Home / Joint Stock Commercial Bank

Regression Analysis Results Table After Removing Variable X3


Table 2.16: Table of regression coefficients




Model


Unstandardized Coefficients


Standardized Coefficients


t


Sig.


Collinearity Statistics

B

Std. Error

Beta

Tolerance

VIF

1

(Constant)


.038


.015



2,600


.022




X1

-.004

.002

-.442

-2.191

.047

.269

3,711


X2

.012

.005

.301

2,216

.045

.594

1,682


X3

-.007

.016

-.070

-.419

.682

.389

2,572


X4

-.327

.094

-.608

-3.470

.004

.357

2,804


X5

-.761

.300

-.390

-2.538

.025

.462

2,163


X6

-.010

.001

-.940

-6.877

.000

.586

1,705


X7

.015

.018

.114

.859

.406

.622

1,608


X8

.019

.009

.254

2,062

.060

.720

1,389

Maybe you are interested!


a. Dependent Variable: ROA


(Source: Author's research results)


In addition, to measure multicollinearity, the tolerance of the variable (Tolerance) and the VIF magnification factor are considered when VIF exceeds 5, it is a sign of multicollinearity. With the results of calculating Tolerance and VIF from the Coefficients table, the phenomenon of multicollinearity between independent variables in the model does not occur because the VIF coefficients are all less than 5.

The histogram of standardized residuals shows that the distribution of residuals is approximately normal (Mean =0 and standard deviation Std Dev=0.787). Therefore, it can be concluded that the assumption of normal distribution of residuals is not violated.


Figure 2.1: Frequency histogram of residuals


Source: Author's research results

The scatter plot between the residuals and the predicted values ​​of the linear regression model shows that the residual values ​​are randomly scattered in the region around the line passing through the zero intercept, demonstrating that the assumption of linear relationship is not violated.

Looking at the regression coefficient table, we see that the Sig value of variable X3 is 0.682 and variable X7 is

0.406 are all greater than 0.1 so the regression model is not suitable. We see that among the variables, the sig value of the beta coefficient of variable X3 has the largest value and is equal to 0.682 so we remove this variable from the model and regress the model without variable X3. The regression results are as follows:


Table 2.17: Regression analysis results table after removing variable X3


Model Summary b



Model


R


R Square


Adjusted R Square


Std. Error of the Estimate

Change Statistics


Durbin-Watson

R Square

Change


F Change


df1


df2

Sig. F

Change

1

.925a​

.856

.784

.00066385

.856

11,868

7

14

.000

1,733

a. Predictors: (Constant), X8, X5, X6, X7, X2, X1, X4


b. Dependent Variable: ROA


Source: Author's research results




Model


Unstandardized Coefficients

Standardized Coefficients


t


Sig.


Collinearity Statistics

B

Std. Error

Beta

Tolerance

VIF

1

(Constant)

.042

.011


3,847

.002




X1

-.005

.001

-.499

-3.461

.004

.496

2.016


X2

.012

.005

.308

2,362

.033

.605

1,653


X4

-.310

.082

-.576

-3.756

.002

.438

2,285


X5

-.797

.278

-.409

-2.864

.012

.505

1,981


X6

-.010

.001

-.943

-7.121

.000

.588

1,701


X7

.018

.016

.135

1.129

.278

.723

1,383


X8

.020

.008

.272

2,435

.029

.823

1,215

a. Dependent Variable: ROA


Source: Author's research results

Looking at the table above, we see that the Sig of the F value = 11.868 < 10%, but the Sig of the beta value of variable X7 is still greater than 10%, so we continue to remove this variable. Removing variable X7 - economic growth rate GR from the model, we have:


Table 2.18: Regression analysis results table after removing variables X3, X7



Model Summary b



Model


R


R Square


Adjusted R Square


Std. Error of the Estimate

Change Statistics


Durbin-Watson

R Square

Change


F Change


df1


df2

Sig. F

Change

1

.918a​

.843

.780

.00066989

.843

13,388

6

15

.000

1,856

a. Predictors: (Constant), X8, X5, X6, X2, X1, X4


b. Dependent Variable: ROA


Source: Author's research results




Model


Unstandardized Coefficients

Standardized Coefficients


t


Sig.


Collinearity Statistics

B

Std. Error

Beta

Tolerance

VIF

1

(Constant)

.042

.011


3,786

.002




X1

-.005

.001

-.482

-3.333

.005

.501

1,995


X2

.011

.005

.289

2,215

.043

.615

1,626


X4

-.282

.080

-.525

-3.550

.003

.479

2,087


X5

-.797

.281

-.409

-2.839

.012

.505

1,981


X6

-.009

.001

-.882

-7.228

.000

.705

1,418


X8

.022

.008

.298

2,691

.017

.857

1,167

a. Dependent Variable: ROA


Source: Author's research results

After removing variables X3 and X7, looking at the regression table above we see that=0.843, F test (ANOVA table) shows the significance level p (in SPSS, denoted by Sig) =0.000. The Sig beta values ​​of variables X1, X2, X4, X5, X6, X8 are all less than 10%, so the regression model is suitable. In other words, the independent variables explain about 84.3% of the variance of the dependent variable.

So we have the optimal regression model as follows:


ROA=0.042-0.005X1+0.011X2-0.282X4- 0.797X5-0.009X6+0.022X8+

In there:

ROA: return on total assets X1: Logarithm base 10 of total assets

X2: Ratio of outstanding credit to total assets

X4: Credit risk provision expense on total outstanding debt X5: Total non-interest income/ total assets

X6: Total operating expenses/Total operating income X8: inflation rate

:Residual or statistical error

Based on the regression model data, we see that the bank size variable has an inverse impact on bank profits.

The sign of the BOPO coefficient is as expected. The (–) sign of BOPO indicates that inefficient banks will generate lower profits. Conversely, efficient banks will generate higher profits.

Inflation rate has a positive impact on bank profits

The sign of Total non-interest income/total assets is not as expected. The negative correlation coefficient of this variable indicates that the more non-interest income, the lower the bank's profit.

2.4.4. Analysis of model results

The estimated results of the model show that the business performance of Vietinbank is affected by bank size, credit balance ratio, credit risk provision, BOPO business performance, inflation rate. Although the variable of non-interest income on total assets has a significant impact on ROA, its impact is contrary to the initial hypothesis H0. The variable of debt, total equity and economic growth rate do not have a statistically significant impact on bank profits.

2.4.4.1. Internal factors of the bank Bank asset size (LogTA)

Bank asset size is negatively correlated with ROA. The negative correlation indicates


that the more the scale is expanded, the more the profit decreases, showing the diseconomies of scale. This result is consistent with the results of previous research Sufian and Razali (2008). Network expansion increases costs, making it more difficult for banks to manage and supervise. The development of management level and human resources does not keep up with the development of scale, which can create more risks and reduce bank profits.

Ratio of outstanding credit to total assets

The ratio of outstanding credit to total assets has a positive correlation with the bank's ROA. The higher the outstanding credit, the higher the profit, which is suitable for Vietnamese commercial banks because credit activities are the main activity and bring 2/3 of the total income for the bank and are a portfolio that contains a lot of risks. Therefore, to effectively increase credit growth, banks need to strengthen their appraisal and risk management capabilities for the credits granted to customers.

Credit risk provision expense on total outstanding debt

The cost of credit risk provisioning on total outstanding loans has a negative correlation with the bank's ROA. The higher the credit risk, the lower the return on assets. This research result is consistent with the author's initial expectation and consistent with previous results of many researchers. The correlation coefficient has correctly reflected the current situation of Vietinbank in the recent past. Rapid credit growth has left bad debts, causing credit risk provisioning to increase, leading to a decline in the bank's profits.

Total operating expenses to operating income

Operating expenses over operating income have a negative correlation with ROA. This negative correlation indicates that the higher the bank's operating expenses, the lower its business performance and vice versa, the more cost-effective the bank is, the higher its profits will be. This research result is consistent with previous research by Zeitun (2012; Syfari (2012)

2.4.4.2. Factors outside the bank Inflation rate (INF)

Inflation rate has a positive correlation with ROA. The study results indicate that the rate


The higher the inflation rate, the higher the bank's profit. In fact, when inflation increases, banks tend to increase their lending interest rates higher than their deposit interest rates, and this trend increases their profits. On the other hand, during periods of liquidity stress, banks also lend on the interbank market at very high interest rates, which brings a sharp increase in interest income. The research results of the model are consistent with the studies of Sufian (2011); Gul, Irshad and Zaman (2011).

2.4.5. Evaluation of factors affecting Vietinbank's business performance

2.4.5.1. Credit risk provision expense on total outstanding debt

The first factor affecting the business performance of ROA is the cost of credit risk provisioning on total outstanding loans. The regression coefficient of this variable is statistically significant at 10%. That means that when LLP/TL is reduced by 1%, ROA will increase by 0.282%. High cost of credit risk provisioning means that banks are growing credit excessively with poor quality loans and increasing bad debts. In the period of 2008-2012, banks are suffering severe consequences from excessive credit growth, the economic recession has led to businesses having difficulty in repaying debts, causing bad debts of banks to increase, affecting bank profits. And during this period, Vietinbank's cost of credit risk provisioning is many times larger, the total accumulated cost of credit risk provisioning is 12,870 billion VND, an increase of more than 4,074 billion VND compared to the previous period. Therefore, it is necessary to implement solutions to minimize credit risk provisioning costs to limit and handle bad debt.

2.4.5.2. Bank size

The second factor affecting ROA is bank size. The regression coefficient β = -

0.005 means that with other factors remaining the same, when the asset size increases by 1%, ROA will decrease by 0.005%, meaning that the growth rate of total assets increases, which has an impact on reducing business performance and vice versa. This is because the bank's size is too large, managing this asset block requires highly qualified human resources and costs a lot in management and operation. Therefore, it reduces bank profits. The widespread network of operations creates favorable conditions for Vietinbank in mobilizing capital.


capital, but if not managed well, diseconomies of scale will appear, and asset growth for the bank in this case will cause disadvantages. Vietinbank's operating costs are quite high compared to other banks. This increase is mainly due to personnel costs and expansion costs. Therefore, banks need to consider carefully before deciding to expand their current operations to avoid the impact of the law of diminishing returns to scale.

2.4.5.3. Credit balance size

The third factor affecting business performance is the size of outstanding loans. TL/TA. The regression coefficient of this variable is statistically significant at the 10% level. That means that with other factors remaining unchanged, a 1% increase in TL/TA will increase ROA by 0.011%. Credit growth contributes to increasing income and increasing profits. This is very suitable because for Vietnamese commercial banks, credit granting is the activity that brings the largest income to the bank. For Vietinbank, lending is still the key activity, accounting for about 65 - 70% of total assets. However, the loan/total asset ratio (LAR) has tended to decrease in recent quarters due to the reason that the growth rate of customer loans has decreased because banks have tightened lending to control bad debt.

2.4.5.4. Operating expense to operating income ratio

The next factor affecting ROA is the ratio of operating expenses to operating income. The regression coefficient of this variable is statistically significant at the 10% level. That means that a 1% decrease in BOPO will increase ROA by 0.009% and vice versa. Vietinbank's operating expense to operating income ratio has continuously increased sharply since 2009. In 2009, this ratio reached a fairly high level of nearly 60% while other banks were only around 40%. In the following years, this ratio has tended to decrease gradually, always remaining below 50% and at a low level compared to the industry. In the period from 2011 to 2013, the BOPO ratio tended to increase because the growth rate of expenses tended to increase while the growth rate of income tended to decrease over the years, causing the CIR ratio to increase. Therefore, banks need to take measures to better control operating expenses in the coming years.

2.4.5.5. Inflation rate

The final factor affecting ROA is inflation. The regression coefficient of this variable is significant.

Comment


Agree Privacy Policy *