UNIVERSITY OF ECONOMICS AND LAW FINANCE AND BANKING INTERNSHIP REPORT The approach of Z-score model to analyze ACB’s risks Asia Commercial Bank Advisor: Ph.D LE TRUNG THANH Student:
Trang 1UNIVERSITY OF ECONOMICS AND LAW
FINANCE AND BANKING
INTERNSHIP REPORT
The approach of Z-score model to analyze ACB’s
risks (Asia Commercial Bank)
Advisor: Ph.D LE TRUNG THANH
Student: HUYNH THU AN
Class: K10404B
Student ID: K104040561
Ho Chi Minh City, April 7th, 2014
Trang 2Page 1 of 19
ACKNOWLEDGEMENT
An endeavor over period can be successful only with the advice and support of well-wisher
I take this opportunity to express my gratitude and appreciation to all those who encourage
me to complete my internship report
I am deeply indebted to PH.D LE TRUNG THANH who introduced me to the approach of quantitative finance, reviewed my mistakes as well as listened and responded all my
questions
I express my profound and sincerely thank to ASIA COMMERCIAL BANK- NGUYEN VAN TROI BRANCH which created a great condition to assist me in completing the report
I also extend thank to my best friends for their support and encouragement
I sincerely thank all!
Trang 3Page 1 of 19
Comment of Internship Organization
Ho Chi Minh City, April 7 th , 2014
Signature of Branch Director
Trang 4Page 1 of 19
Comment of Advisor
Ho Chi Minh City, April 7 th , 2014
Signature of Advisor
CONTENTS
Trang 5Page 1 of 19
Trang 6Page 6
ABSTRACT
In this study, ACB’s data are used in Vietnam in the period 2008-2012 and quantitative methods are adopted to determine the impact of specific factors on insolvency risk These factors are LLR -Loans loss reserve ratio, NIR- Net interest income ratio, LEV Leverage
Ratio, LDR- Loan to Deposit Ratios, and LAD – Liquidity assets to deposit ratios Using
NIR instead of NIM helps to complete the research in the past The study also confirmed the increase of equity as prerequisites to protect banks from insolvency risk
1 INTRODUCTION
At the moment, the financial crisis in 2008 still affects our economy, resulting in the sign of slow growth As a consequence, many companies were out of business Acting as veins in the economy, banks are also facing difficulties, which disbursement and unfreezing the capital flow to businesses The competition among banks is even more intense and banks are suffering many risks, such as liquidity risk, credit risk, interest risk, etc Thus, bank managers should pay more attention to a particular risk but also all risks affecting the operation of banks That leads to the decision whether a bank continues to operate or merge with another bank Within the past 3 years, several banks have lost the brand-name on the market forever which shows the view of the State Bank of Vietnam: "Mergers and consolidation are an indispensable trend to enhance the competitiveness of the bank." On December,15 th 2011, consolidation project of 3 banks: Sai Gon Commercial Bank , Tin Nghia Bank and Ficombank were approved by the General Meeting of Shareholders and named Saigon Commercial Bank ( SCB) Habubank officially took the name away when it was merged into Sai Gon - Ha Noi Bank ( SHB ) on August ,28 th 2012 On October, 4th , PVcomBank officially came into operation , meaning that the Western Bank lost an official name On November, 18th 2013, the State Bank of Vietnam decided to merge Dai A Bank with Ho Chi Minh Develop Bank (HD Bank)
Student: HUYNH THU AN - Advisor: LE TRUNG THANH
Trang 7Page 7
Asian Commercial Bank (ACB) is one of the group which was establish on June, 4th,
1993 Since December, 31st, 2012, ACB's capital has been 9,376,965,060,000 VND During more than 20 years of development, the bank has gained remarkable achievements: The best bank of Viet Nam in 2011,2012; The strongest bank in 2010 But over the last few years, the bank has been in trouble because of many competitors such as Techcombank, Sacombank, Eximbank, etc In addition, there are more unexpected information that impacts the internal development of the bank
Applying former research is very essential in assessing all activity risks of the ACB and comparing with G12 banks After this research, I can identify the basic factors affecting the insolvency risk of ACB, analyze and explain them based on the real condition From there, I compare and review the overall difference between ACB’s risk and G12’s risk This research will help bank managers have an objective and accurate view to plan strategies for sustainable development
2 LITERATURE REVIEW
A number of researchers have attempted to discriminate between financial characteristics of successful firms and those facing failure The objective has been developed a model that uses financial ratios to predict which firms have the greatest likelihood of becoming insolvent in the near future Altman is perhaps the best known of these researchers The Z-score model, commonly referred to as the Altman Z-Score, was developed by Professor Edward I, Altman in 1968 The Z-Score is constructed from six basic accounting values and one market-based value These seven values are combined into five ratios which are the pillars that comprise the Z-score The five pillars are combined to result in a company’s Z-score (Altman 2002) Z –score model of Altman has accurately estimated 66 % of bankrupt companies and 78 % of companies that did not go bankrupt for
a previous year period Thanks to the fairly accurate prediction of this model should only be used some popularity in many countries around the world However, this model does not indicate the expected time of bankruptcy because the bankruptcy of the business depends on the economic situation and the crisis
Student: HUYNH THU AN - Advisor: LE TRUNG THANH
Trang 8Page 8
Banks are a non-manufacturing sector so the application of Z-score model is somewhat different It is the study of Boyd and Graham (1986) Define bank insolvency as a state where (CAR+ROA) ≤ 0, with CAR the bank’s capital-asset ratio and ROA its return
on assets Then if ROA is a normal distributed random variable such as, Boyd and Graham (1986) noted that the probability of insolvency can be given as:
Where the Z–score is defined as >0 and is the cumulative distribution function (CDF)
of standard normal distribution N (0, 1)
According to Cihak (2008), the model that is used to show relationship between insolvency risk and other bank-specific risk is described:
(1)
where i = 1,…, N indexes banks, Xj, j = 1,…,J , denote macroeconomic variables which are identical across banks and affect all the banks in the same fashion through while Zik, k = 1,
…, K, denote bank-specific variables with corresponding pooled effects Lagged z is included into specification as an attempt to capture capital reserves built in previous period All the specifications are estimated applying the ordinary least squares with robust White errors Given a relatively large number of cross sections within each country in this sample we do not estimate the fixed effects specification and restrict the analysis to pooled intercepts only
3 METHODOLOGY
A regression model which based on following variables will be built below:
Zit = β0 + βi *Xit + eit
Zit: z score of bank i at time t, measured by
Xit: (i=1, 2, 5): stand for independent variables
Input:
X1: LLR - Loans loss reserve ratio
X2: LLP - Loan Loss Provision
Student: HUYNH THU AN - Advisor: LE TRUNG THANH
Trang 9Page 9
X3: LEV –Leverage Ratio
X4: NIR – Net interest income ratio
X5: LDR – Loan to Deposit Ratios
X6: LAD – Liquidity assets to deposit ratios
Output:
, ,,, , and their correlations with insolvency risk measured by Z-score
Then the detailed model will be presented by the equation below:
=
According to many researches, there are many variables which have a lot to do with insolvency risk Some of those factors are LLR, LLP, LEV, NIR, LDR, LAD that stand for credit risk, interest risk, and liquidity risk Credit risk which is related to asset portfolio represented by LLR variable (Loan Loss Reserve) Interest risk is shown by the NIR variable (Net Income Ratio) Both LDR and LAD from asset portfolio and capital reflect the liquidity supply and demand LEV represents leverage of bank LLP is bad debt expense, which can be combined with LLR to evaluate credit risk
LLR: Loans loss reserve ratio.
Whalen’s study (1988) shows that loan loss reserve to total loans ratio moves with insolvency risk The increase of bad debt makes loan loss reserve increase However, Halling’s study (2006) said that they are in inverse correlation with each other Banks which have good financial condition usually increase loan loss reserve Banks facing financial difficulties will reduce the amount of loan loss reserve to the lowest I expect that the correlation with Z-score will get -/+ and +/- with insolvency risk
LLP – Loan Loss Provision
According to Whalen (1988), loan loss provision to total assets has positive correlation with risks but it has no statistical significance Halling (2006) supposed that loan loss provision to income from an active business moves with risk However, due to changes in the process of regression, this ratio doesn’t have significance I expect that the correlation with Z-score will get - and + with insolvency risk
LEV – Leverage Ratio
From the research of Logan (2001), leverage ratio which means total loan to total equity (D/E) has negative correlation with bank insolvency risk The higher this leverage ratio is, the higher possibility of risk we can get In the developing financial system, he
Student: HUYNH THU AN - Advisor: LE TRUNG THANH
Trang 10Page 10
realized that weak banks must raise capital so that the shareholders continue to fund them According to Montgomery (2004), leverage ratio which means equity to total loans ratio (E/D) moves with insolvency risk However, it has no statistical significance The results of Jordan (2011) with leverage ratio are calculated by Tier 1 capital to total asset ratio This leverage ratio has inverse correlation to risk Therefore, I expect that the correlation with Z-score will get +/- and -/+ with insolvency risk
NIR – Net interest income ratio
Logan‘s research (2001) gives the empirical result that the net interest income to total assets moves with the risk of insolvency Halling (2006) has the same result as Logan While the results of Jordan (2011) show that non-interest income to interest income ratio moves with bank insolvency risk, or diversification of income makes banks riskier because those banks could not keep their traditional customers Therefore, I expect that the correlation with Z-score will get - and + with insolvency risk
LDR – Loan to Deposit Ratios
According to Montgomery Research (2004), total loan to total deposits ratio has positive correlation with insolvency risk This is because when banks face difficulties, they will concentrate on growing the credit to get more profit and they tend to lend with higher interest rates for the customers who are not qualified enough As a result, it causes bad debt
in banks Therefore, I expect that the correlation with Z-score will get - and + with insolvency risk
LAD – Liquidity assets to deposit ratios
Montgomery’s research (2004) shows that the ratios between liquid assets and total deposit have negative relationship with insolvency risk If banks had more liquidity assets, they could reduce liquidity risk Therefore, I expect that the correlation with Z-score will get + and - with insolvency risk
This is summarized in the following Table 1:
Table 3.1: The summarization of risky variables and their relation to insolvency risk
Student: HUYNH THU AN - Advisor: LE TRUNG THANH
Trang 11Page 11
To
start the regression, I use a quantitative research model which is called the multi- variable
Student: HUYNH THU AN - Advisor: LE TRUNG THANH
Original
number Variable Formula Researc-hers Relation-ship
with Z-core
Relation -ship with Insolvency risk
1 LLR - Loans loss
Halling (2006)
-+
+
-2 LLP - Loan Loss
Provision
Whalen (1988)
Halling (2006)
Leverage
Ratio
Logan (2001
Montgo mery (2004)
+
-+
interest income
ratio
Logan (2001) Halling (2006)
5 LDR – Loan to
(2004)
6 LAD – Liquidity
assets to deposit
ratios
Montgo mery (2004)
Trang 12-Page 12
regression equation It is based on Marco‘s study (2004), Cihak (2008) and Whalen’s research (1988) in which the dependent variable and the independent variable will be qualified From empirical studies results, variables which affect the insolvency risk such as instability, bankruptcy that will be prior The basic rule is to keep the essence of variable and formula unchanged Unsuitable variables for Viet Nam‘s condition will be adjusted due
to scientific arguments The advantages and disadvantages of this regression will be analyzed to clarify how the specific risks affect on insolvency risk
4 RESULT
Table 5.1: Result of ACB’s model 1
Residual standard error: 2.137
Multiple R-squared: 0.7718, Adjusted R-squared: 0.5437
F-statistic: 3.383 , P-value: 0.08184
For the table (*) and (**) denote significance at 10% and 1% respectively.
Result:
=
The following steps will be presented to test the hypothesis about model 1:
The null and alternative hypotheses are:
Student: HUYNH THU AN - Advisor: LE TRUNG THANH
Trang 13Page 13
Assuming that variables X and Y are normally distributed, I will use the t distribution
to perform this test about the linear correlation coefficient
According to the result of model 1, P (F – statistic) < 0.1 H0 is denied, therefore, ACB’s model 1 is accepted
With the same method, the statistical significance of the independent variables in model 1 will be tested with =0.1, I have:
• LLR: The coefficient is 8788, so for every unit increase in LLR, a unit decrease
in Z-score is predicted, holding all other variables constant Besides, P-value = 0.0176 < 0.1 => the coefficient for LLR is statistically significant
• LLP: The coefficient is so for every unit increase in LLP a unit increase in
Z-score is predicted, holding all other variables constant Besides, P-value = 0.6207 >
0.1 => the coefficient for LLP is not statistically significant.
• LEV: The coefficient is so for every unit increase in LEV a unit increase in
Z-score is predicted, holding all other variables constant Besides, P-value=0.0762 < 0
1 => the coefficient for LEV is statistically significant
• NIR: The coefficient is so for every unit increase in NIR a unit increase in
Z-Score is predicted, holding all other variables constant Besides, P-value=0.0691> 0
1 => the coefficient for NIR is statistically significant
• LDR: The coefficient is so for every unit increase in LRD a unit decrease in
Z-Score is predicted, holding all other variables constant Besides, P-value = 0.7843>
0.1 => the coefficient for LDR is not statistically significant.
• LAD: The coefficient is so for every unit increase in LAD a unit decrease in
Z-score is predicted, holding all other variables constant Besides, P-value = 0.9163>
0.1 => the coefficient for LDR is not statistically significant.
After the results of model 1, unsuitable variables such as LLP, LDR, LAD will be eliminated as table below:
Table 5.2: Result of ACB’s model 2
Student: HUYNH THU AN - Advisor: LE TRUNG THANH