Professional Documents
Culture Documents
Huaipeng Li
Department of Mathematics & Institute of Finance
Engineering, Jinan University
Guangzhou, China
lihp2006@163.com
Sulin Pang
Department of Public Administration & Institute of Finance
Engineering, School of Management, Jinan University
Guangzhou, China
pangsulin@163.com
AbstractBased on studying the KMV default distance model,
the work of this paper is to calculate the default distance using
BCC model in DEA methodology to quantify the credit risk.
The principle idea is to use DEA value of the company to
replace the market value of the company in KMV model, and
use average DEA points of ST companies within the industry
instead of default point, then get the Default Distance. The
third part is an empirical analysis which using 17 Chinese
textile companies in 2007 and 2008, the result shows that
default distance calculated with the DEA method does not
completely reflect the company's credit risk, but as a
discussion of method, the default distance with the DEA
method still have research value.
Keywords-credit risk; default distance; DEA;
BCC model
I. INTRODUCTION
Credit risk is that one party in a deal is in the risk of loss
while the other party is unable to keep the promise, but also
that the default bring the creditor risk of loss caused by
debtor who is unable to repay its debt. It can be translated
into the measurement of the financial situation of enterprises
that banks measure its customers credit risk, because the
formation of credit risk - whether the enterprise can be
scheduled debt service depends on the financial situation of
enterprises. There are many factors affecting the financial
situation, so, to assess the financial situation, we should use
multi-dimensional indicators affecting the financial situation,
but not a particular indicator [1].
So far, scholars analysis enterprises credit risk by using
various methods which generally can be attributed to the
following two ways[2]: one is a comprehensive analysis
based on financial ratios, of which the typical
representations are linear probability model, Logistic model,
Probit model, and discriminate analysis model, etc. The
other is a analysis phase based on modern financial theory
and information science theory including the KMV model,
Credit risk + model, Credit metrics models.
In recent years, a classification method called non-
parameter data envelopment analysis (DEA) is paid attention
widespread and studied deeply. It integrates multi-input
indicators to a composite input index, and multi-output
indicators to a composite output index, then determines the
two composite indexes rationality and effectiveness and
divides the evaluated subjects into two major categories of
relatively effective and ineffective, representing counter-
performance and default, respectively.
Yeh was one of the first scholars using financial ratios to
analysis DEA [3]; she combined DEA method with the
analysis of financial ratio to study bank performance. By
empirical analysis ,she found that DEA method can be used
to analysis bank's financial operating strategy through
dividing financial ratios into different dimensions according
to different meanings effectively . Emel and others used
DEA method to evaluate the company's credit risk according
to financial indicators [4]. Troutt and others, in 1996,
proposed a decision-making system based on an credit
accepted cases set given by experts, which formed piece
wise linear separating hyper plane as the credit boundary of
accepted set and refused set, talked about how to determine
the credit status of new units only knowing the credit stand
of good credit flats [5]. In 1998, Teiford, who improved the
system, proposed a DEA-type linear programming model, so
that we can judge new cases credit easier, and can easily
determine the location of new case relative to the boundary
[6]. Zhaohan Sheng, Guangmou Wu and others have studied
that in the DEA model, with the number of evaluation
indicators increasing, the effectiveness factor of each
decision making unit will increase. Most or all of DMU
efficiency would reach 1, if the number of indictors is up to
a certain extent [7]. In order to get a reasonable degree of
differentiation evaluation result we usually take that the
number of objects DMU is not less than two times of the
total number of input and output indicators [8], there are still
researchers thinking that the number of DMU should be at
least three times of the total number of input and output
indicators [9]. When the number of DMU changes largely,
we must re-select the evaluation index, which is obviously
undesirable.
II. THE KMV / DEA MODEL OF CREDIT RISK
A Introduction of KMV model
KMV model is developed as a credit risk measurement
tool by risk management firm called KMV. In the KMV
model, default risk is defined as probability that enterprises
asset value is less than default point. Default distance is a
ratio index measuring the size of default risk, it can embody
the degree that how company's assets is close to default
point before the financial crisis. Default distance is
calculated in KMV model as follows:
2010 International Conference on Computational Intelligence and Security
978-0-7695-4297-3/10 $26.00 2010 IEEE
DOI 10.1109/CIS.2010.25
82
2010 International Conference on Computational Intelligence and Security
978-0-7695-4297-3/10 $26.00 2010 IEEE
DOI 10.1109/CIS.2010.25
81
2010 International Conference on Computational Intelligence and Security
978-0-7695-4297-3/10 $26.00 2010 IEEE
DOI 10.1109/CIS.2010.25
81
'
( )
( ')*
E V DP
DD
E V
. (1)
In formula (1), represents expected asset value,
( ') E V
(2)
The model (2) is input-oriented BCC model.
Banker(1984) pointed out that there can exists negative
output factors in input-oriented BCC model and negative
input factors an in output-oriented BCC model according to
the theorem of invariance under the transformation. The
optimal value
*
. (3)
The optimal value
*