Determinants of RiskTaking of Vietnamese Banks after Global Financial Crisis45279

Determinants of Risk-Taking of Vietnamese Banks after Global Financial Crisis Tu T.T Tran (1)*, Yen T Nguyen (2), Dick Beason (3) VNU University of Economics and Business, Vietnam National University, Hanoi, Vietnam University of Vinh, Nghe An, Vietnam (3) Alberta University, Alberta, Canada * Corespondence: tuttt@vnu.edu.vn (1) (2) Abstract: The paper deals with hypotheses evaluating the impact of factors on the risk-taking of the commercial bank in Vietnam in a 10-year period, starting from 2008 Using the OLS regression method for analysis by running Eviews and ANOVA test in SPSS, it was found that: Bank performance plays an important role in reducing risks of the Vietnamese bank; The corporations led by female faces with higher risk level than led by male; The bail-out activities of the State Bank of Vietnam in 2015 does not influence on risk-taking of the bank system; The merge and acquisition does not support the bank to reduce risk, it increases the risk for acquiring banks; The global crisis 2008 exerts dire consequence on the bank system in Vietnam; There is the difference of risk-taking among the groups of the bank having a different number of years of operation Basing on this result, the paper also makes recommendations to Government, The State Bank of Vietnam and the commercial banks for effective risk management Keywords: Commercial bank; risk management; risk-taking Introduction The term “Risk” stems from the Italian word “risco” with its meaning of danger Risks of banks are unintended losses that cause economic losses, increased costs, reduced incomes and profits, relative to the original estimate Banks also deal with risks from the micro and macro environment Besides, being a special type of business, banks encounter specific risks created by the nature of financial industry such as credit risk, interest rate risk, liquidity risk, risks from off-balance-sheet activities, etc Risks and expected profits are two co-variables In other words, bank risk is a part of the banking business, it is unavoidable that risks must be accepted, managed and quantified so that banks can balance between expanding the scale, increasing profit and ensuring banking safety That is, risk management in the banking business and identification of determinants of bank risk-taking is an initial step of risk management The financial crisis of 2008 has a strong influence on Vietnam The government implemented many policies to prevent the impact of this crisis One of the remedies was the restructuring of banks The last decade has witnessed many breakthroughs in the restructuring of banks in Vietnam Restructuring of financial institutions was not only a solution to overcome the crisis but also an evitable trend to enhance sustainable stability of the financial system (Tran & Truong, 2014) Project 254 relating to “Restructuring the system of credit institutions in the period of 2011 – 2015” was considered as a milestone in marking the Vietnamese government’s efforts to address this problem Following 2011, many useful policies were executed by the State Bank of Vietnam (SBV) One of the initial gunshots of the restructuring process in Vietnam was a mass of merger and acquisition activities (M&A) among many Vietnam banks, namely a consolidation of three banks (SCB, Ficombank and TinnghiaBank) and a merger of Habubank into SHB in 2011; two M&A deals: the merger of WesternBank and PVFC into Mass Commercial Joint Stock Bank (PVcomBank) and the deal of DaiABank and HD Bank merged into HD Bank in 2013 Not only that, three M&A transactions were conducted in 2015 Besides, the SBV undertook the mandatory acquisition of three weak banks – Vietnam Construction Bank, Ocean Bank and GP Bank – at zero VND years ago That was the first time in which the SBV conducted restructuring of the commercial bank by weak bank bail-out solution According to the SBV, this solution has had immediate positive effects on the market's sentiment The depositor has stopped withdrawing money and returned to deposit money, which plays a significant role in addressing the liquidity risk of the three banks (Thai, 2018) There is little research studying factors affecting banks ‘overall risk Moreover, there has been no quantitative research studying the effect of restructuring measures on the risk of the commercial bank in Vietnam Therefore, based upon these two research gaps, this paper will construct models to examine determinants of risk-taking in Vietnamese commercial banks in the period of from 2008 to 2018 and especially, to consider whether these actions of the SBVs to restructure the Vietnamese financial system affect commercial banks' risk-taking The remainder of this paper consists of four parts as follows The second session reviews researches relating to the bank’s risk and factors contributing to bank risk-taking The third section proposes the methodology to build models to address alternative hypotheses The fourth session presents the research results and discussions The final session proposes some recommendations to improve the stability of Vietnamese commercial banks in particular and the Vietnamese financial system in general Methodology 2.1 Hypothesis development Khan et al (2015) studied the relationship between funding liquidity and bank risktaking Using a dataset of US bank from 1986 to 2014 in quarterly intervals, the authors concluded that banks with higher ROA, capital buffers take the less overall risk, compared to smaller capital buffer banks because shareholders would lose out in the case of default They showed that banks having higher levels of deposits take less risk in contrast to the banks with lower levels of deposits OLS regression results showed that the rise of GDP, the unemployment rate and changes in the house price index impacts positively on a decrease of banks’risk-taking (Khan, et al., 2017) From the research, therefore, we can identify our first hypothesis with regard to the Vietnamese banking system: H1: Performance of Vietnamese commercial bank has a negative impact on risktaking Skala and Weill (2018) studied the impact of CEO gender for bank risk with a dataset of 365 Polish banks They explored an interesting relationship between CEO gender and bank risk They found that banks led by female CEOs are less risky due to higher capital adequacy and equity to asset ratios (Skata & Weill, 2018) Consistent with the study, (Bellucci, et al., 2010) concluded that female CEO’s are more risk-averse and less selfconfident while male CEOs usually have more venture projects Consistent with Skala and Weill (2018), we hypothesize that: H2: The female-led banks exert a negative impact on risk-taking of commercial bank in Vietnam John Sedunov et al (2016) studied whether bank bail-outs have a positive or negative impact on systemic risk The result showed that TARP played a significant role in reducing the risk of banks Contrary to this viewpoint, (Davila & Walther, 2017) concluded that bank bail-outs have negative consequences on the bank's financial status because it creates the wrong motives for internal risk management and breeds moral hazard The larger banks have assumed that their actions influence directly on the government’s bailout response Therefore, they tend to use a larger leverage ratio than the smaller banks In 2015, The State Bank of Vietnam bought three weak banks without compensation In sum, following (Sedunov, et al., 2016), we hypothesize that: H3: Bail-out activity by the State Bank of Vietnam exerts a positive impact on the security of commercial bank in Vietnam Using a sample of 714 agreements involving EU acquirers and objectives located throughout the world over a 14-year period, starting from 1991, some researchers found that M&A operations cause a slight decline in return on equity, cash flow return and profit competence and contemporaneously a marked enhancement in cost efficiency (Beccalli & Frantz, 2009) Studying European commercial banks from 1997 to 2005, Duangkamol Prompitak (2009) assumed that the merged banks are likely to encounter with less risk than the no-merged banks Both papers showed that M&A activity plays a significant role in reducing the lending rate, which is beneficial to the customers (Prompitak, 2009) Conversely, Jens Hagendorff et al (2012) concluded an opposite viewpoint that M&A actions have a trivial effect on security and soundness indicators of the European banks (Hagendorff, et al., 2012) Base on (Beccalli & Frantz, 2009) and (Prompitak, 2009) and the authors’ practical observation, we thus propose the following hypothesis: H4: M&A activity play a significant role in enhancing the stability of the banking system in Vietnam There are many papers studying determinants influencing to risk-taking of banks over the world Vietnam is no exception The studies in Vietnam, however, focus primarily on studying the factors affecting each particular type of risk, for instance, papers about factors impacting on the bank’s liquidity risk such as (Truong, 2013), (Vu, 2015) or (Mai & Bui, 2018); the bank’s credit risk such as (Vo & Bui, 2014), (Phan & Nguyen, 2017), etc Very few studies have considered impacts these elements on the overall risk of banks There is only the paper by (Le, 2016) studying about factors affecting the overall risk of 27 Vietnamese commercial banks from 2005 to 2014 Le (2016) provides empirical evidence showing that the higher rate of operating cost, the rate of asset growth, debt and profit are, the more risk that the bank encounters By contrast, bank size and equity ratio have a negative correlation with bank risk-taking Other macroeconomic variables, namely economic growth and inflation rate have adverse impacts on the bank’s stability However, this paper did not examine whether weak bank bail-out and M&A activities have an influence on bank risk Based on this gap, we build a model to evaluate the effects of these determinants on the overall risk of the Vietnamese commercial bank system, which is one of the highlights of this report Moreover, the 2008 crisis affected seriously on economic in Vietnam At the end of the fiscal year of 2008, while many commercial banks in the US and Europe suffered big losses, most commercial banks in Vietnam made thousands billion VND profit (Ton, 2009) The commercial bank was the opposite: 2008 witnessed strong growth, ROA reached approximately 11.8%; however falling to 10.8% in 2009 and about 10% in 2010 (Phuong Dong, 2011) The bad debt ratio in 2009 and the following years increased significantly According to the Vietnamese accounting standard, the bad debt of the whole banking system in 2008 accounted for about 3.5% By contrast, using international accounting standards for the classification of bad debts, the figure was up to 8% and roughly 11% in 2010 and rose another 2% in 2011 (Nguyen, et al., 2012) Therefore, the authors question whether the crisis in 2008 had a late impact on commercial banks in Vietnam and will examine the lag of this influence We hypothesize that: H5: The 2008 crisis had a lagged influence on risk of Vietnamese commercial banks 2.2 Empirical model 2.2.1 For Hypothesis Testing the impact of ROA on risk-taking: RISKit = β0 + β1CAPit+ β2DEPit + β3LEVit + β4STAit + β5SIZEit + β6GDPit + β7INFit + β8LTRit + β9ROAit + εit Testing the impact of ROE on risk-taking: RISKit = β0 + β1CAPit+ β2DEPit + β3LEVit + β4STAit + β5SIZEit + β6GDPit + β7INFit + β8LTRit + β9ROEit + εit 2.2.2 For Hypothesis 2, 3, 4, For Hypothesis 2: RISKit = β0 + β1CAPit+ β2DEPit + β3LEVit + β4STAit + β5SIZEit + β6GDPit + β7INFit + β8LTRit + β9CEOit + εit For Hypothesis 3: RISKit = β0 + β1CAPit+ β2DEPit + β3LEVit + β4STAit + β5SIZEit + β6GDPit + β7INFit + β8LTRit + β9BOBit + εit For Hypothesis 4: RISKit = β0 + β1CAPit+ β2DEPit + β3LEVit + β4STAit + β5SIZEit + β6GDPit + β7INFit + β8LTRit + β9M&Ait + εit For Hypothesis 5: RISKit = β0 + β1CAPit+ β2DEPit + β3LEVit + β4STAit + β5SIZEit + β6GDPit + β7INFit + β8LTRit + β9CRISISit + εit Where: RISKit is independent variable indicating overall risk for bank i on year t CAPit is the capital buffers for bank i at the time t DEPit is the deposits for bank i at the time t LEVit is the levarage ratio for bank i at the time t STAit is the public or private status for bank i at the time t SIZEit is the size for bank i at the time t CEOit is the gender of CEO for bank i at the time t M&Ait is the merger and acquistion situation for bank i at the time t BOBit is the status of banks which are received bailout assistance from State Bank of Vietnam GDPit is the gross domestic product of Vietnam on the year t INFit is the overall inflation rate in Vietnam on the year t LTRit is the long-term lending interest rate on the year t CRISIS is the financial crisis in 2008 ROAit, is the net income on total asset on the year t ROEit is the net income on total equity on the year t εit is a random error term Not only that, one of this paper’s outstanding point is the usage of a control variable, namely NYO which is an abbreviation of the number of years of operation of bank i at the time t The number of operating years of Vietnam commercial banks is classified into specific groups, namely operating for less than 10 years, 10-20 years, 20-30 years, 30-40 years, 40-50 years and over 50 years A dummy variable is used representing these groups By using ANOVA testing, we examine whether there are differences in the number of years of operation to the bank risk-taking H6: There is a difference of risk-taking status among the bank groups having different operating years The dependent variable Z-score has been increasingly utilized in recent literature Drakos et al (2016) used two different measures to evaluate the risk-taking of banks, namely risk assets and Z-score when conducting a research on bank risk in 10 CEE economies and Russian Federation in the 14 years, starting from 1997 The former aim to express the risk of a commercial bank from current activities while the latter conveys not only the current bank risk-taking behaviour but also the risk accumulated from the past period’s bank operations The lower this value is, the higher the probability of default is and as a result, the higher the bank insolvency risk is (Drakos, et al., 2016) These variables were also used by Khan et al (2017) The ratio of risk-weighted assets to total assets is a proxy to measure the quality of the asset and reflect credit risk while Z-score was used to measure the overall risk of banks (Khan, et al., 2017) Skala and Weill (2017) utilized variables as dependent variables to examine impact of CEO gender on risk of 365 Polish commercial banks, namely the total capital adequacy ratio, equity to asset, Z-score, the ratio of non-performing loans to total loans (NPL) and the ratio of loan loss provisions to total loans (LLP) Z-score based on ROA indicates the bank’s stability status Both NPL and LLP show credit risk and reflect the asset quality of banks (Skata & Weill, 2018) Consistent with the researches of Drakos and et al (2016), Khan and et al (2017), Skala and Weil (2017), we used four indexes, namely two Zscore indexes calculated by ROA and ROE which has been the standard in the recent literature for measuring overall risk of banks; the ratio of non-performing loans to total loans (NPL) and the ratio of loan loss provisions to total loans (LLP) Then, we evaluated which model has the highest explanatory power The indexes are computed as (1), (2), (3), (4) The higher Z-score value, the greater bank stability In other words, lower Z-scores imply higher bank risk By contrast, the higher NPL and LLP are, the higher the bank risk-taking is In other words, an increase in the NPL and LLP implies higher bank risk The independent variables Based on the literature, the report has used 13 variables which are expected to influence bank risk-taking, of which are interbank factors and the remaining are macroeconomic indicators In addition to the variables used in previous studies such as size, capital buffer, deposit, the leverage ratio, CEO gender, ownership status, bank performance, GDP, inflation, long-term interest rate; the authors have added new variables, namely CRISIS, M&A and the bank bailout Bank specific factors Capital buffers (CAP): The previous studies assumed that banks with higher capital buffers take the less overall risk, compared to smaller capital buffer banks because shareholders will lose out in the case of default (Khan, et al., 2017), (Bonstandzic & Weib, 2018) We expect a negative correlation between the capital buffer and bank risk We define the capital buffers according to formula (5) Deposits (DEP): It is proved that having higher levels of deposits take less risk, compared to the banks with lower levels of deposits This is explained as banks with higher deposits are less likely to face a funding shortfall immediately and bank manager’s aggressive risk-taking behaviour is less likely to be audited (Khan, et al., 2017) We expect this variable to have an unfavourable relationship between it and the risk-taking of banks The leverage ratio (LEV): The bank with higher leverage tends to more vulnerable and likely to fail when faced with a crisis (Vazquez & Federico, 2015) The leverage ratio is calculated by the ratio of total liability to total assets Using a high level of liability exerts both positive and negative impacts on the bank’s performance If it is used effectively and properly, it will bring advantages and promote profit If not, it is likely to cause an increase in losses, which breeds a decline of gains, even risk of bankruptcy We expect that the higher the leverage ratio is, the higher the risk-taking of the bank Ownership structure (STA): There are a variety of reasons why bank ownership structure plays an important role in risk-taking of commercial banks First, the state-owned banks listed has a tendency to access information rapidly Therefore, the reactions to regulatory policy changes are likely to be faster for public banks, rather than private banks Another reason is that public bank, unlike other banks, is subject to stronger market discipline and more stringent regulatory requirements Moreover, the publicly traded banks engage in less risky activities than their private counterparts (Anis Samet and et al, 2018) According to these arguments and application in Vietnam, we believe that exposure to risk in public banks is less than that in private banks Bank performance: ROA and ROE represents bank performance They are the popular ratios measured by net income on total assets and net income on equity respectively ROA, ROE represents the level of efficiency when using the total funds and equity respectively In other words, it reflects how much interest the company earns on dong of asset and equity respectively The higher these figures are, the better the ability to use assets and equity are CEO gender (CEO): The larger gender difference is the higher risk aversion of women (Powell & Ansic, 1997) Other gender differences, such as in management styles, personality traits, etc of CEOs have also effected on making decisions, which can have a detrimental or positive impact on bank risk Based on the studies of (Beccalli & Frantz, 2009) and (Skata & Weill, 2018), we suppose that banks led by female CEOs are likely to undertake less risky, compared to banks led by male Size of bank (SIZE): The previous studies have different priors about the relationship between bank size and bank risk In line with the too-big-to-fail hypothesis, many studies such as (Aspachs, et al., 2005), (Valla, et al., 2006), (Vodova, 2012) suggested that the greater total asset, the less risk-taking of a bank Large-scale banks may have many advantages It is easier in raising capital from customers, lending on the interbank market or receiving support from the central bank as lender of last resort However, (Bunda & Desquilbet, 2008) gave contradictory outcomes The large-scale commercial banks typically enjoy implicit advantages with low cost for mobilizing capital They often invest boldly in many venture projects Hence, the risk of large-scale banks is also higher compared to smaller-scale banks In this model, the authors expect a positive relationship between the size of the bank and the bank’s stability Macroeconomic factors Economic growth (GDP): (Haas & Lelyveld, 2008) concluded that GDP growth exerts a negative influence on subsidiaries’ credit growth in the host country, seemingly indicating a portfolio adjustment toward higher return opportunities According to one of the most basic tenets of investing, that high risk, high return, it is likely to increase risk-taking by banks Some have argued that GDP growth should have a significant effect on the quality of credit Therefore, this should imply a lower risk density In this paper, when considering the current situation in Vietnam, the authors expect a negative correlation between economic growth and risk of banks Inflation (INF): Perry (1992) suggested that the effects of inflation on the bank’s expectations about inflation in the near future If inflation is expected to rise, banks will adjust interest rates to increase interest income faster than the speed of interest expense However, the banks face uncertainty in predicting inflation It not only increases costs, reduces the bank’s net profit but also creates difficulties in mobilizing funds Based on the studies of (Bunda & Desquilbet, 2008), (Vodova, 2012), (Fola, 2015), the writers expect that this relationship is positively correlated Long-term lending interest rate (LTR): The higher the long-term lending interest rate, the more profit the banks earn As an evitable consequence, they will reduce the liquidity reserve for supporting long-term loans, which increases the risk for banks These previous researches such as (Valla, et al., 2006), (Bunda & Desquilbet, 2008), (Vodova, 2012) suggested that the lending rate is inversely correlated with the liquidity of banks In this paper, we anticipate that an increase of long-term lending interest rate causes exposure to risk for banks Merger & Aquisition (M&A) is represented by a dummy variable It equals to if that year witnessed M&A activity of the Vietnamese bank with the number of shares from 25% or more – the threshold to become shareholders entitled to certain influence right such as the right to request a meeting of the Members' Council to resolve matters under their authority according to Law of Enterprises No.68/2014/QH13 The dummy variable is equal to otherwise The research will test the relationship between this variable and the risktaking of the commercial bank in Vietnam Bail-out bank (BOB) is a dummy which equals to if that year witnessed a bail-out from the Center Bank of Vietnam and otherwise This report will examine changes in risktaking behaviour of the bank before and after this rescue event CRISIS is a dummy which represents for the financial crisis in 2008 (indicator variable with for the 2008 crisis and otherwise) The authors utilize this variable to test the lagged impact of the 2008 crisis on the commercial banks in Vietnam 2.2 Data and sample Our analysis is based on yearly bank-level data from balance sheet database of commercial banks in Vietnam with data provided by the FiinGroup company which is the leading firm in financial and business information in Vietnam All data used in this research published by the commercial banks, audited by external auditors and supplied by the most prestige financial information corporation The banks included in the research make up over 85% of a number of banks in the Vietnam banking system After collecting process, we choose 216 observations on 31 banks from 2008 to 2018, which ensures a good sample needed for regression models Z-score (ROA) = ( Z-score (ROE) = ) ( ) = ( = ) ( ) (1) (2) With: ROA is equal to earnings before taxes and loan loss provision divided by total assets ROE is equal to earnings before taxes and loan loss provision divided by total equity CAR is equal to equity divided by total assets ∂ (ROA) and ∂ (ROE) are the standard deviation of ROA and ROE respectively NPL = (3) Within: Bad debt is the total of debts classified from Group to Group according to the Decision 493/2005 / QD-NHNN dated April 22, 2005 of the State Bank of Vietnam LLP = (4) Loan loss provision is the total loss provision of five debt groups according to the Decision 493/2005 /QD-NHNN dated April 22, 2005, of the State Bank of Vietnam = ∗ (5) Within: Risk-weighted asset is sum of all assets from group to group according to Decision 493/2005/QĐ-NHNN Results The paper runs the model to test the proposed hypotheses by using Eview 10 according to ordinary least squares (OLS) Our results follow: Results of the hypotheses Results of Hypothesis Result in Table shows that the eight regressions are statistically significant at 1% The six models with dependent variables Z-score(ROA), LLP and NPL experience the impact of ROA and ROE However, the results are contradictory In model I and II, ROA and ROE have a negative relationship with Zscore (ROA) and Zscore (ROE) On the other hand, the dependent variable and risk of the bank have an adverse relationship Subsequently, ROA and ROE have a positive correlation with the bank risk-taking In other words, the higher the return on asset or the return on equity is, these banks face higher risk, which is contrary to our expectation Dep.Var Constant CAP DEP LEV STA SIZE GDP INF LTR ROA ROE Prob (F-Statistic) R-squared No.obs Table 1: Regression results of the eight regression models – coefficient for Hypothesis Hypothesis Zscore(ROA) Zscore(ROE) LLP Model I Model II Model III Model IV Model V Model VI 387.1369*** 154.2336 54.95332*** 55.13667*** -0.004771 -0.018400 -6.42E-13 -6.63E-13 -2.04E-14 -2.04E-14 1.48E-16*** 1.47E-16*** -133.8217*** -137.3254*** -10.79865*** -10.77743*** 0.000724 0.000345 -22.18300 155.2705 -0.767932 -0.763885 -0.005899 0.003129 71.51527*** 71.70089*** 7.064767*** 7.072412*** 0.004124** 0.004061** -7.402931 -4.828519 -1.348635*** -1.356594*** 0.000537 0.000744 6.73E-15 6.07E-15 7.07E-16** 7.11E-16** -5.16E-19 -5.82E-19 -10.87503 24.51968 2.657956 2.487944 -0.024621 -0.021209 -5.932437 -44.04619 -9.270030 -9.016918 0.071740*** 0.067406** -2390.635*** 0.308481 -0.125056* -246.6817*** 0.223752 -0.014714** 0.000000 0.000000 0.000000 0.000000 0.000000 0.000000 0.251316 0.261637 0.352593 0.352604 0.292209 0.297571 216 216 216 216 216 216 Note: Statistically significant at * 10%, ** 5%, *** 1% NPL Model VII 0.137271*** 5.20E-16*** 0.009512 -0.028001 -0.002794 -0.003675** 8.18E-19 -0.072880* 0.206506*** -0.417722** 0.000000 0.260417 216 Model VIII 0.101257*** 5.17E-16*** 0.009534 0.003829 -0.002527 -0.003458** 8.07E-19 -0.071747* 0.207426*** -0.037243** 0.000000 0.240177 216 The remaining models give the opposite result With the models whose the dependent variables are NPL and LLP, ROA and ROE have a negative correlation with NPL and LLP which has a positive relationship with the bank risk-taking Therefore, ROA and ROE have negative impacts on the risk of the bank system In other words, the higher the bank performance is, the higher the stability of banks is This result is consistent with the authors’ prediction This result is consistent with the authors’ prediction It can be understandable because of the difference in the dependent variables representing the risktaking of the commercial bank in Vietnam We carry out the verification by Akaike (AIC) and Schwarz (SC) information standard The smaller the AIC and SC values, the better the model is We find that Model VI has the smallest AIC and SC values, which proves that this model is stronger than the other models Model VI explains 29.75% of the change of the dependent variable The adjusted R2 and R2 results show the strength of this model compared to other models, so we have a basis to choose model VI for the best model to evaluate the effects of bank performance After using the Wald Test (Table B1.2), P-value (F-statistic) > 0.05, we can omit variables, namely C, DEP, LEV, GDP and INF and have a new regression model (Table B1.3) Test for multicollinearity We use the Variance Inflation Factor to test for multicollinearity, the result is shown in Table B1.4 VIFs of LTR and SIZE are more than 10 Therefore, multicollinearity is a serious problem of the model LTR is removed and the test result is re-estimated VIF of all variables of the new model is low (Table B1.5), which proves that the multicollinearity error has been successfully addressed Test for Heteroskedasticity The White test is used to assess whether the model has a heteroskedasticity problem or not Basing on the result table B1.6, we see that P-value (F-statistic) > 0.05, which implies that Heteroskedasticity test is not a problem of this model Test for autocorrelation Serial Correlation LM is run to test for autocorrelation of this model (Table B1.7) We see that Prob (F-statistic) < 0.05, which suggest that autocorrelation is an issue After controlling AR(1) in the model and testing the new model, the authors recognize that Durbin-Watson statistic is equal to 1.902153 which belongs to a range from to It shows that the model solved the autocorrelation error After addressing all diagnostics, we have the best model as Table In conclusion, bank performance has a negative correlation with LLP having a positive relationship with the Vietnamese bank’s risk-taking Consequently, the higher return on equity of the bank is, the lower the bank risk-taking is Table 2: The final regession model result evaluating effect of bank performace on stability of the Vietnamese bank system Dependent Variable: LLP Method: ARMA Maximum Likelihood (BFGS) Date: 05/27/19 Time: 15:11 Sample: 216 Included observations: 216 Convergence achieved after iterations Coefficient covariance computed using outer product of gradients Variable Coefficient Std Error t-Statistic Prob CAP 2.61E-16 2.44E-17 10.66679 0.0000 STA 0.004769 0.001195 3.991178 0.0001 SIZE 0.000421 3.06E-05 13.75228 0.0000 ROE -0.016197 0.005064 -3.198277 0.0016 AR(1) 0.565928 0.040442 13.99352 0.0000 SIGMASQ 2.50E-05 1.90E-06 13.18944 0.0000 R-squared 0.467381 Mean dependent var 0.013917 Adjusted R-squared 0.454700 S.D dependent var 0.006870 S.E of regression 0.005073 Akaike info criterion -7.700665 Sum squared resid 0.005404 Schwarz criterion -7.606907 Log likelihood 837.6718 Hannan-Quinn criter -7.662787 Durbin-Watson stat 1.902153 Inverted AR Roots 57 3.1.2 Results of Hypothesis The result table shows that the four models’ Pros (F-statistic) are lower than 0.01, which proves that the models are statistically significant at 1% When Zscore (ROE) is represented for risk-taking of the bank system, the CEO gender impacts on the stability of the bank with statistically significant at 5% We show that Female CEO is significantly negative with Zscore(ROE) Thus, it supports the view that banks led by female directors are associated with higher insolvency risk Table 3: Regression result – coefficient for Hypothesis Dep.Var Hypothesis Zscore(ROA) Zscore(ROE) LLP NPL Constant 380.7115*** 58.75997*** -0.003757 0.129186*** CAP -6.63E-13 -1.99E-14 1.47E-16*** 5.16E-16*** DEP -105.1040*** -10.08804*** 0.002480 0.013224 LEV 201.0426* 0.643038 0.006288 0.008370 STA 81.82671*** 7.098141*** 0.004675** -0.001056 SIZE -16.00186*** -1.524544*** 2.46E-05 -0.004854*** GDP 1.08E-14** 7.49E-16** -2.89E-19 1.45E-18 INF -181.4051 3.549258 -0.033233* -0.104267*** LTR 279.9089 -8.668417 0.086919*** 0.255284*** Female CEO -11.82394 -1.526702** -0.001161 0.000730 Prob (F-Statistic) 0.000001 0.000000 0.000000 0.000000 R-squared 0.196034 0.365125 0.284377 0.237153 No.obs 216 216 216 216 Note: Statistically significant at * 10%, ** 5%, *** 1% Using the Wald Test, the authors see that four variables namely CAP, LEV, INF, LTR are not statistically significant because P-value of F-statistic = 0.9290 > 0.05 (Table B2.1) Therefore, we can eliminate these variables and have a new model (Table B2.2) Test for multicollinearity We see that the VIFs of the dependent variables are lower than 10, which implies that multicollinearity is not the issue of this model (Table B2.3) Test for Heteroskedasticity The White test shows that P-value (F-statistic) = 0.0000 < 0.05, which implies that the model has Heteroskedasticity problem (Table B2.4) By using the weighting method according to the opinion of Breusch & Pagan (1979), we address this problem with the new model (Table B2.5) After examining again by the White test, we see that Prob (F-statistic) = 0.2549 > 0.05, which points out that the model is fixed the Heteroskedasticity error (Table B2.6) Test for autocorrelation The model has an autocorrelation problem because Prob (F-statistic) = 0.0000 < 0.05 (Table B2.7) After making good the model ‘s problem, the model shows that Durbin-Watson statistic = 1.997853 which ranges from to It implies that the model solved the autocorrelation error The best regression model result testing impact of CEO on risk-taking of the commercial bank in Vietnam is shown in Table The female CEO variable has a negative correlation with the dependent variable (ZscoreROE) Moreover, the dependent variable has an inverse relationship with the risk-taking of the bank system Consequently, the banks led by female directors are associated with higher insolvency risk This result is opposite to the authors’ hypothesis, the previous researches of Bellucci et al (2010) and Skata & Weill (2018) However, considering the Vietnamese working environment, the authors realize that this result is reasonable The Vietnamese businesses run by women are usually less financial efficiency than the led-male corporation because the operator of a woman often has few negotiation experience and skill as well as time fund for work Table 4: The final regession model result testing impact of CEO on risk-taking of the commercial bank in Vietnam Dependent Variable: ZSCOREROE Method: ARMA Maximum Likelihood (OPG - BHHH) Date: 05/28/19 Time: 13:37 Sample: 216 Included observations: 216 Convergence achieved after 22 iterations Coefficient covariance computed using outer product of gradients Variable Coefficient Std Error t-Statistic Prob C 65.49182 6.912546 9.474341 0.0000 DEP -8.847122 1.597531 -5.537998 0.0000 STA 5.709724 0.542677 10.52141 0.0000 SIZE -1.767835 0.224068 -7.889715 0.0000 GDP 7.40E-16 1.91E-16 3.880727 0.0001 CEO -1.202541 0.567028 -2.120779 0.0351 AR(1) 0.585115 0.065317 8.958134 0.0000 SIGMASQ 6.502726 0.430521 15.10433 0.0000 R-squared 0.572225 Mean dependent var 5.949800 Adjusted R-squared 0.557828 S.D dependent var 3.907937 S.E of regression 2.598621 Akaike info criterion 4.786113 Sum squared resid 1404.589 Schwarz criterion 4.911123 Log likelihood -508.9002 Hannan-Quinn criter 4.836617 F-statistic 39.74808 Durbin-Watson stat 1.997853 Prob(F-statistic) 0.000000 Inverted AR Roots 59 3.1.3 Results of Hypothesis Table shows that the four models are significant at 1% However, with a significant level of 5%, the BOB variable is not statistical meaning Using the Wald test to examine the importance of the BOB variable, we see that Pro (F-statistic) = 0.6307 > 0.05 Hence, we can eliminate this variable Table 5: Regression result – coefficient for Hypothesis Dep.Var Hypothesis Zscore(ROA) Zscore(ROE) LLP NPL Constant 353.2177*** 55.20087*** -0.007743 0.129186*** CAP -6.58E-13 -1.94E-14 1.42E-16*** 5.16E-16*** DEP -111.1377*** -10.86475*** 0.002222 0.013224* LEV 190.1251* -0.767661 0.005070 0.008370 STA 81.44233*** 7.049050*** 0.004714** -0.001056 SIZE -14.70669*** -1.356965*** 0.000201 -0.004854*** GDP 1.06E-14** 7.19E-16** -3.77E-19 1.45E-18 INF -177.6829 3.981138 -0.039757** -0.104267*** LTR 268.4987 -10.11153 0.090065*** 0.255284*** BOB 3.206137 0.399024 -0.001799 0.000730 Prob (F-Statistic) 0.000001 0.000000 0.000000 0.000000 R-squared 0.192407 0.353320 0.286821 0.242841 No.obs 216 216 216 216 In other words, the bail-out activity of the Vietnamese state bank does not affect the risk of the bank system This contrasts with the authors’ expectation and the results of Lammert Jan Dam and Michael Koetter (2012) and John Sedunov et al (2016) They concluded that the bail-out activity had a negative or positive impact on the bank's risk when studying in German and United State respectively However, the result is incorrect when applied to the case of Vietnam 3.1.4 Results of Hypothesis We can see that the four models in Table are statistically significant With a significant level of 10%, the M&A variable is statistical meaning at model IV In other words, the M&A activity of the Vietnamese bank has a negligible influence on the risk of the bank system This result is not consistent with the paper of Jens Hagendorff et al (2012) when studying the European bank Table 6: Regression result – coefficient for Hypothesis Dep.Var Hypothesis Zscore(ROA) Zscore(ROE) LLP NPL Constant 342.5694*** 55.52610*** -0.007301 0.137789*** CAP -6.83E-13 -1.94E-14 1.46E-16*** 5.29E-16*** DEP -110.7559*** -10.79346*** 0.001928 0.013662 LEV 184.6932* -0.454935 0.004803 0.013122 STA 82.03919*** 7.032048*** 0.004685** -0.001414 SIZE -14.21635*** -1.374873*** 0.000190 -0.005263*** GDP 1.04E-14** 7.08E-16** -3.23E-19 1.49E-18 INF -196.4478 3.224186 -0.034518** -0.097363** LTR 284.2551 -9.912962 0.087132*** 0.248344*** M&A -10.23805 0.668864 -0.000770 0.007988* Prob (F-Statistic) 0.000001 0.000000 0.000000 0.000000 R-squared 0.193522 0.353718 0.282516 0.247695 216 216 216 216 No.obs Note: Statistically significant at * 10%, ** 5%, *** 1% When using the Wald test, we can eliminate the variables DEP, LEV, STA, GDP out of this model because of Prob (F-statistic) > 0.05 (Table B4.1) Test for multicollinearity After using VIF to examine this problem, we see that VIF of INF and LTR are quite high, therefore, the multicollinearity problem may be happening (Table B4.2) We removed the INF variable and retested, we see that VIF of these variables is lower than (Table B4.3) It implies that the model overcomes multicollinearity error Test for Heteroskedasticity The White test shows that Prob (F-statistic) is higher than 0.05, which point out that Heteroskedasticity problem is not an issue of this model (Table B4.4) Test for autocorrelation The autocorrelation problem is examined by Serial Correlation LM which points out that Prob (F-statistic) = 0.0000 < 0.05 Therefore, the model has an autocorrelation problem After overcoming by control directly AR(1) into command, we have the new model addressing the autocorrelation error The best regression model is shown in Table Table 7: The final regession model result testing impact of M&A on risk-taking of the commercial bank in Vietnam Dependent Variable: NPL Method: ARMA Maximum Likelihood (BFGS) Date: 05/28/19 Time: 20:14 Sample: 216 Included observations: 216 Convergence achieved after iterations Coefficient covariance computed using outer product of gradients Variable Coefficient Std Error t-Statistic Prob C 0.148372 0.050648 2.929478 0.0038 CAP 6.19E-16 9.43E-17 6.568722 0.0000 SIZE -0.004338 0.001523 -2.848170 0.0048 LTR 0.082251 0.037254 2.207866 0.0283 MA01 0.009559 0.003296 2.899885 0.0041 AR(1) 0.339498 0.040628 8.356198 0.0000 SIGMASQ 0.000167 8.35E-06 19.98235 0.0000 R-squared 0.266171 Mean dependent var 0.023825 Adjusted R-squared 0.245104 S.D dependent var 0.015114 S.E of regression 0.013131 Akaike info criterion -5.795190 Sum squared resid 0.036038 Schwarz criterion -5.685806 Log likelihood 632.8805 Hannan-Quinn criter -5.750998 F-statistic 12.63458 Durbin-Watson stat 1.954676 Prob(F-statistic) 0.000000 Inverted AR Roots 34 To sum up, the M&A variable has a positive correlation with NPL with statistical meaning at 1% Moreover, the dependent variable has a positive relationship with the risktaking of the bank system In other words, M&A activity exerts a dire consequence to the bank, which is opposite to the proposed hypothesis This result is understandable because although considered as a method to remain the stability of the bank, M&A activity brings many risks for the bank such as the risk of personnel costs, the merger of corporate culture, the burden of bad debt, etc 3.1.5 Results of Hypothesis In Table 8, we see that Pro (F-statistic) of the models is lower than 0.01, therefore, the models are statistically significant at 1% However, the CRISIS variable at the model III and model IV whose the dependent variable is LLP and NPL are statistical meaning at 10% and 5% respectively Both models show the same result, that is CRISIS variable has a positive relationship with LLP and NPL Therefore, the crisis 2008 has an immediate negative impact on the stability of the Vietnamese bank in 2008, not have a late effect as the proposed hypothesis Table 8: Regression result – coefficient for Hypothesis Dep.Var Hypothesis Zscore(ROA) Zscore(ROE) LLP NPL Constant 356.5812*** 55.12491*** -0.005321 0.134625*** CAP -6.02E-13 -1.84E-14 1.63E-16*** 5.60E-16*** DEP -119.7449*** -11.08793*** 0.000337 0.007355 LEV 201.6129* -0.427479 0.008117 0.017035 STA 78.52733*** 6.968268*** 0.003892* -0.003104 SIZE -15.08088*** -1.361794*** 4.70E-05 -0.005244*** GDP 1.06E-14** 7.11E-16** -2.88E-19 1.56E-18 INF -259.3174 0.444780 -0.051659*** -0.152353*** LTR 336.6533 -7.368702 0.101828*** 0.297616*** CRISIS 24.64426 0.774104 0.006151* 0.016791** Prob (F-Statistic) 0.000001 0.000000 0.000000 0.000000 R-squared 0.195019 0.353140 0.293210 0.254171 No.obs 216 216 216 216 Note: Statistically significant at * 10%, ** 5%, *** 1% We choose the model with LLP which is the dependent variable because Akaike info criterion and Schwarz criterion of this model is lower than the remaining model, which means that this model is stronger than the other models After using the Wald test, we can eliminate the variables which are not statistically meaning namely C, DEP, LEV SIZE, GDP because Prob (F-statistic) is equal to 0.8732 > 0.05 (Table B5.1) Test for multicollinearity According to the result table of the variables, we see that VIF of INF and LTR are more than 2, therefore, the multicollinearity problem may be happening (Table B5.2) We conducted to remove LTR, the regression table shows that STA is not statistical significance After using the Wald Test, we eliminate this variable out of the model The new model overcame the multicollinearity error because VIFs of all variables are lower than (Table B5.3) Test for Heteroskedasticity The result table shows that Prob (F-statistic) is higher than 0.05, which point out that Heteroskedasticity problem is not an issue of this model (Table B5.4) Test for autocorrelation Tested by Serial Correlation LM, the model has autocorrelation problem now that Pro.(F-statistic) is lower than 0.05 (Table B5.5) After repairing the model, we have the best regression model as Table To conclude, CRISIS has a positive correlation with the dependent variable which has a similar relationship with the bank risk-taking Therefore, the global financial crisis in 2008 has dire consequences on the stability of the commercial bank in Vietnam Table 9: The final regession model result testing impact of the global crisis 2008 on the risk-taking of the Vietnamese commercial bank Dependent Variable: LLP Method: ARMA Maximum Likelihood (BFGS) Date: 05/31/19 Time: 15:16 Sample: 216 Included observations: 216 Convergence achieved after iterations Coefficient covariance computed using outer product of gradients Variable Coefficient Std Error t-Statistic Prob CAP 4.11E-16 3.42E-17 12.03171 0.0000 INF 0.007965 0.007901 1.008114 0.3146 CRISIS 0.003415 0.002160 1.581231 0.0153 AR(1) 0.893757 0.025597 34.91709 0.0000 SIGMASQ 3.12E-05 1.84E-06 16.92312 0.0000 R-squared 0.336260 Mean dependent var 0.013917 Adjusted R-squared 0.323678 S.D dependent var 0.006870 S.E of regression 0.005649 Akaike info criterion -7.484204 Sum squared resid 0.006734 Schwarz criterion -7.406072 Log likelihood 813.2940 Hannan-Quinn criter -7.452639 Durbin-Watson stat 2.195268 Inverted AR Roots 3.1.6 Results of Hypothesis 89 In table 10, the variance of the dependent variable has Prob > 0.05, so at a 95% confidence level, the assumption of equal variance is accepted (there is no difference in variance) Therefore, the ANOVA analysis results can be used Table 10: Test of homogeneity of variances Levene statistic df1 df2 Prob 0.563 210 0.729 After running the ANOVA test, the result table 11 shows Prob Coefficient < 0.05 It can be confirmed that there is a statistical difference in risk-taking between the bank groups among the number of years of operation Table 11: The results of the ANOVA test Sum of Squares df Mean Square F Prob Between Groups 18.523 3.705 9.380 0.00000 Within Groups 82.939 210 0.395 Total 101.462 215 Conclusions and Policy Implications In this study, we examined six hypotheses relating to the risk-taking of the Vietnamese bank after the global crisis in 2008 To achieve this goal, we used a unique dataset of 216 observations from 31 Vietnamese banks from 2008 to 2018 We used panel data to run regression models by Eviews and test ANOVA by SPSS We have the following results: This research showed that bank performance has a positive impact on the bank’s stability, which matches with the author’s proposed hypothesis The higher profit of the banks is, the lower loan loss provision and the ratio of non-performing loans are One of the selling points of this study is the finding of the relationship between the gender of the CEO and the bank risk-taking in Vietnam It emphasized that the banks led by female usually face up with higher risk than that led by a male This result is opposite with many previous pieces of research, however, it is reasonable when applying in the Vietnamese case According to a survey of Navigos Group which is one of the biggest groups providing personnel recruitment services, the female usually copes with invisible barriers The female CEOs in Vietnam deal with difficulties in balancing work and family and lack of support and sympathy from their family Moreover, some social prejudices still exist in Vietnam Many people still believe that women are too soft to run their businesses, which breeds not-beneficent results in negotiating and engaging in large contracts (Nguyen, 2018) Our conclusion is of the particular importance of the banking system which implies that demographic characteristics of bank managers should be considered carefully with the position be under pressure like the banking industry Moreover, we also evaluated one of the methods of SBV in restructuring the Vietnamese bank system The bail-out purchases of SBV in 2015 is considered as the last method to save these banks out of bankruptcy However, the result indicated that it does not influence the stability of the bank system Our conclusion suggests that the government should be considered as bankruptcy law which allowing the weak banks to go bankrupt, instead of saving banks in all ways Depositors will weigh between profit and risk for deposits as an investment Therefore, banks will transparent information and healthy financial to attract customers' trust instead of racing interest rate Furthermore, it improves the bank's responsibility for its investments Thanks to these benefits, allowing banks bankruptcy is the inevitable trend of the bank system in the coming period Therefore, commercial banks should well execute risk management One of the findings has gone beyond the authors’ speculation is that merger and acquisition activity have a dire consequence to the bank, which is contradictory to our previous expectations It makes the ratio of non-performing loans increase significantly, which plays a crucial role in raising the risk of banks Moreover, the mix of corporate culture, the shift of human resources, conflicts between shareholders, etc are the risks that banks may face when implementing M&A deals Hence, commercial banks should carefully consider before making M&A decisions The global crisis of 2008 has an immediate effect on the stability of the bank system in Vietnam It is one of the main culprits making loan loss provision and non-performing loans increased considerably, which rises risk The openness of Vietnam's economy had not been large yet at that time, which helps the economic consequences react slower than in other countries However, the banking system suffered immediately because of the close relationship between domestic banks and foreign banks Nowadays, now that the openness of the Vietnamese economy is larger than before, each fluctuation of the global economy impacts immediately and seriously Its consequences for the bank system could not be anticipated Therefore, risk management should be one of the priority tasks of the bank, especially when the banking bankruptcy law is executed Finally, we found that there is a difference of risk-taking among the banks having different years of operation The seniority plays an important role in risk-taking of the commercial banks because knowledge and experience will be cultivated day by day References Aspachs, O., Nier, E & Tiesset, M., 2005 bis.org [Online] Available at: https://www.bis.org/bcbs/events/rtf05AspachsNierTiesset.pdf [Accessed 12 December 2018] B.Jones, R., 2012 20% Chance of Rain: Exploring the concept of risk s.l.:s.n Barucci, E & Milani, C., 2018 Do European banks manipulate risk weights? International review of Financial Analysis, pp 47-57 Beccalli, E & Frantz, P., 2009 M&A Operations and Performance in Banking Journal of Financial Services Research Bellucci, A., Borisov, A & Zazzaro, A., 2010 http://docs.dises.univpm.it [Online] Available at: http://docs.dises.univpm.it/web/quaderni/pdfmofir/Mofir047.pdf [Accessed 10 December 2018] Bonstandzic, D & Weib, G., 2018 Why some banks contribute more to global systemic risk? Jounrnal of Financial Intermediation, pp 17-40 Bunda, I & Desquilbet, J.-B., 2008 The bank liquidity smile across exchange rate regimes International Economic Journal, 22(3), pp 361-386 Chen, M., Wu, J., Jeon, B N & Wang, R., 2017 Do foreign banks take more risk? Evidence from emerging economies journal of banking and finance, pp 20-39 Dam, L & Koetter, M., 2012 Bank Bailouts and Moral Hazard: Evidence from Germany The Review of Financial Studies, 25(8), p 2343–2380 Davila, E & Walther, A., 2017 Does Size Matter? Bailouts with Large and small bank s.l.:s.n Drakos, A A., Kouretas, G P & Tsoumas, C., 2016 Ownership, interest rates and bank risk-taking in Central and Eastern European countries International Review of Financial Analysis, pp 308-319 Fola, B., 2015 http://etd.aau.edu.et [Online] Available at: http://etd.aau.edu.et/bitstream/handle/123456789/2834/Belete%20Fola.pdf?sequence=1&is Allowed=y [Accessed 25 December 2018] Haas, R D & Lelyveld, I V., 2008 ebrd.com [Online] Available at: https://www.ebrd.com/downloads/research/economics/workingpapers/wp0105.pdf [Accessed 28 November 2018] Hagendorff, J., Nieto, M J & Wall, L D., 2012 Ideas [Online] Available at: https://ideas.repec.org/p/fip/fedawp/2012-13.html [Accessed January 2019] Khan, M S., Scheule, H & Wu, E., 2017 Funding liquidity and bank risk taking Journal of Banking and Finance, pp 203-216 Le, T T., 2016 Tapchitaichinh [Online] Available at: http://tapchitaichinh.vn/nghiencuu trao-doi/trao-doi-binh-luan/tai-cau-truc-he-thong-ngan-hang-ket-qua-va-lo-trinhcho-giai-doan-moi-106802.html [Accessed 22 December 2018] Le, T X., 2016 dainamedu [Online] Available at: http://dainam.edu.vn/sap-nhaphop-nhat-mua-lai-xu-huong-tat-yeu-cua-cac-nhtm-viet-nam-trong-qua-trinh-phattrien.htm [Accessed 26 December 2018] Madura, J., 2010 Financial Markets and Institutions Canada: South-western Cengage Learning Mai, T P T & Bui, T D., 2018 Tapchitaichinh [Online] Available at: http://tapchitaichinh.vn/kinh-te-vi-mo/yeu-to-anh-huong-den-rui-ro-thanh-khoan-cua-cacngan-hang-thuong-mai-viet-nam-148223.html?mobile=true [Accessed January 2019] Mishkin, F S., 2016 The Economics of Money, Banking and Fianancial Markets England: Pearson Education Nguyen, H S., Tran, T T T & Tran, T H Y., 2014 Bank restructuring- International Perspectives and Vietnam Practices Accounting and Finance Research, 3(1), pp 36-50 Nguyen, H S., Tran, T T T & Tran, T H Y., 2014 Bank Restructuring–International Perspectives and Vietnam Practices Accounting and Finance Research, 3(2), pp 36-50 Nguyen, N A et al., 2012 Financial and economic report in Vietnam in 2011, s.l.: Ministry of Finance Nguyen, T., 2018 Lãnh đạo nữ đối mặt với nhiều rào cản vơ hình [Online] Available at: https://tinnhanhchungkhoan.vn/doanh-nhan/lanh-dao-nu-doi-mat-voi-nhieurao-can-vo-hinh-232136.html Orsholits, D., 2014 The causes and consequences of the 2008 Financial crisis, Geneve: Universite de Geneve Pham, N D & Dinh, X H., 2014 Financial - Currency Textbook Hanoi: Financial Publishing House Phan, D K & Nguyen, V T., 2017 Các yếu tố vi mô ảnh hưởng đến rủi ro tín dụng: Trường hợp ngân hàng thương mại cổ phần sở hữu nhà nước Hậu Giang Tạp chí khoa học Trường Đại học Cần Thơ, IV(48), pp 104-111 Phuong Dong, c., 2011 ORS [Online] Available at: http://www.ors.com.vn/194/News/tin-nhan-dinh-chuyen-gia/194/85775/roaroe-cua-ngan-hang-giam-lien-tuc.aspx [Accessed Feburary 2019] Powell, M & Ansic, D., 1997 Gender Differences in Risk Behaviour in Financial Decision-Making: An Experimental Analysis Journal of Economic Psychology , Volume 18, pp 605-628 Prompitak, D., 2009 ujcontent.uj.ac.za [Online] Available at: https://ujcontent.uj.ac.za/vital/access/manager/Repository/uj:24309;jsessionid=47C1482022 B8777A5602F754C5477B87?exact=sm_creator%3A%22Ragavan%2C+Merusha%22 [Accessed 10 january 2019] S.Rose, P., 1999 Commercial bank management In: United States: Texas A&M University, p 212 S.Rose, P., 2003 Commercial Bank Management United State: McGraw-Hill Saif-Alyousfi, A., Asish, S & Md-Rus, R., 2018 The impact of bank competition and concentration on bank risk-taking behavior and stability: Evidence from GCC countries North American Journal of Economics and Finance Samet, A., Boubakri, N & Boubaker, S., 2018 Does public-private status affect bank risk taking? Worldwide evidence Journal of International Financial Markets, Institutions & Money, pp 287-306 Sedunov, J., Berger, A N & Roman, R A., 2016 Do Bank Bailouts Reduce or Increase Systemic Risk? The Effects of TARP on Financial System Stability s.l.:s.n Shaban,M.; James,G.A., 2017 The effects of ownership change on bank performance and risk exposure: Evidence from Indonesia Journal of Banking and Finance, pp 483-497 Skata, D & Weill, L., 2018 Does CEO gender matter for bank risk? Economic System, pp 64-74 Thai, P., 2018 cafef.vn [Online] Available at: http://cafef.vn/thong-doc-noi-ve-viecmua-lai-3-ngan-hang-gia-0-dong-20180516223916616.chn [Accessed Feburary 2019] Tin, B Q., 2015 cafef.vn [Online] Available at: http://cafef.vn/tai-chinh-nganhang/ma-ngan-hang-thuong-mai-duoc-va-mat 20150311142701405.chn [Accessed 13 January 2019] Ton, T T., 2009 sbv.gov.vn [Online] Available at: https://www.sbv.gov.vn/webcenter/portal/vi/menu/rm/apph/tcnh/tcnh_chitiet?leftWidth= 20%25&showFooter=false&showHeader=false&dDocName=CNTHWEBAP01162521370&r ightWidth=0%25¢erWidth=80%25&_afrLoop=4176563192277577#%40%3F_afrLoop%3 D4176563192277577%26cente [Accessed Feburary 2019] Tran, T T T & Truong, T H T., 2014 Xử lý nợ xấu - Cách nhìn từ nước khu vực hàm ý cho Việt Nam, s.l.: s.n Truong, Q T., 2013 Cantholib [Online] Available http://www.cantholib.org.vn/Database/Content/2904.pdf [Accessed January 2019] at: Tung, L & Kim, T., 2016 cafef.vn [Online] Available at: http://cafef.vn/3-ngan-hang0-dong-ngay-ay-bay-gio-20161005162004248.chn [Accessed Feburary 2019] Valla, N., Saes-Escorbiac, B & Tiesset3, M., 2006 bis.org [Online] Available at: https://www.bis.org/ifc/publ/ifcb28g.pdf [Accessed 11 December 2018] Vazquez, F & Federico, P., 2015 Bank funding structures and risk: Evidence from the global financial crisis Journal of Banking & Finance, pp 1-14 Viet, T., 2017 SBV to stop acquiring weak banks at zero đồng [Online] Available at: https://vietnamnews.vn/economy/374649/sbv-to-stop-acquiring-weak-banks-at-zerodong.html#5sVcK7QIdOIkihkO.97 [Accessed January 2019] Vodova, 2012 researchgate.net [Online] Available at: https://www.researchgate.net/publication/268077176_Liquidity_of_Czech_and_Slovak_Co mmercial_Banks [Accessed 24 November 2018] Vo, T Q & Bui, N T., 2014 Các yếu tố ảnh hưởng đến rủi ro tín dụng hệ thống ngân hàng thương mại Việt Nam Tạp chí khoa học Trường Đại học mở TP Hồ Chí Minh, 3(36), pp 16-25 Vu, T H., 2015 Các yếu tố ảnh hưởng đến khoản ngân hàng thương mại Việt Nam Nghiên cứu & Trao đổi, July, Volume 33, pp 32-49  ... profit of the banks is, the lower loan loss provision and the ratio of non-performing loans are One of the selling points of this study is the finding of the relationship between the gender of. .. relating to the risk-taking of the Vietnamese bank after the global crisis in 2008 To achieve this goal, we used a unique dataset of 216 observations from 31 Vietnamese banks from 2008 to 2018 We... that, one of this paper’s outstanding point is the usage of a control variable, namely NYO which is an abbreviation of the number of years of operation of bank i at the time t The number of operating