample. 
The former including 400 customers will be used 
to estimate discriminant function while the later 
including 100 customers will be used to check 
the validity of the model.
As data used in this study is numerical data, 
of which value can be measured numerically 
(Saunders et al, 2007), quantitative approach 
was applied. Particularly, quantitative approach 
was used to measure differences in means of independent variables between two groups. Moreover, quantitative analysis was also used to look 
for connections and spot relationships between 
independent variables.

3.2. Statistical analysis and checking 
assumptions

Before running discriminant analysis, it is 
important to describe characteristics of all variables used in this study and check assumptions 
to make sure that study’s fi ndings are accurate. 
In this study, data was processed by SPSS 21.
Firstly, as data in this study are continuous 
variables, descriptive was used to explore basic 
statistics such as mean, maximum, minimum, 
standard deviation of predictors in each group. 
Besides, independent sample T test SPSS was 
also used in this study to compare mean score 
on predictors between non defaulted and already defaulted group (Pallant, 2013).
Secondly, it is required that data used in discriminant analysis must be independent and 
normally distributed (Khemakhem and Boujelbene, 2015); therefore, like other researches this 
study also accesses normality of data’s distribution by the Kolmogorov-Smirnov test on SPSS.
Thirdly, not only normal distribution, but 
outliers and multicollinearity were also tested 

Review of Business and Economics Studies  
 
Volume 4, Number 4, 2016

to make sure results of further tests are accurate 
(Field, 2009; Pallant, 2013). It is clear that the 
presence of an outlier, which is defi ned as cases 
of which values are quite higher or lower than 
majority of other cases’ ones (Pallant, 2013), 
might make researchers miss important information and receive confusing results; therefore, 
it is essential to recognize outlier (Dielman, 
2001). Tails of distribution presented in graph 
named histogram was used to fi nd out there is 
potential outliers in this study or not. There are 
some observations are out at the outlier labelling rule, which after that will be eliminated. 
Besides, the existence of multicollinearity or 
explanatory variables are correlated might lead 
to estimates of parameter values are not reliable, and it is diffi cult for researchers to access 
the contributions of each independent variable 
to overall R2 (Gujarati, 1999). Therefore, this 
study used results obtained from correlation 
matrix, which presents not only correlation between dependent variable and predictors, but 
also between independent variables to test for 
multicollinearity. Particularly, Pearson produced 
moment correlation coeffi cient will be used. The 
highest absolute value of correlation coeffi cient 
between each of independent variable should 
be less than 0.7 to ensure that multicollinearity 
does not happen in this study.
After checking and correcting problems related to data, the next step is to apply discriminant 
analysis to the analysis sample. However, it is 

worth noting that there are two common methods for discriminant analyses, which are direct 
method and stepwise discriminant analysis. In 
this study, which is based on the previous research and theoretical model, the direct method 
will be used.

4. RESULTS
As can be seen from the table named group 
statistics, group means and standard deviations 
are calculated for each variable of the default 
and the non-default groups, which after that 
contributes to see whether the variables can differentiate between default customers and regular customers. It is true that, except for salary 
clear differences are witnessed in group means 
for the groups for the variables age, years at 
present job, number of dependents and loan 
amount. Particularly, average age for creditworthy borrowers, which is about 36 years old, 
is relatively higher than average age for the bad 
ones which is only a little above 30 years old. 
This result supports for conclusion of Vasanthi 
and Raja (2006) who said that the probability of 
default is higher with a younger borrower. The 
same pattern is also witnessed in term of number of dependents. This might be explained by 
the fact that the more people borrowers have to 
support fi nancially, the less money they have to 
pay loan or borrowers are likely not to pay loan 
in time. Moreover, there is big difference in years 
at present job between borrowers who are con
Table 1. Group Statistics

Ability to 
pay loan
N
Mean
Std. Deviation
Std. Error Mean

Age
Not good
120
30.719
3.9987
.3161

Good
280
36.772
5.5364
.2922

Salary
Not good
120
13.0419
4.19951
.33200

Good
280
14.0351
4.92672
.26410

Years at present job
Not good
120
5.38
2.454
.194

Good
280
10.26
3.679
.194

Number 
of independents

Not good
120
2.03
.812
.064

Good
280
1.53
.854
.045

Loan amount
Not good
120
398677156.250
165445876.1431
13079644.9524

Good
280
469608695.652
261697678.4845
14089329.3907

Review of Business and Economics Studies  
 
Volume 4, Number 4, 2016

sidered as credit worthy and not. Table 1 shows 
that average value of years at present job of no 
defaulted borrowers is nearly twice already defaulted borrowers’ ones. By contrast, the dissimilarity in monthly salary between good and 
bad borrowers is slight, which income among the 
defaulters is only one million VND lesser than 
the non-defaulters. More importantly, this difference might contribute to explain why loan 
amount of non-defaulters is relatively higher 
than defaulters.
As mentioned above, data used in discriminant analysis should be normally distributed 
(Khemakhem and Boujelbene, 2015); therefore, 
K-S test was used to fi nd out whether distribution of data used in study is normal or not.
The test statistic for the K-S test is presented 
in table 2 showing that the percentage of age D 
(396) = 0.095, p=.000, which was smaller than 
0.05; therefore, the distribution is not normal 
(Pallant, 2013). The same pattern also was witnessed in salary, years at present job, number 
of dependents and loan amount. To correct this 
problem, according to Field (2009), transforming 
data is one of popular options. Therefore, in this 
study, all variables were transformed into log 
transformation, which is as the same as method 
used by Hörkkö (2010). More importantly, Reichrt (1983), Hand et al (1996) and Uddin (2013) 
proved that discriminant analysis still get good 
result in case data used is not normally distributed. As a result of that, this problem in this 
study is not serious.
Besides, by looking at the tails of distribution 
presented in graph named histogram (Appendix 6), this study found that there are potential 
outliers because there are some observations are 
out at the outlier labelling rule. However, when 
considering information in descriptive table, the 

difference between 5 % trimmed mean (4.719) 
and mean (4.7161) values is extremely small; 
therefore, outlier problem in this study is not 
serious and might be solved by eliminating outliers.
According to Pallant (2013), multicollinearity happens when absolute value of correlation 
coeffi cient between each of independent variables is 0.7 or more. The correlations between 
variables used in this study (Table 3) showed the 
fi rst largest bivariate correlation was listed for 
relationship between age and years at present 
job. Unfortunately, this pair-wise correlation was 
only 0.770, which was clearly higher than 0.7; 
therefore, multicollinearity does happen and age 
will be omitted from regression.
As the sig. (2-tailed) value for predictors are 
below the required cut-off of 0.05; there is statistically signifi cant difference in salary, YAPJ, 
number of dependents and loan amount between the defaulters and non-defaulters.
Wilks’ lambdas and the F rations are estimated to test the equality of the group means. 
The value of the Wilks’ lambda () varies between 0 and 1. While the large value of  indicates that group means are not different, 
small value of  indicates that the group means 
are different or in other words the smaller the 
Wilks’s lambda, the more important the independent variable to the discriminant function. 
Wilks’s lambda is signifi cant by the F test for 
all independent variables. The lower signifi cant 
ratio for the corresponding F ratio means — the 
variable is very signifi cant in the case of determining group membership. Therefore, based on 
results presented in Table 4, it is obvious that 
dependents and years at present job may best 
discriminate between the two groups of borrowers.

Table 2. Tests of Normality

Kolmogorov-Smirnova
Shapiro-Wilk

Statistic
Df
Sig.
Statistic
Df
Sig.

Age
.095
400
.000
.968
396
.000

Salary
.097
400
.000
.948
390
.000

YAPJ
.079
400
.000
.962
392
.000

Dependents
.317
400
.000
.833
397
.000

Loan amount
.088
400
.000
.910
385
.000