851

Int. J Latest Trends Fin. Eco. Sc. Vol‐5 No. 1 March, 2015

Loan Products and Credit Scoring by

Commercial Banks (India)
Rais Ahmad Itoo #1, A. Selvarasu #2, José António Filipe*3
#
Department of Business Administration, Annamalai University, India
*
Instituto Universitário de Lisboa (ISCTE-IUL), Portugal
1
[email protected]
2
[email protected]
3
[email protected]

Abstract - This study describes the loan products offered the economic development. Indian economy is
by commercial banks and credit scoring techniques among the fastest growing economies; and probably
used for classifying risks and granting credit to the main reason behind this was the financial and
applicants in India. Loan products offered by
banking sector reforms that have taken place after
commercial banks are from several kinds, since housing
1991. Banking sector has faced many circumstances
loans, personal loans, business loans until education
loans or vehicle loans, among many other types. All loan where growth has taken place but others with more
products are categorized either as secured or unsecured difficult situations. This kind of ups and downs were
loans. Credit scoring techniques used for both secured verified several times in the post independence era.
and unsecured loans are broadly divided into two Retail banking has always been important in India,
categories: Advanced Statistical Methods and coming back to the times when banks were
Traditional Statistical Methods. Some methodologies nationalized with the objective of reaching the
are presented to discuss Indian situation and masses.
understand the different kind on retail loans offered by
banks and the different credit scoring methods for The growth in retail banking has been quite
personal finance used by commercial banks in India. prominent retail in the recent years. Retail banking
Keywords ‐  Retail banks, Secured and Unsecured Loans, has been supported by growth in banking technology
Credit Scoring Techniques. and automation of the banking process.
Technological development has played the main role
1. Introduction for the rapid growth and spread of retail banking.
This banking segment has vast opportunity as well as
1.1 Initial Considerations
challenges in a growing economy as it is the case of
The human History allows well to understand Indian economy. The company A.T. Kearney, a
the risks of lending money. Borrowing and lending global management consulting firm, has identified
have long history related to human behavior (Thomas India as the second most attractive retail destination
et al., 2002). A credit risk component may be out of 30 emergent markets.
associated with lending transactions, in order to
reduce credit risk. Credit scoring methods may be The major relevant policy issues to retail
used to assess the credit worthiness of borrower. The banking may be pointed as financial capability,
information filled in the personal finance application consumer protection, regulation and responsible
form is used to develop a numerical score for each lending.
applicant (Lewis, 1992; Hand and Jacks, 1998; Retail banking is an interesting innovation in the
Thomas et al., 2002) and these scores may be used to 21st century in India and has experienced a fast
discriminate on bad and good loans (Durand, 1941). growth. As retail banking is a banking service that
Credit scores are broadly classified into two has increased primarily towards individual customer,
categories based on the method used to obtain scores, it focuses mainly on buyer markets. Retail banking
i.e. deductive or judgmental credit scoring and uses to apply mass-market strategy being the target
empirical or statistical credit scoring (Muller, 1997; the individual customer. The individual customer
Liu, 2001; Caire, 2004). uses local branches of larger commercial banks. The
1.2 Retail Banking In India main services offered by retail banks are saving and
checking accounts, mortgages, personal loans, debit
Any national economy has to have a good
cards, credit cards, among others. Moreover, it deals
banking system. It is very important for sustaining
also with various customers’ banking needs. Retail
____________________________________________________________________________________ banking provides several kind of features related to
International Journal of Latest Trends in Finance & Economic Sciences
IJLTFES, E‐ISSN: 2047‐0916
Copyright © ExcelingTech, Pub, UK (http://excelingtech.co.uk/)
 852
Int. J Latest Trends Fin. Eco. Sc. Vol‐5 No. 1 March, 2015

multiple products, channels and customer groups. In in accumulating NPAs in banking sector and may
India most of the banks are providing retail banking accentuate the indebtedness of households in the
services. medium term.
The considerable recent retail banking growth in 1.3 Credit Scoring
India is expected to continue in the future. There are Credit evaluation is one of the most crucial
some consistent reasons for this growth since processes in banks credit management decisions. This
technological factors, private and foreign banks process includes collecting, analyzing and classifying
introduced in India, the increased competition, the different credit elements and variables to assess the
significant innovation in banking products and credit decisions. The quality of bank loans is the key
services, the economic growth, the interest rates determinant of competition, survival and profitability.
deregulation, the consumerism, changes in life style One of the most important kits, to classify bank
of working/middle class, the focus on productivity customers, as a part of the credit evaluation process
and profitability, the driving towards low NPAs, the to reduce the current and the expected risk of a
changing consumer demographics. customer being bad credit, is credit scoring. Hand &
Retail banking sector in the banking industry is Jacka (1998) stated that “the process of modeling
continuously undergoing innovations, product creditworthiness by financial institutions is referred
reengineering, adjustments and alignments. As to as credit scoring”. It is also useful to provide
mentioned above, Indian retail banking segment is further definitions of credit scoring.
developing considerably and includes nowadays Anderson (2007) suggested that to define credit
many services. Besides credit cards, debit and ATM, scoring, the term should be broken down into two
housing loans, personal loans, consumption loans, components, credit and scoring. Firstly, simply the
education loans, vehicle loans, it includes also for word credit means “buy now, pay later”. It is derived
example insurance services, online services, among from the Latin word “credo”, which means “I
many other services. believe” or “I trust in”. Secondly, the word “scoring”
For the purposes of this study, a restriction of refers to “the use of a numerical tool to rank order
analysis is made being the study focused basically on cases according to some real or perceived quality in
loan products. order to discriminate between them, and ensure
objective and consistent decisions”. Therefore, scores
Retail lending is the exhortation in India. Most
might be presented as “numbers” to represent a single
banks have the retail segment on around 20% of their
quality, or “grades” which may be presented as
total lending portfolio, being this segment growing at
“letters” or “labels” to represent one or more qualities
an unnatural rate of 30 to 35% per annum. Retail
(Anderson, 2007). Consequently, credit scoring can
lending has been the key profit driver in the banking
be simply defined as “the use of statistical models to
sector in recent times. Retail banking used to be
transform relevant data into numerical measures that
synonymous of savings account and fixed deposits
guide credit decisions. It is the industrialization of
with cheque based/deposit slip based transactions.
trust; a logical future development of the subjective
Retail loans were usually restricted to housing loans.
credit ratings (Beynon, 2005) first provided by
This situation has changed considerably in the last
nineteenth century credit bureau, that has been driven
decade in India.
by a need for objectives, fast and consistent
RBI's report on Trend and Progress of India decisions, and made possible by advances in
(2003-04) has shown that the retail lending ranges technology” (Anderson, 2007). Furthermore, “credit
between Rs.1 20000 to Rs. 100 Lakh2, which are scoring is the use of statistical models to determine
generally for the duration of 5-7 years with housing the likelihood that a prospective borrower will default
loans granted for 15 years. It reveals some new trends on a loan. Credit scoring models are widely used to
in growth of credit. There is upsurge in retail credit as evaluate business, real estate, and consumer loans”
against corporate advances, which may reveal itself (Gup & Kolari, 2005). Also, “credit scoring is the set
of decision models and their underlying techniques
                                                             that aid lenders in the granting of consumer credit.
1
 1 USD= 59.310 Indian Rupees (27th October 2014).  These techniques decide who will get credit, how
2A lakh or lac is a unit in the Indian numbering system
equal to one hundred thousand (100,000; scientific much credit they should get and what operational
notation: 105). In the Indian convention of digit grouping, it strategies will enhance the profitability of the
is written as 1,00,000. For example, in India 150,000 borrowers to the lenders” (Thomas et al., 2002).
rupees becomes 1.5 lakh rupees.
 853
Int. J Latest Trends Fin. Eco. Sc. Vol‐5 No. 1 March, 2015

Credit scoring models (Lewis, 1992; Bailey, evaluating the classification capacity of proposed
2001; Mays, 2001; Malhotra & Malhotra, 2003; scoring model. ROC curves also known as Lorentz
Thomas et al., 2004; Chuang & Lin, 2009; Sustersic curves is a two dimensional graph that represents the
et al, 2009) are some of the most successful proposition of sensitivity (1-type II error) on y-axis
applications of research modeling in finance and and specificity (1-type I error) on x-axis. The
banking, as reflected in the number of scoring maximum distance between ROC and diagonal is
analysts in the industry, which is continually equal to constant times K-S statistics.
increasing. “However, credit scoring has been
Bellotti and Crook (2009) developed a credit
important in allowing the phenomenal growth in
score model with inclusion of time varying macro-
consumer credit over the last five decades. Without
variables like interest rate and unemployment rate
credit scoring techniques, as an accurate and
using Survival Analysis. Survival Analysis is
automatically operated risk assessment tool, lenders
competitive in comparison to LR as a credit scoring
of consumer credit could not have expanded their
method for prediction. The inclusion of macro-
loan effectively” (Thomas et al, 2002).
variables gave a statistically significant improvement
2 Literature Review in predictive performance.

There are many studies developing this subject. Kessy (2011) examined the link between the
Some studies are particularly interesting in order to loan processing and monitoring in banks, and asset
frame the analysis and to create some references for growth and empowerment of individual customers.
the developments that are intended to develop in this Findings revealed that better allocation and utilization
study. Some modelling cases follow around studies of financial institution’s economic capital not only
on this area. facilitates outreach to more under banked and
unbanked productive poor people but also
Hand and Crowde (2005), for example, used empowering them by stimulating investments and
latent-variable technique for measuring underlying increasing productivity in a cost-effective way for
aspects of credit customer behavior. The latent- poverty reduction.
variable model separates the observed variables into
primary characteristics (x) and behavioral Azam et al. (2012) evaluated the significance of
characteristics (y). Then the study summarizes them loan applicant socioeconomic attributes on personal
into overall measure of credit consumer scores. loan decision in banks using descriptive statistics and
logistic regression. The model identified that out of
Samreen et al. (2013) summarized the six independent variables only three variables
development of a credit scoring model known as (region, residence status and year with the current
Credit Scoring Model for Corporations (CSMC), organization) have significant impact on personal
which could be used to evaluate the creditworthiness loan decision.
of corporate borrowers before granting loan. Type I
and type II errors of proposed model (CSMC) have Matthew and Sarah (2013) investigated credit
more accuracy rate with no errors as compared to LR risk and default among Ghanaian banks. It was
and DA. suggested that banks should tighten their credit
assessment tool, i.e. CAMPARI (Character, Ability,
Li and Zhong (2012) introduced ensemble Model, Purpose, Amount, Repayment and Insurance)
learning model for credit scoring. This model points model. It was recommended that the Central Bank
out moving from static credit scoring to dynamic should facilitate the establishment of a vibrant credit-
behavioral scoring and maximizing revenue by referencing bureau in order to provide the credit
decreasing Type I and Type II errors. The challenges history for banks customers.
faced in building credit scoring models are half-
baked applicant’s information, missing values and Koh et al. (2006) illustrated the use of data
inaccurate information. mining techniques to construct credit scoring models.
The construction of this model has five steps:
Hussein and Pointon (2011) reviewed 214 defining the objective, selecting variables, selecting
articles/books of credit scoring applications. The sample and collecting data, selecting modelling tools
important and key determinants of credit scoring and constructing models, validating and assessing
models have been investigated. The matrix (ACC rate models.
criterion) measures the proportion of correctly
classified cases. ACC rate is a significant criterion in Nancy et al. (2013) studied the Credit Risk
Assessment Model of SBI Bank. SBI loan norms are
 854
Int. J Latest Trends Fin. Eco. Sc. Vol‐5 No. 1 March, 2015

flexible and differ from case to case. Loans applicant professionals, self-employed, businessmen and
information will be checked from RBI willful farmers can apply for this kind of loan.
defaulters lists.
4. Banks give loans to professional people like
3 Types of Retail Loans doctors or engineers, for example. These loans are
designed specially to cater to the financial needs of
Retail banks offer different types of loan to the the professional. A loan can be availed as a
individual customer to meet their diverse needs. The demand/term loan or overdraft as per customer’s
different types of loans offered by a retail banks to an preference. The loan is available for purchasing
individual customer may be home loans, loans office equipments viz. computers, fax, air-
against property, education loans, personal loans, conditioners, furniture, etc.
business loans, gold loans, credit card loans, or yet
even others. 5. Also, loans are available for
expansion/renovation/modernization of existing
Description of some loan types offered by retail premises. Traders loan facilities enable individuals,
banks is given below: proprietorships, bodies such as partnership firms and
1. Home Loan - available for purchasing co-operative societies to avail of working capital or
new/old dwelling unit, construction of a house, undertake development of shop by way of
purchasing a plot of land for construction of a house. loan/overdraft.
Banks have also designed housing loans facilities for 6. Dealers in gold/silver jewellary may get
NRI/PIO. Customer opt for Flexi Rate plan to hedge benefit of a loan. The loan is provided against the
the interest rate risk by breaking a loan into two security of tangible collateral securities in the form of
separate accounts, free property insurance and mortgage of land (not agricultural land) and building.
personal accident insurance. Borrower does not pay There is wedding in the family. May be a person likes
pre-payment/foreclosure charges for a part as well as to surprise the spouse with a priceless gift. People
full prepayment (when repaid from own sources by simply may like to pamper their family with an
the borrower). extended vacation. These are situations for which a
2. Education is the most important investment person may need a helping hand.
one may make in life. Higher studies and A bank offers personal loans to meet personal
specialization in certain fields call for additional requirements. A bank helps a customer to take care of
financial support from time to time. Whether all kind of expenses at a short notice. A loan may be
customers are planning school education for their availed to meet expenses related to marriage, a travel,
children (nursery to standard XII) pursuing a the honeymoon, a holiday or medical expenditures or
graduate or post-graduate degree, the bank gives yet for any other personal use. It is also available to
Education Loans, to fulfill customers’ ambitions and pensioners/defense pensioners. A loan is also
goals. available for Earnest Money Deposits for buyers of
3. In nowadays fast paced world, a vehicle is a home/flat/plot. A bank gives loans to a customer as
common necessity. Yet there are some expenses and an innovative combination of a loan and over draft
plans in life that take priority, the dream of owning a facility with flexible repayment options against the
car is now also important. Whether as a comfortable immovable property security of the customer.
and dependable means of transport or as a status Benefits of this loan are an ideal use of idle property -
symbol in society, it is believed that anyone deserves generating additional income from idle property,
today to have a vehicle. There are benefits for this customer withdraw money as per their needs and save
kind of loans. Available up to `15 Lakhs for any car on interest costs, deposit surplus money/regular
model (inclusive of gas-kit), loans can be availed for income/salary and save interest, flexibility to
new and second-hand vehicles (not more than 3 years withdraw money deposited earlier. Banks also
old). Repayment period of a loan is as long as 7 provide either as overdraft or demand loans as per the
years. For those individuals who prefer to travel more customer’s need.
conservatively or to get to their destinations faster, a
two-wheeler is as much a boon as it is to a car owner.
With newer models coming out each year, the options 4 Credit Scoring Procedures and
available to the customer are both attractive as well as Techniques
convenient. All resident Indians, salaried people,
 855
Int. J Latest Trends Fin. Eco. Sc. Vol‐5 No. 1 March, 2015

4.1 Credit Scoring Methodology Probably the earliest use of statistical scoring to
Credit scoring was primarily dedicated to distinguish between “good” and “bad” applicants was
assessing individuals who were granted loans, both made by Durand. Bankers will assign some weight
existing and new customers. Credit analysts, based on age to the loan application form filled by borrower.
pre-determined scores, reviewed customers’ credit Every bank has its own credit score cut points, which
history and creditworthiness to minimize the will cluster the customer into different risk groups.
probability of delinquency and default. A wide range of statistical techniques are used in
Basically, credit scoring is a method which can building scoring models. Most of these models are
be used to classify or quantify the risk factors statistical, being some of them non-linear; models are
relevant for a borrower's ability and willingness to applicable to build an efficient and effective credit
repay the loan. Credit scoring allows lenders to scoring system that is effectively used for predictive
predict likely loan outcomes based on the use of purposes. Techniques, such as weight of evidence
statistical techniques, which allow objective measure, regression analysis, discriminant analysis,
predictions as to whether a loan will produce a good probit analysis, logistic regression, linear
or bad outcome. Credit scoring can be used on a programming, Cox’s proportional hazard model,
standalone basis or as a part of the credit evaluation support vector machines, decision trees, neural
process. When used on a standalone basis, credit networks, k-nearest-neighbour, genetic algorithms
scoring assists in classifying applicants into and genetic programming, are all widely used
accept/reject groups or good/bad credits; when used techniques in building credit scoring models by credit
as part of the credit evaluation process, credit scoring analysts, bankers, lenders and computer software
can help to measure the credit risk of the applicants developers and providers.
(Thomas et al, 2002; Bhatia, 2006). 4.2 Advanced statistical methods vs. traditional
statistical methods
Durand (1941) was the first to recognise that one
could use the same techniques to discriminate Advanced statistical techniques, such neural
between good and bad loans. “Credit scoring is networks and genetic programming, provide an
essentially a way of recognizing the different groups alternative to conventional statistical techniques, such
in a population when one cannot see the as discriminant analysis, Probit analysis and logistic
characteristic that separates the groups” (Thomas, regression. The point of using sophisticated
2000). Commercially, the credit scoring was first techniques, such as neural nets, is their capability of
developed in the 1950s by Bill Fair and Earl Isaac, modeling extremely complex functions, and, of
but has only come into increasing use in the last two course, this stands in contrast to traditional linear
decades (Thomas, 2000). The main aim of the credit techniques, such as linear regression and linear
scoring model is to build a single aggregate risk discriminant analysis. Probabilistic neural nets
indicator for a set of risk factors from analysis of data usually trains presented cases faster than multi-layer
representative of the lender's own previous lending feed-forward nets, and classifies them in the same
experience (Thomas, 2000; Bhatia, 2006). way or better than multi-layer feed-forward nets,
even through multi-layer feed-forward nets have been
As per the information collected from the
shown to be excellent classifiers (Palisade, 2005;
bankers, the credit scoring for personal loans is done
Irwin, et al., 1995). However, a range of sophisticated
in line with the RBI guidelines. Almost all banks are
algorithms for neural nets training - making them an
following BASEL II and III guidelines. As per the
attractive alternative to the more conventional
experience of bankers, credit scoring process includes
techniques - has become available (Masters, 1995;
collecting, analyzing and classifying different credit
Palisade, 2005). Also, genetic programming is one of
elements and variables to assess the credit decisions.
the most successful alternatives to traditional
The quality of bank loans is the key determinant of
techniques recently used in this field. Genetic
competition, survival and profitability. It is one of the
programming is utilized to automatically determine
most important kits, to classify a bank’s customers, as
the sufficient discriminant functions and the
a part of the credit evaluation process to reduce the
applicable features simultaneously. Dissimilar neural
current and the expected risk of a customer as being a
networks may only suit large datasets, but genetic
bad credit. The objective of credit scoring models is
programming can positively perform well even with
to assign loan customers to either good credit or bad
small data-sets (Nath et al, 1997). Different credit
credit or predict the bad creditors. Therefore, scoring
problems are related to the classification analysis.
 856
Int. J Latest Trends Fin. Eco. Sc. Vol‐5 No. 1 March, 2015

scoring tools and techniques used by banks are the legal obligations (or conditions) of a loan, he is
discussed in existing literature. said to have defaulted on his/her loan. These defaults
increase the level of non-performing assets. In order
4.3 Credit Scoring Tools and Techniques in the
Literature to decrease the level of non-performing assets, the
bank has to develop loan application screening
Follows some references which are important methods, which distinguishes applicants as bad and
where this topic is developed by pursuing some good applicants, considering the credit scoring
different scoring tools and techniques: methods. Almost all banks have credit scoring
 Logistic Regression: Orgler (1971), Lucas methods, but still they have non-performing assets.
1992, Henley, 1995; Arminger et al.,1997, Desai et This study is carried out to know the credit scoring
al.,1997, Hand and Henley, 1997; Hand and Jacka, methods used by banks and list loan products offered
1998, West 2000, Baesens et al., 2003, Abdou et to customers.
al.,2009c 5.2 Objectives
 Discriminant Analysis: Durand (1941), The main objectives of the present study are the
Altman (1968), Boyle et al. (1992), Henley (1995), following ones:
Desai et al. (1997), Hand and Henley (1997);
1. To know different types of retail loans
Caouette et al. (1998); West (2000), Baesens et al.
offered by banks.
(2003), Malhotra and Malhotra (2003), Sarlija et al.
(2004); Abdou and Pointon (2009). 2. To know the different credit scoring
methods for personal finance available and used by
 Probit Analysis: Grablowsky and Talley
commercial banks.
(1981), Guillen and Artis (1992), Pindyck and
Rubinfeld (1997), Maddala (2001). 6 Current Credit Scoring Procedure
 DT or CART or recursive partitioning: The different types of loans offered by Indian
Baesens et al. (2003), Stefanowski and Wilk (2001), banks are home loans, auto loans, personal loans,
Fritz and Hosemann (2000), Hand and Jacka (1998), business loans, loans against property, gold loans and
Henley and Hand (1996) and Coffman (1986), credit cards loans. The process of loans sanctioning
Paleologo et al. (2010), Breiman et al. (1984); involves personal loans product promotions,
Arminger et al. (1997), Breiman et al. (1984), customers contacting the bank, filling up the
Rosenberg and Gleit (1994). application form. Physical evidences for the support
of a loan process are application forms, agreement,
 Neural Network: Bishop (1995); Masters
loan balance statement, and acknowledgement of
(1995); Arminger et al. (1997), Stefanowski and Wilk
repayment.
(2001), Lee et al. (2002), Malhotra and Malhotra
(2003), Kim and Sohn (2004), Zekic Susac et al. The interface (people) responsible for carrying
(2004), Lee and Chen (2005), Yim and Mitchell the whole loan process is a bank branch, a bank loan
(2005), Blochlinger and Leippold (2006), Seow and executive, a bank manager, a credit scoring
Thomas (2006), Trinkle and Baldwin (2007). executive, a CIBIL and a credit manager. Customer
requirements are: availability for different loan
 Genetic Programming: Koza (1994); Teller
products, low interest rates, low EMI, maximum
and Veloso (2000), Xia et al. (2000), McKee and
repayment time periods, diminishing rates of
Lensberg (2002), Nunez Letamendia (2002), Chen
interests, easy documentation, higher LTV and the
and Huang (2003), Zhang and Bhattacharyya (2004),
credit scoring. Once banks receive the filled
Ong et al. (2005), Lensberg et al. (2006), Huang et al.
application form of applicant for personal finance, it
(2006), Huang et al. (2007), Etemadi et al. (2009).
undergoes through various stages. The application
5 Methodology form will be sent to the credit risk department where
credit scores are calculated.
5.1 Statement of the Problem
There are two types of credit scoring done by
Considering the exposed above, it can be said
banks in India i.e. internal and external scoring. The
that the main income for retail banking is the interest
ranges as well as groups vary from bank to bank,
generated for the loans and advances. If this interest
while as the external CIBIL scores are the same for
or loans are not paid regularly it becomes a big
all banks. The CIBIL scores vary from 300 to 900 or
problem for the bank. When a borrower fails to meet
NA (Not applicable) or NH (No History), 300 being
 857
Int. J Latest Trends Fin. Eco. Sc. Vol‐5 No. 1 March, 2015

the lowest and 900 the highest. Banks prefer 7 Final Notes
applicants with a score higher than 700. The external
scoring is carried out to check the applicant’s banking In a new or emerging market, the operational,
with other banks and other financial institutions. The technical, business and cultural issues should be
components of the CIBIL credit score and credit considered with the implementation of the credit
report are payment history (35%), Amount Owed scoring models for retail loan products. The
(30%), Length of Credit History (15%), New Credit operational issues relate to the use of the model and it
(10%) and types of Credit used (10%) for scoring the is imperative that the staff and the management of the
applicant. CIBIL has all the transaction details of all bank understand the purpose of the model.
borrowers. Normally the whole process will take Application scoring models should be used for
place within 10 working days. making credit decisions on new applications and
behavioral models for retail loan products to
The commonly assessed customer details are: supervise existing borrowers for limiting the
bank history, income, banking (Annual Quarterly expansion or for marketing new products. The
Balance AQB), stability, etc. The whole processing technical issues relate to the development of proper
fee will be paid by banks. The internal credit scoring infrastructure, maintenance of historical data and
will be done within the bank, while as for the external software needed to build a credit scoring model for
scoring bank has to pay towards CIBIL. Bank will retail loan products within the bank. The business
not charge the processing fee to the applicant. The issues relate to whether the soundness and safety of
credit score can be improved considering the banks could be achieved through the adoption of
following principles: quantitative credit decision models, which would
1. Always pay your dues on time. send a positive impact in the banking sector. The
cultural issues relate to making credit irrespective of
2. Always keep your credit balance low on race, colour, sex, religion, marital status, age or
your credit card. ethnic origin. Further, models have to be validated so
3. Maintain a healthy mix of credit as to ensure that the model performance is compatible
in meeting the business as well as regulatory
4. Monitor your and your guarantor accounts
requirements.
balance frequently.
Thus, the above issues have to be considered
There are 4 main factors which mainly affect the
while developing and implementing credit scoring
score:
models for retail loan products.
1. payment history,
2. high utilization of credit limit,
3. higher percentage of credit cards or personal
finance and
4. many new accounts opened recently.

Bibliography Risk Management and Decision Automation.

New York: Oxford University Press.
[1] Abdou, H. (2009c). Genetic programming for [5] Arminger, G., Enache, D., Bonne, T. (1997).
credit scoring: The case of Egyptian public Analyzing Credit Risk Data: A Comparison of
sector banks. Expert Systems with Applications Logistic Discriminant, Classification Tree
36 (9): 11402-11417. Analysis, and Feedforward Networks.
[2] Abdou, H., Pointon, J. (2009). Credit scoring Computational Statistics 12 (2): 293-310.
and decision-making in Egyptian public sector [6] Baesens, B., Gestel, T. V., Viaene, S.,
banks. International Journal of Managerial Stepanova, M., Suykens, J., Vanthienen, J.
Finance 5 (4): 391-406. (2003). Benchmarking State-of-the-Art
[3] Altman, E. I. (1968). Financial Ratios, Classification Algorithms for Credit Scoring.
Discriminant Analysis and the Prediction of Journal of the Operational Research Society 54
Corporate Bankruptcy. The Journal of Finance (6): 627-635.
XXIII (4): 589-609. [7] Bailey, M. (2001). Credit scoring: the principles
[4] Anderson, R. (2007). The Credit Scoring and practicalities. Kingswood, Bristol: White
Toolkit: Theory and Practice for Retail Credit
 858
Int. J Latest Trends Fin. Eco. Sc. Vol‐5 No. 1 March, 2015

Box Publishing. Consumer Instalment Financing. New York:

[8] Beynon, M. J. (2005). Optimizing object National Bureau of Economic Research.
classification under ambiguity/ignorance: [21] Etemadi, H., Rostamy, A., Dehkordi, H.
application to the credit rating problem. (2009). A genetic programming model for
Intelligent Systems in Accounting, Finance and bankruptcy.
Management 13(2). pp. 113-130. [22] Fritz, S., Hosemann, D. (2000). Restructuring
[9] Bhatia, M. (2006). Credit Risk Management and the credit process: behaviour scoring for
Basel II: An implementation guide. London: german corporates. Intelligent Systems in
Risk Books. Accounting, Finance and Management 9(1): 9-
[10] Bishop, C. M. (1995). Neural Networks for 21.
Pattern Recognition. New York: Oxford [23] Grablowsky, B. J., Talley, W. K. (1981).
University Press Inc. Probit and discriminant functions for
[11] Blochlinger, A., Leippold, M. (2006). classifying credit applicants: a comparison.
Economic Benefit of Powerful Credit Scoring. Journal of Economic and Business, 33 (3):
Journal of Banking & Finance 30(3): 851-873. 254-261.
[12] Boyle, M., Crook, J. N., Hamilton, R., [24] Guillen, M., Artis, M. (1992). Count Data
Thomas, L. C. (1992). Methods for credit Models for a Credit Scoring System: The
scoring applied to slow payers. In Credit European Conference Series in Quantitative
Scoring and Credit Control, Thomas, L. C., Economics and Econometrics on Econometrics
Crook, J. N., Edelman, D. B., eds., Oxford of Duration, Count and Transition Models.
University Press, Oxford, 75-90. Paris.
[13] Breiman, L., Friedman, J. H., Olshen, R. A., [25] Gup, B. E., & Kolari, J. W. (2005).
Stone, C. J. (1984). Classification and Commercial Banking: The management of
Regression Trees. Belmont: The Wadsworth. risk. Alabama: John Wiley & Sons, Inc.
[14] Caire, D. (2004) Building Credit Scorecards [26] Hand, D. J. & Jacka, S. D. (1998). Statistics in
for Small Business Lending in Developing Finance, Arnold Applications of Statistics:
Markets. London: Bannock Consulting, London.
[15] Caouette, J. B., Altman, E. I., Narayanan, P. [27] Hand, D. J., Henley, W. E. (1997). Statistical
(1998). Managing Credit Risk: The Next Great Classification Methods in Consumer Credit
Financial Challenge. New York: John Wiley & Scoring: A Review. Journal of the Royal
Sons Inc. Statistical Society: Series A (Statistics in
[16] Chen, M., Huang, S. (2003). Credit scoring Society) 160 (3): 523-541.
and rejected instances reassigning through [28] Hand, D. J., Jacka, S. D. (1998). Statistics in
evolutionary computation techniques. Expert Finance, Arnold Applications of Statistics:
Systems with Applications 24(4): 433-441. London.
[17] Chuang, C. and Lin, R. (2009). Constructing a [29] Henley, W. E. (1995). Statistical aspects of
reassigning credit scoring model. Expert credit scoring. Ph.D. Thesis, The Open
Systems with Applications. 36 (2/1). pp.1685- University,Milton Keynes.
1694. [30] Huang, C., Chen, M., Wang, C. (2007). Credit
[18] Coffman, J. Y. (1986). The proper role of tree scoring with a data mining approach based on
analysis in forecasting the risk behaviour of support vector machines. Expert Systems with
borrowers. MDS Reports 3, 4, 7 and 9. Applications 33 (4): 847-856.
Management Decision Systems, Atlanta. [31] Huang, J., Tzeng, G., Ong, C. (2006). Two-
Cramer. stage genetic programming (2SGP) for the
[19] Desai, V. S., Conway, D. G., Crook, J. N., credit scoring model. Applied Mathematics
Overstreet, G. A. (1997). Credit scoring and Computation 174 (2): 1039-1053.
models in the credit union environment using [32] Irwin, G. W., Warwick, K., Hunt, K. J. (1995).
neural networks and genetic algorithms. IMA Neural networks applications in control.
Journal of Mathematics Applied in Business London: The Institution of Electronic
and Industry 8 (4): 323-3463. Engineers.
[20] Durand, D. (1941). Risk Elements in [33] Kim, Y. S., Sohn, S. Y. (2004). Managing
Consumer Instalment Financing, Studies in Loan Customers Using Misclassification
 859
Int. J Latest Trends Fin. Eco. Sc. Vol‐5 No. 1 March, 2015

Patterns of Credit Scoring Model. Expert [47] Muller, M. (1997). Computer-Assisted

Systems with Applications 26 (4): 567-573. Generalised Partial Linear Models, Proceedings
[34] Koza, J. R. (1994). Genetic Programming II of Interface'97, 29/1, pp.221-230.
Automation Discovery of Reusable Programs. [48] Nath, R., Rajagopalan, B. and Ryker, R.
Cambridge, MA: MIT Press. (1997). Determining the saliency of input
[35] Lee, T., Chen, I. (2005). A Two-Stage Hybrid variables in neural network classifiers.
Credit Scoring Model Using Artificial Neural Computers and Operations Researches. 24 (8).
Networks and Multivariate Adaptive pp. 767–773.
Regression Splines. Expert Systems with [49] Nunez-Letamendia, L. (2002). Trading
Applications 28 (4): 743-752. Systems Designed by Genetic Algorithms.
[36] Lee, T., Chiu, C. Lu, C., Chen, I. (2002). Managerial Finance 28 (8): 87-106.
Credit Scoring Using the Hybrid Neural [50] Ong, C., Huang, J., Tzeng, G. (2005). Building
Discriminant Technique. Expert Systems with Credit Scoring Models Using Genetic
Applications 23 (3): 245-254. Programming. Expert Systems with
[37] Lensberg, T. Eilifsen, A., McKee, T. (2006). Applications 29 (1): 41-47.
Bankruptcy theory development and [51] Orgler, Y. E. (1971). Evaluation of Bank
classification via genetic programming. Consumer Loans with Credit Scoring Models.
European Journal of Operational Research Journal of Bank Research 2 (1): 31-37.
169 (2): 766-697. [52] Paleologo, G., Elisseeff, A., Antonini, G.
[38] Lewis, E. M. (1992). An Introduction to Credit (2010). Subagging for credit scoring models.
Scoring. California: Fair, Isaac & Co., Inc. European Journal of Operational Research
Liang, Q. 2003. Corporate Financial Distress 201 (2): 490-499.
Diagnosis in China: Empirical Analysis Using. [53] Palisade Corporation. (2005). Neural Tools:
[39] Liu, Y. (2001). New Issues in Credit Scoring Neural Networks Add-In for Microsoft Excel.
Application, Technical Report 16/2001, Version 1.0. New York: Palisade Corporation.
Institulfur Wirtschaftsinformatik, Universitat [54] Pindyck, R. S., Rubinfeld, D. L. (1997).
Gottingen. Econometric Models and Economic Forecasts.
[40] Lucas, A. (1992). Updating scorecards: McGraw-Hill/Irwin.
removing the mystique. In Credit Scoring and [55] Rosenberg, E., Gleit, A. (1994). Quantitative
Credit Control, Thomas, L. C., Crook, J. N., methods in credit management: a survey.
Edelman, D. B., eds., Oxford University Press, Operations Research 42 (4): 589-61 3.
Oxford, 180-197. [56] Seow, H., Thomas, L. C. (2006). Using
[41] Maddala, G. S. (2001). Introduction to Adaptive Learning in Credit Scoring to
Econometrics. Chichester: John Wiley & Sons Estimate Take-Up Probability Distribution.
Inc. European Journal of Operational Research
[42] Malhotra, R., Malhotra, D. K. (2003). 173 (3): 880-892.
Evaluating consumer loans using Neural [57] Stefanowski, J., Wilk, S. (2001). Evaluating
Networks. Wiley & Sons Inc. business credit risk by means of approach-
[43] Masters, T. (1995). Advanced Algorithms for integrating decision rules and case-based
Neural Networks: AC++ Sourcebook. New learning. Intelligent Systems in Accounting,
York: Omega the International Journal of Finance and Management 10(2): 97-114.
Management Science. 31(2). pp. 83-96. [58] Sustersic, M., Mramor, D., and Zupan J.
[44] Mays, E. 2001. Handbook of Credit Scoring. (2009) Consumer credit scoring models with
Chicago: Glenlake Publishing Company, Ltd. limited data. Expert Systems with Applications.
[45] Mays, E. (2004). The Rule of Credit Scores in 36 (3). pp. 4736-4744.
Consumer Lending. In E. Mays, Credit [59] Teller, A., Veloso, M. (2000). Internal
Scoring. John Wiley & Sons, Inc. reinforcement in a connectionist genetic
[46] McKee, T., Lensberg, T. (2002). Genetic programming approach. Artificial Intelligence
programming and rough sets: A hybrid 120 (2): 165-198.
approach to bankruptcy classification. [60] Thomas, L. C. (2000). A survey of credit and
European Journal of Operational Research behavioural scoring: forecasting financial risk
138 (2): 436-451.
 860
Int. J Latest Trends Fin. Eco. Sc. Vol‐5 No. 1 March, 2015

of lending to consumers. International Journal International Journal of Business and

of Forecasting 16 (2): 149-172. Commerce 2(5): 1-26.
[61] Thomas, L. C., Edelman, D. B., Crook, J. N. [71] Xiao-Lin Li, Yu Zhong (2012). An Overview of
(2004). Readings in Credit Scoring: recent Personal Credit Scoring: Techniques and
developments, advances, and aims. New York: Future Work. International Journal of
Oxford University Press. Intelligence Science 2: 181-189.
[62] Thomas, L. C., Edelman, D. B., Crook, L. N. [72] Hussein A. Abdou and John Pointon (2011).
(2002). Credit Scoring and Its Applications. Credit Scoring, Statistical Techniques And
Philadelphia: Society for Industrial and Evaluation Criteria: A Review Of The
Applied Mathematics. Literature. Intell. Sys. Acc. Fin. Mgmt. 18,59–
[63] Trinkle, B. S., Baldwin, A. A. (2007). 88
Interpretable credit model development via [73] T Bellotti and J Crook (2009). Credit scoring w
artificial neural networks. Intelligent Systems ith macroeconomic variables using survival
in Accounting, Finance and Management 15 analysis. Journal of the Operational Research
(3-4): 123-147. Society 60, 1699-1707.
[64] West, D. (2000). Neural Network Credit [74] Rehan Azam, Muhammad Danish and Syed
Scoring Models. Computers & Operations Suleman Akbar (2012). The significance of
Research 27 (11-12): 1131-1152. socioeconomic factors on personal loan
[65] Xia, Y., Liu, B., Wang, S., Lai, K. K. (2000). decision a study of consumer banking local
A model for portfolio selection with order of private banks in Pakistan. IQRA University.
expected returns. Computers & Operations [75] Severine S. Kessy (2011). Consumer Loan
Research 27 (5): 409-422. Processing and Monitoring in Tanzanian
[66] Yim J., Mitchell H. (2005). Comparison of Banks: A Strategy for Growth and
country risk models: hybrid neural networks, Empowerment of Micro and Small
logit models, discriminant analysis and cluster Enterprises. Journal of Marketing
techniques. Expert Systems with Applications Development and Competitiveness 5(7). 45:52.
28 (1): 137-148. [76] Matthew Ntow-Gyamfi and Sarah Serwaa
[67] Zekic-Susac, M., Sarlija, N., Bensic, M. Boateng (2013). Credit risk and loan default
(2004). Small Business Credit Scoring: A among Ghanaian banks: An exploratory study.
Comparison of Logistic Regression, Neural Management Science Letters 3 753–762
Networks, and Decision Tree Models. 26th [77] Nancy Arora, Arti Gaur And Babita (2013).
International Conference on Information Credit Appraisal Process of SBI: A Case Study
Technology Interfaces. Croatia. of Branch of SBI in Hisar. Arth Prabhand: A
[68] Zhang, Y., Bhattacharyya, S. (2004). Genetic Journal of Economics and Management 2(1):
programming in classifying large-scale 10-26.
data:an ensemble method. Information [78] Hian Chye Koh, Wei Chin Tan and Chwee Peng
Sciences 163 (1-3) : 85-101. Goh (2006) A Two-step Method to Construct
[69] David J Hand and Martin J Crowder (2005). Credit Scoring Models with Data Mining
Measuring customer quality in retail banking. Techniques. International Journal of Business
Statistical Modelling 5: 145–158. and Information 1(1): 96-118
[70] Asia Samreen, Farheen Batul Zaidi and Aamir
Sarwar (2013). Design and Development of
Credit Scoring Model for the Commercial
Banks in Pakistan: Forecasting
Creditworthiness of Corporate Borrowers.