a credit risk model for bank’s loan portfolio optimize … · process in terms of a markov chain...

ijcrb.webs.com

INTERDISCIPLINARY JOURNAL OF CONTEMPORARY RESEARCH IN BUSINESS

COPY RIGHT © 2012 Institute of Interdisciplinary Business Research

125

MAY 2012

VOL 4, NO 1

A CREDIT RISK MODEL FOR BANK’S LOAN PORTFOLIO &

OPTIMIZE THE VAR

Meysam Salari 1 ‡

, S. Hassan Ghodsypour2,Mohammad Sabbaghi Lemraski 3, Hamed

Heidari.1

1 M.Sc. in industrial Engineering, Department of Industrial Engineering, Amirkabir

University of Technology, Tehran, Iran

2 Professor of industrial engineering, Department of Industrial Engineering, Amirkabir

University of Technology, Tehran, Iran

3 M.A. student in financial management, University of Tehran, Tehran, Iran

Abstract The New Basel accord has highlighted the need for models of the credit risk in portfolios of

corporate loans. There are really no such models of the risks in corporate loan portfolios even

though there is a well established industry – credit scoring – in modelling the risk of

individual loans. This paper proposes a method to calculate portfolio credit risk. Individual

default risk estimates are used to compose a value-at-risk (VaR) measure for credit risk. This

empirical research study attempts to measure credit risk of a bank’s corporate loan portfolio,

including firms from 5 different business sectors. Credit loss distributions for each segment

are selected and used in a Monte Carlo simulation process to generate the loss distribution of

the portfolios. We use samples of Agriculture, Manufacturing and mining, Construction and

housing, Exports, Trade services and miscellaneous to obtain efficiency estimates for

individual firms in each industry. The monthly observations of the total amount of corporate

loans and defaults across various sectors were supplied by a major commercial bank in Iran.

This period covers 36 monthly observations. The observed sectoral default rates are needed to

be simulated to obtain a nicely shaped distribution. Monte Carlo simulation is applied for

1,500 times. Based on the simulated default rates, the expected sectoral default rates are

computed. The sectoral weights in the whole loan portfolio are multiplied by the expected

sectoral default rates matrix, considering cross-sectoral correlations to get the total amount of

the bank’s credit risk and capital requirement. Finally, we model credit risk of portfolio to

optimize the percentage of loan allocation to each cluster in order to minimize the VaR of the

portfolio.

Keywords: IRB approach, Risk management, Default rate, Credit risk, Value at Risk

1. Introduction

The growth in liquidity of credit markets and the active management of credit risk are among

the most significant developments in commercial banking in recent years. These

developments hold the potential to permanently reduce the risk profile and improve the

financial performance of commercial banks.

Almost all banking activities inherently incur credit risk. So, accurate measurement of credit

risks banks bear are critical since the Basel standards І,ІІ and ІІІ requires all banks to allocate

sufficient amount of capital for all risks they incur through their banking activities. On the

‡ Corresponding author,‡MeysamSalari, M.Sc., industrial Engineering, Department of Industrial Engineering and

Management Systems, Amirkabir University of Technology, Tehran, Iran

ijcrb.webs.com



126

MAY 2012

VOL 4, NO 1

other hand, the newly proposed Basel Accord III encourages banks to use either foundation

or advanced internal ratings base (IRB) approach in measuring credit risks of their portfolios.

The use of technical and advanced approaches in the measurement of credit risk of banks’

portfolios has nowadays become a very hot issue. The most recent technical report issued by

the Basel Committee has concentrated heavily on the measurement of credit risk. Using either

foundation or advanced Internal Ratings Base (IRB) approaches. The foundation IRB

approach employs predetermined parameters of “exposure at default” and “loss given

default,” and requires the estimation of bank specific default rates. On the other hand,

advanced IRB approach requires that all underlying parameters are estimated by individual

banks. The application of IRB approaches needs a large scale of detailed and classified

historical observations. IRB models may be easily applied by simulating historical

observations a sufficient number of times to obtain nicely shaped normal distributions.

Hence, the simulated sectoral default distributions can be used to determine credit risk of a

bank’s loan portfolio and amount of capital requirement (Teker, Akcay, & Turan, 2006).

While credit portfolio management was created for a defensive purpose (to reduce the losses

in the loan portfolio), the rational management of economic capital is driving this function to

embrace return and risk in its performance objectives. The continued progress toward rational

asset management will require commercial banks to overcome formidable obstacles that

impede or prevent them from fully integrating their activities into the market.

Models of portfolio credit risk predict the distribution of credit loss or value for a portfolio.

The loss or value distribution can be used for many applications in risk management such as

Economic Capital and Value-at-Risk assessment, performance evaluation and marginal risk

estimation.

The quantitative loan portfolio management process solutions facilitate the decrease in

portfolio risk through better risk identification and risk diversification, and increase portfolio

profitability through the reduction of portfolio volatility and the increase in customer

profitability.

Credit portfolio management grew out of the need to improve the financial performance of

the large corporate loan portfolios in commercial banks. Lending is the backbone of

commercial banking, so lending is what banks should do best. Yet these portfolios proved to

be the source of recurring problems and the cause of failure for many institutions.

Credit policies, procedures, systems and controls do not always assure asset quality and

earnings will be maintained at acceptable levels. A quantitative approach is necessary for

effective loan portfolio management.

Objectives of measuring credit risk of a bank’s corporate loan portfolio is to design a capital

adequacy framework thatresponds dynamically to changes in credit quality, alerts bank

management, supervisors, and others to emerging problems more quickly than under current

capital accord, better suited to the complex activities of large, capable of adapting to market

and product evolution, encourages banks to invest additional resources in risk management

activities.

Corporate credit risk models split into structural and reduced form modeling and these are

reviewed in more detail in next Section. Structural models were introduced by Merton

(Merton, 1974), while reduced form models were developed from the work of Jarrow and

Turnbull (Jarrow, Lando, & Turnbull, 1995).

Through effective management of credit risk exposure, banks not only support the

viability and profitability of their own business but also contribute to systemic stability and to

an efficient allocation of capital in the economy.

As mentioned before, in Section 2 we review different models for estimating the credit

risk of portfolios of corporate loans, in particular the structural and reduced form models. In

section 3 we explain the empirical data and methodology employed for simulating historical

ijcrb.webs.com



127

MAY 2012

VOL 4, NO 1

observations, estimating future sectoral default rates and calculating bank’s capital

requirement. Section 4 describes the data and discusses the empirical findings, we model the

VaR of portfolio to optimize it in this section and finally section 5 covers the implications

and conclusions.

2. Literature Review

Corporate bonds are the way lending to large companies is structured and losses occur with

these (a) because the company defaults; (b) because the rating of the loan is lowered by rating

agencies, like Standard and Poor or Moodys, and hence the loan is worth less in the market

because of the increased risk of default; (c) the difference between the value of a bond of a

given rating and that of the highest grade bond increases. This last problem affects the whole

bond market and so is considered to be a market risk rather than a credit risk. The first split in

models is between those which only try to estimate the chance of default, called default mode

(DM) models, and those which try to estimate both the chance of default and of ratings

changed, called mark-to-market (MTM) models. The second split is between structural

models which relate the value of a bond and its default probability to the capital and debt

structure of the firm, and the reduced form models, which model the default process and/or

the rating change process directly using historical estimates. Lastly there is split between

static models which can only model the credit over a fixed time interval and dynamic models

which allow one to model over any time interval. (Thomas, 2009)

A variety of analytical techniques have been used for credit risk assessment. They include

statistical methods, models based on contingent claims and asset value coverage of debt

obligations, operational research (OR) such as linear or quadratic programming, data

envelopment analysis (DEA) and neural networks models.

Since the pioneering work of Altman (Altman, 1968) a plethora of statistical bankruptcy

prediction studies have been done using multivariate discriminant analysis (Deakin, 1972)

(Blum, 1974), logistic regression (Martin, 1977) (Ohlson, 1980) (Zavgren, 1985) (Keasey,

McGuinness, & Short, 1990), and probit analysis (Zmijewski, 1984) (Skogsvik, 1990). More

recent works include Kolari who developed an early-warning system for bank failure based

on logit analysis and trait recognition (Kolari, Glennon, Shin, & Caputo, 2002), Jones and

Hensher used a mixed logit model to predict financial distress (Jones & Hensher, 2004), and

Canbas who combined discriminant analysis, probit, logit and principal component models to

create an integrated early-warning system for bank failure (Canbas, Cabuk, & Kilic, 2005).

The reduced approach developed from the idea of Pye (Pye, 1974) that one could use an

actuarial approach and model the default rate of a group of bonds over time as is done in

mortality tables for humans. This is a static, one-period or infinite period model if one thinks

of the one period being infinitely repeated). Jarrow and Turnbull were able to reach out the

idea to a dynamic model by replacing the fixed probability of default by a hazard rate which

depicts the changes in the default probability over the duration of the loans. (Jarrow &

Turnbull, 1995) The basic approach is a DM model but by modelling the dynamics of the

process in terms of a Markov chain model on the ratings grades including a defaulted ratings

grade, they were about to develop a MTM version of this approach. (Jarrow, Lando, &

Turnbull, 1995)

As Guo et al (Guo, Jarrow, & Zeng, 2005)indicate structural models are based on the

information on asset values and liabilities available only to the firm’s management while

reduced form models are based on information available in the market, that there is only

indirect information on a firm’s assets and liabilities. Reconciliation models are being

developedwhich do not have complete information on the dynamics of the process that

ijcrb.webs.com



128

MAY 2012

VOL 4, NO 1

triggers default in the structural models and hence lead to cumulative rates of default as in

reduced form models.(Cuny & Lejeune, 2003)

Finally there are the static scorecard models which search for estimate the probability of

default in a fixed period of time as a function of accounting ratios of firms, and now

including other market information. These are DM models.(Thomas, 2009). DM models are

the static scorecard models which seek to estimate the probability of default as a function. It

was pioneered by Altman. (Altman, 1968)

This empirical research attempts to measure the credit risk of a sample bank’s corporate loan

portfolio using advanced IRB approach. The foundation of this method is based on assigning

a credit rating to each individual loan and individual firm considering both historical or

expected default rates and cross-sectoral correlations. Historical observations for each type of

loan and firm must be sufficiently large to have a nicely shaped probability distribution. If

not, the observed default rates may be simulated a sufficient number of times to obtain a

smooth probability distribution. Finally we model the default rate of portfolio and optimize it

in order to minimize the risk of portfolio as well as satisfying Constraints.

3. Data and methodology

We used data involving installment loans of corporates. These data were supplied by a major

commercial bank in Iran. The data included all the payment behavior registered for these

corporates from March 2007 up to March 2009 covering a total of 36 months. Table 1 shows

the name of the sectors and the numbers assigned to.

Table 1: Sector Names and Numbers Assigned

Sector Name Sector Number Assigned

Agriculture 1

Exports 2

Construction and housing 3

Manufacturing and mining 4

Trade services and miscellaneous 5

A sector of the portfolio can be viewed as a group of expositions, and its default rate in a

specific period can be obtained by calculating the weighted average of the default rate of the

expositions in that period, using the exposition amounts for weighting. Using this process, we

constructed time loss series for each of the 5 sectors that the bank assigns its loans.

The monthly default rate for each corporate is computed as:

Li =

Where:

Li = Default rate of month i

Hi = The amount of exposition i

ijcrb.webs.com



129

MAY 2012

VOL 4, NO 1

And Ai is the penalty of exposition i and computed by:

Ai =

Where:

aj = The amount of exposition i paidtj days after due date.

R1 = The interest rate of bank

R2 = The penalty rate

tj = Number of days after due date of exposition.

And the monthly sectoral default rates in each sector are computed as:

Mi =

Where:

n = Number of corporates

Li = Corporate default rate of month i

Hi = The amount of exposition of month i

Mi = Sectoral default rate of month i

We used a random sample of 9000 expositions (each exposition represents one installment of

a loan) to estimate monthly default rates for 36 months. Table 2 shows the computed sectoral

default rate matrix (36 × 10). Since the numbers of historical observations are not sufficient

to obtain a smooth probability distribution, the sectoral default rates need to be simulated. We

use Monte Carlo simulation as simulation method. First of all we determine the type of

statistical distribution function of defaults using a simulation software package, “Arena”.

The results support that the default rate series are fit nicely into lognormal distribution at 93%

confidence level. The mean and standard deviation of each sector are computed in table 2.

Then sectoral variance-covariance and Cholesky Decompositionmatrixes of variance-

covariance are computed (Table 3 and 4).

Employing a variance-covariance matrix as an input, Monte Carlo simulation is applied for

1,500 times. Based on the simulated default rates, the mean and the standard deviation of

expected sectoral default rates are computed. Then, a credit quality rating scale is fitted into

the sectoral default rates distributions.

Therefore, credit quality ratings are assigned for each sector based on the expected defaults in

the future (Graphs 1-5).

Table 2: Historical Sectoral Default Rates

Month-year

SECTORS

1 2 3 4 5

Apr-2006 0.2542 0.0663 0.1011 0.1114 0.0207

May-2006 0.2523 0.0732 0.1164 0.1484 0.0826

Jun-2006 0.2439 0.0723 0.1231 0.1405 0.0512

Jul-2006 0.2642 0.0772 0.1499 0.2018 0.0540

Aug-2006 0.2447 0.0324 0.1288 0.1283 0.0633

Sep-2006 0.2469 0.0340 0.0703 0.1408 0.0861

Oct-2006 0.2415 0.0716 0.1347 0.1955 0.0795

Nov-2006 0.2545 0.0423 0.0847 0.1083 0.0430

Dec-2006 0.2474 0.0269 0.0808 0.2034 0.0617

Jan-2007 0.2439 0.0933 0.0729 0.0981 0.0583

ijcrb.webs.com



130

MAY 2012

VOL 4, NO 1

Feb-2007 0.2316 0.0490 0.0875 0.1754 0.0389

Mar-2007 0.2538 0.2096 0.0757 0.1463 0.0749

Apr-2007 0.2407 0.0841 0.0878 0.1311 0.0923

May-2007 0.2273 0.1479 0.0809 0.1127 0.0808

Jun-2007 0.2278 0.0971 0.0706 0.1219 0.1164

Jul-2007 0.2619 0.0400 0.1095 0.1903 0.0610

Aug-2007 0.2344 0.0558 0.0725 0.1289 0.0596

Sep-2007 0.2296 0.0836 0.0693 0.1394 0.0462

Oct-2007 0.2534 0.1029 0.0653 0.1296 0.0576

Nov-2007 0.2381 0.0545 0.0773 0.1579 0.0524

Dec-2007 0.2227 0.1061 0.0971 0.1465 0.0760

Jan-2008 0.2260 0.0252 0.0666 0.1516 0.0390

Feb-2008 0.2550 0.0431 0.0840 0.1184 0.0635

Mar-2008 0.2402 0.0366 0.0991 0.1305 0.0377

Apr-2008 0.2187 0.0669 0.0739 0.1327 0.0393

May-2008 0.2526 0.1019 0.0720 0.1060 0.0809

Jun-2008 0.2365 0.0742 0.0864 0.1743 0.0769

Jul-2008 0.2228 0.2016 0.0760 0.1760 0.1169

Aug-2008 0.2476 0.0401 0.0789 0.1928 0.0331

Sep-2008 0.2399 0.0817 0.0833 0.1426 0.0445

Oct-2008 0.2537 0.1201 0.1108 0.1569 0.0643

Nov-2008 0.2597 0.0860 0.0830 0.1480 0.0706

Dec-2008 0.2361 0.0710 0.1120 0.1129 0.0588

Jan-2009 0.2491 0.0264 0.1022 0.1170 0.0953

Feb-2009 0.2512 0.0817 0.0887 0.2204 0.0855

Mar-2009 0.2474 0.0736 0.1103 0.2102 0.0502

Mean 0.2431 0.0764 0.0912 0.1485 0.0643

SD 0.0118 0.0426 0.0211 0.0329 0.0221

Simulation

Results

Mean 0.2431 0.0766 0.0916 0.1493 0.0645

SD 0.0115 0.0422 0.0210 0.0334 0.0215

Table 3: VARIANC-COVARIANCE MATRIX

Sector

Number 1 2 3 4 5

1 0.00014 -0.00008 0.00009 0.00005 -0.00003

2 -0.00008 0.00176 -0.00014 -0.00006 0.00041

3 0.00009 -0.00014 0.00043 0.0002 -0.00003

4 0.00005 -0.00006 0.0002 0.00105 0.00001

5 -0.00003 0.00041 -0.00003 0.00001 0.00048

ijcrb.webs.com



131

MAY 2012

VOL 4, NO 1

TABLE 4. CHOLESKY DECOMPOSITION MATRIX

Sector

Number 1 2 3 4 5

1 0.01166 0 0 0 0

2 -0.00653 0.04146 0 0 0

3 0.00749 -0.0023 0.01922 0 0

4 0.00467 -0.00079 0.0086 0.0309 0

5 -0.00239 0.00944 0.00051 0.00062 0.01951

4. Empirical results and implications

It appears that the values of Monte Carlo simulated expected sectoral mean default rates and

standard deviations are close to the values of actual sectoral mean default rates and standard

deviations. For example, the actual mean and standard deviations for sector 1 on Table 2 are

24.3% and 1.18%, respectively. The simulated sectoral mean and standard deviation for

sector 1 are 24.3% and 1.15%, respectively. Therefore, it may be argued that the simulations

are a good representative of historical observations. The distributions of simulated sectoral

default rates are presented in Graphs 1 to5.

As regards to resulted sectoral default rates, suppose that the Bank has disbursed $30 million

in Agriculture sector, we can infer that at 95% confidence level less than $7.8 million will

never be paid or will not be paid on time. So the bank should reserve $7.8 million as VaR.

To show a practical application for the simulated sectoral default rates, the Monte Carlo

simulation presented the situation where a sample bank formed a corporate loan portfolio in

the amount of $100 million which was composed of 5 different sectors with weights:

{30,20,15,25,10}. The credit risk of the bank’s loan portfolio is computed by multiplying the

sectoral weights in the portfolio and expected sectoral default probability distribution.

Incorporating the effects of cross-sectoral correlations on defaults rates, the simulated default

rates are multiplied by Transposed Cholesky Decomposition matrix (Table 4) to calculate the

total credit risk of the loan portfolio (Graph 6).

The results show that the capital requirement for the credit risk of a $100 million corporate

loan portfolio is $18.04 million at 99%, and $16.89 million at 95% confidence level (Graph

6). So if bank needs to avoid bankruptcy at a 95% confidence level should reserve $16.98

million in the bank.

In order to assign optimum amount of loan in each sector the portfolio is modeled as bellow:

Min Z1 = AX

Min Z2 = tXSX

S.T

Hx

n

i

i 1

Ixi 2.0 ni ,..,1

Where:

Xi=The fraction of loan in sector i

S = Variance-covariance matrix

ijcrb.webs.com



132

MAY 2012

VOL 4, NO 1

A=Average default rate matrix

H=Total amount of loan portfolio

I=The amount of money paid to the bank by the Central Bank of Iran for loaning.

Assume that the bank has allotted $100 million for loaning, $70 million of which is provided

by the central bank, thus:

H=$100 million

I=$70 million

Solving the model with lingo, the result is as bellow:

X1 =14.00000

X2 =14.00000

X3 =20.19588

X4 =14.00000

X5 =37.80412

Therefore the optimum loan assignment is $14 million for Agriculture, $14 million for

Manufacturing and mining, $20.19 million forConstruction and housing,$14 million

forExports, $37.80 million for Trade services and miscellaneous.

5. Conclusion

The aim of the negotiations on revising the current Capital Accord (Basel I) was to develop a

morerisk-sensitive framework for determining capital requirements.

By means of Monte-Carlo simulation, we propose a VaR measure for the sample portfolio of

loans. We also show how calculating VaR can enable financial institutions to evaluate

alternative lending policies on the basis of their implied credit risks and loss rates.

This research paper attempts to measure the credit risk and capital requirement of a sample

bank’s corporate loan portfolio using advanced IRB approach. Since the amount of historical

default rates are limited, the expected sectoral default rates are simulated 1,500 times using

Monte Carlo approach to get smooth sectoral probability distributions. The simulated default

rates are then applied to a theoretical bank’s corporate loan portfolio in the value of $100

million. The calculation results show that the bank needs to allocate $19.05 million capital for

credit risk for its loan portfolio using IRB approach. Finally we model VaR of bank in order

to minimize it before lending. We consider all constraints that a commercial bank in Iran

might have and solve the model for $100 million of lending.

The consequence of this research is a fully-fledged range of solutions to deliver outstanding

credit performance, with proven results, which are now available to serve VaR of portfolio.

In summary, VaR calculating platform has been conceived, designed, created and activated to

achieve objectives and to provide credit grantors that can elevate them to a position of credit

excellence with the confidence and assurance of achieving excellent results.

ijcrb.webs.com



133

MAY 2012

VOL 4, NO 1

References

Altman, E. (1968). Financial ratios, discriminant analysis and the prediction of corporate bankruptcy.

Journal of Finance , 589–609.

Blum, M. (1974). Failing company discriminant analysis. Journal of Accounting Research , 1–25.

Canbas, S., Cabuk, A., & Kilic, S. (2005). Prediction of commercial bank failure via multivariate

statistical analysis of financial structure: The Turkish case. European Journal of Operational

Research , 528–546.

Cuny, X., & Lejeune, M. (2003, February). Statistical modelling and risk assessment. Safety Science ,

29-51.

Deakin, E. (1972). A discriminant analysis of predictors of business failure. Journal of Accounting

Research , 167–179.

Guo, X., Jarrow, R., & Zeng, Y. (2005). Information Reduction in Credit Risk Models. Ithaca:

Working Paper, School of Operations Research and Industrial, Cornell University,.

Jarrow, R., & Turnbull, S. (1995). Journal of Finance , 53–86.

Jarrow, R., Lando, D., & Turnbull, S. (1995). A Markov model for the term structure of credit risk

spreads. Review of Financial Studies , 481-523.

Jones, S., & Hensher, D. (2004). Predicting firm financial distress: A mixed logit model. Accounting

Review , 1011–1038.

Keasey, k., McGuinness, P., & Short, H. (1990). Multilogit approach to predicting corporate failure –

Further analysis and the issue of signal consistency. Omega , 85–94.

Kolari, J., Glennon, D., Shin, H., & Caputo, M. (2002). Predicting large US commercial bank failures.

Journal of Economics and Business , 361–387.

Martin, D. (1977). Early warning of bank failure: A logit regression approach. Journal of Banking and

Finance , 249–276.

Merton, R. (1974). On the pricing of corporate debt: the risk structure of interest rates. Journal of

Finance , 449-470.

Ohlson, J. (1980). Financial ratios and the probabilistic prediction of bankruptcy. Journal of

Accounting Research , 61–80.

Pye, G. (1974). Gauging the default process. Financial Analysts Journal , 49–52.

Skogsvik, K. (1990). Current cost accounting ratios as predictors of business failure: The Swedish

case. Journal of Business Finance and Accounting , 137–160.

Teker, S., Akcay, B., & Turan, M. (2006). Measuring Credit Risk of a Bank's Corporate Loan

Portfolio Using Advanced Internal Ratings Base Approach. Journal of Transnational Management ,

17-40.

ijcrb.webs.com



134

MAY 2012

VOL 4, NO 1

Thomas, L. C. (2009). Modelling the credit risk for portfolios of consumer loans: Analogies with

corporate loan models. Mathematics and Computers in Simulation , 2525-2534.

Zavgren, C. (1985). Assessing the vulnerability to failure of American industrial firms: A logistic

analysis. Journal of Business Finance and Accounting , 19– 45.

Zmijewski, M. (1984). Methodological issues related to estimation of financial distress prediction

models. Journal of Accounting Research , 58–59.

ijcrb.webs.com



135

MAY 2012

VOL 4, NO 1

Annexure

0

50

100

150

200

250

0.20

22

0.20

63

0.21

04

0.21

45

0.21

87

0.22

28

0.22

69

0.23

10

0.23

52

0.23

93

0.24

34

0.24

75

0.25

16

0.25

58

0.25

99

0.26

40

0.26

81

0.27

22

0.27

64

0.28

05

0.28

46

Graph 1. Simulated Default Probability of Agriculture

ijcrb.webs.com



136

MAY 2012

VOL 4, NO 1

0

50

100

150

200

250

300

350

0.0139

0.0308

0.0476

0.0645

0.0813

0.0981

0.1150

0.1318

0.1487

0.1655

0.1824

0.1992

0.2160

0.2329

0.2497

0.2666

0.2834

0.3002

0.3171

0.3339

0.3508

Graph 2. Simulated Default Probability of exports

ijcrb.webs.com



137

MAY 2012

VOL 4, NO 1

0

50

100

150

200

250

0.03

85

0.04

60

0.05

36

0.06

12

0.06

88

0.07

63

0.08

39

0.09

15

0.09

91

0.10

66

0.11

42

0.12

18

0.12

94

0.13

69

0.14

45

0.15

21

0.15

97

0.16

72

0.17

48

0.18

24

0.19

00

Graph 3. Simulated Default Probability of construction and housing

ijcrb.webs.com



138

MAY 2012

VOL 4, NO 1

0

50

100

150

200

250

0.0710

0.0821

0.0931

0.1041

0.1152

0.1262

0.1373

0.1483

0.1594

0.1704

0.1815

0.1925

0.2036

0.2146

0.2257

0.2367

0.2477

0.2588

0.2698

0.2809

0.2919

Graph 4. Simulated Default Probability of manufacturing and mining

ijcrb.webs.com



139

MAY 2012

VOL 4, NO 1

0

50

100

150

200

250

300

0.0221

0.0306

0.0392

0.0477

0.0562

0.0648

0.0733

0.0819

0.0904

0.0990

0.1075

0.1161

0.1246

0.1332

0.1417

0.1503

0.1588

0.1674

0.1759

0.1845

0.1930

Graph 5. Simulated Default Probability of trade services and miscellaneous

ijcrb.webs.com



140

MAY 2012

VOL 4, NO 1

Graph 6. Default Distribution of $100 Million Corporate Loan Portfolio

a credit risk model for bank’s loan portfolio optimize … · process in terms of a markov chain...

Documents