CHAPTER IV ANALYSIS AND DISCUSSION

CHAPTER IV ANALYSIS AND DISCUSSION 4.1 Choosing The Model As noted in chapter 3, model to process data for this panel should be selected before analy...
Author: Oscar Lawson
CHAPTER IV ANALYSIS AND DISCUSSION

4.1 Choosing The Model As noted in chapter 3, model to process data for this panel should be selected before analyzing the data. To choose the most appropriate models for this research, the method test is performed. The test is done twice, the first Likelihood ratio test to choose between common effect models and fixed effect model. The second test is Hausman test to choose between fixed effect models and random effect models. The result of likelihood ratio test of the calculations is shown in the table below.

Table 4.1 Likelihood Ratio Test Result Redundant Effects Tests

Fixed

Pool: APOOL Test cross-section fixed effects Effects Test

Statistic

d.f.

Prob.

Cross-section F

436.653376

(25,1604)

0.0000

Refer to Table 4.1 the ! for the likelihood ratio test is 0.0000. That shows that ! 0.05 in other words, H0 is accepted. So that, random effect model is chosen to analyze this research. Although the test results show that the random effect is the right model for this research, there are some exceptions regarding the selection of the model. It is a theory of fixed effect and random effect models. Quoted from Gujarati (pp.650, 2004) based on observation made by Judge et al., “If T (the number of time series data) is large and N (the number of cross-sectional units) is small, there is likely to be little difference in the values of the parameters estimated by Fixed Effect and Random Effect. Hence the choice here is based on computational convenience. On this score, Fixed Effect may be preferable.” Based on the theory above, it can be concluded that fixed effect method is preferable for this research.

4.2 Unit Root Test There are two natural assumptions that we can make when doing unit root test. First, one can assume that the persistence parameters are common across cross-sections. The Levin, Lin, and Chu (LLC) and Breitung tests all employ this assumption. The others is the persistence parameters that vary freely across cross-sections. The Fisher-ADF and Fisher-PP tests are of this form.

27

The null and alternative hypotheses for the tests may be written as: ! < 0,05 Ho : There is unit root in the data H1: There is no unit root in the data

Table 4.3 Unit Root Test Result

Common Levin, Lin & Chu t*

Variables LCR DEMD SAVD TIMED LOAN CASH RESV SEC CAP

Statistic Prob.** Statistic Prob.** Statistic Prob.** Statistic Prob.** Statistic Prob.** Statistic Prob.** Statistic Prob.** Statistic Prob.** Statistic Prob.**

-184.281 0.0327 -116.290 0.0000 -986.691 0.0000 -101.607 0.0000 -872.565 0.0000 -107.068 0.0000 -897.446 0.0000 -117.160 0.0000 -108.035 0.0000

Unit Root Process Individual Im, Pesaran ADF - Fisher PP - Fisher and Shin WChi-square Chi-square stat -457.672 981.715 185.206 0.0000 0.0001 0.0000 -186.438 438.526 1059.98 0.0000 0.0000 0.0000 -168.505 384.012 1061.90 0.0000 0.0000 0.0000 -170.459 393.419 1010.96 0.0000 0.0000 0.0000 -141.758 307.895 838.312 0.0000 0.0000 0.0000 -187.934 440.728 1044.74 0.0000 0.0000 0.0000 -159.946 358.141 963.144 0.0000 0.0000 0.0000 -155.706 348.490 938.827 0.0000 0.0000 0.0000 -155.133 349.035 926.463 0.0000 0.0000 0.0000

Based on the calculation shown in Table 4.3, it shows that H1 is accepted. It can be summarized that all the data in this research is stationary.

28

4.3 Descriptive Statistic Analysis The normality of residual can be known from the probability of Jarque Bera value. ! < 0.05 Ho: The data is not normally distributed H1: The data is normally distributed

Table 4.4 Normality Test Result

LCR_? DEMD_? SAVD_? TIMED_? LOAN_? CASH_? RESV_? SEC_? CAP_?

Jarque-Bera 139185,4 1357434 3485756 5872466 3337777 9788054 6123809 5969268 1263578

Probability 0.000000 0.000000 0.000000 0.000000 0.000000 0.000000 0.000000 0.000000 0.000000

Refer from the table above; all the probability of variables is 0.0000. It shows that ! < 0,05, so H1 is accepted. It means that all the data used in this research is normally distributed.

4.4 Multiple Linear Regression This analysis is intended to determine the influence between independent variable to dependent variable. The purpose is to predict or estimate the value of the dependent variable in a causal relationship to the value of another variable. By using fixed effect model (FEM), result obtained in the following calculation.

29

Table 4.5 Fixed Effect Method Multiple Regression Result

Variable

Coefficient Std. Error

t-Statistic

Prob.

C

0.135868

0.009061

14.99417

0.0000

DEMD_?

0.000648

0.001632

0.397033

0.6914

SAVD_?

0.001143

0.001271

0.899252

0.3687

TIMED_?

0.004480

0.001323

3.387056

0.0007

LOAN_?

0.002055

0.001481

1.387139

0.1656

CASH_?

0.004198

0.001673

2.508926

0.0122

RESV_?

-0.017507 0.001954

-8.960523

0.0000

SEC_?

0.002729

2.144779

0.0321

CAP_?

-0.004010 0.000973

-4.123358

0.0000

0.001272

Cross-section fixed (dummy variables) Weighted Statistics R-squared

0.880181

Mean dependent var

0.457460

0.877716

S.D. dependent var

2.695936

S.E. of regression

0.955737

Sum squared resid

1465.146

F-statistic

357.0560

Durbin-Watson stat

1.311623

Prob(F-statistic)

0.000000

4.4.1 Correlation Analysis Refer to Table 4.5 above; the value of the R-squared is 0.88. So the R-value that shows coefficient correlation is 0.94. The values are then interpreted based on objective criteria as follows:

30

Table 4.6 Coefficient Correlations and The Estimation Coefficient Interval

Relationship

0.00 - 0.199

Very Low

0.20 - 0.399

Low

0.40 – 0.599

Average

0.60 – 0.799

Strong

0.80 – 1.000

Very Strong Source : Sugiyono, 2002:183

Based on the interpretation of the correlation coefficient table presented above, the correlation coefficient of 0.94 indicates a very strong relationship between the independent variable (X) with the LCR (Y).

4.4.2 Multiple Linear Regression Model From the output in Table 4.5 known that constanta and regression coefficient values can be set up so that multiple linear regression equation as follows: Y = 0.13587 + 0.00065 X1 + 0.00114 X2 + 0.00448 X3 + 0.00206 X4 + 0.0042 X5 0.01751 X6 + 0.00273 X7 - 0.00401 X8 + e…………………...…………..(Equation 4.1)

Above equation can be interpreted as follows: b0 = 0.13587 means that if the variables X1, X3, X3, X4, X5, X6, X7, X8 is zero, then the Liquidity Coverage Ratio will be worth 0.13587. b1 = 0.00065 indicates that for every 1 % increase in Current accounts growth (X1) will cause a increase by 0.00065 % in Liquidity Coverage Ratio. b2 = 0.00114 indicates that for every 1 % increase in saving deposit growth (X2) will cause a increase by 0.00114 % in Liquidity Coverage Ratio.

31

b3 = 0.00448 indicates that for every 1 % increase in time deposit growth (X3) will cause a increase by 0.004480% in Liquidity Coverage Ratio. b4 = 0.00206 indicates that for every 1 % increase in loans growth (X4) will cause a increase by 0.00206 % in Liquidity Coverage Ratio. b5 = 0.0042 indicates that for every 1 % increase in cash growth (X5) will cause a increase by 0.0042 % in Liquidity Coverage Ratio. b6 = -0.01751 indicates that for every 1 % increase in Reserves growth (X6) will cause a decrease of 0.01751 % in Liquidity Coverage Ratio. b7 = 0.00273 indicates that for every 1 % increase in securities growth (X7) will cause a increase by 0.00273 % in Liquidity Coverage Ratio. b8 = -0.00401 indicates that for every 1 % increase in capital growth (X8) will cause a decrease by 0.004010 % in Liquidity Coverage Ratio.

4.4.3 Coefficient of Determination Analysis Reffer to Table 4.5 known that the R-Squared values is 0.880181. Thus, the coefficient of determination value obtained at 88% which shows the sense that the current account (X1). Saving deposit (X2), Time deposits (X3), Loan (X4), Cash (X5), Reserves (X6), Securities (X7), and Capital (X8) give effect simultaneously (together) for 88% to the LCR (Y). While the rest of 12% influenced by other factors that are not observed.

4.4.4 Partial Hypothesis Test Based on the significant level ! of 5 %, the hypothesis of partial test for independend variables as follow. ! < 0.05 H0

: Independend variables have no significant effect on the LCR

H1

: Independend variables have significant effect on the LCR

32

From Table 4.5, there are five independent variables that have significant effect to LCR. It contain of Time Deposit, Cash, Reserves, Securities, and Capital. And the rest, Current accounts, Saving Deposits and Loan have an effect, but not significant to LCR. There are three variables that not significant to LCR. Thus, those variables have to be eliminated. The most insignificant variable, that is DEMD (current account) need to eliminate first. After eliminated that variable, still, there are two variables that have no significant effect to LCR, which is saving deposit and loan. The result of multi linear regression by eliminate insignificant variables is shown in appendix 6. Finally after eliminating three insignificant variables, the result of this research is shown below.

Table 4.7 Fixed Effect Method Multiple Regression Result (Final Result)

Variable

Coefficient Std. Error

t-Statistic

Prob.

C

0.135868

0.009061

14.99417

0.0000

TIMED_?

0.004542

0.001301

3,491,571

0.0005

CASH_?

0.004088

0.001659

2,464,422

0.0138

RESV_?

-0.017729 0.001890

-9,382,943

0.0000

SEC_?

0.002966

2,411,509

0.0160

CAP_?

-0.004111 0.000942

-4,365,885

0.0000

0.001230

Cross-section fixed (dummy variables) Weighted Statistics R-squared

0.878350

Mean dependent var

0.448643

0.876079

S.D. dependent var

2,684,850

S.E. of regression

0.957248

Sum squared resid

1,472,533

F-statistic

3,867,673

Durbin-Watson stat

1,312,045

Prob(F-statistic)

0.000000

33

4.4.5 Correlation Analysis of Final Result Refer to Table 4.7 above; the value of the R-squared is 0,878. So the R-value that shows coefficient correlation is 0.937. Based on the interpretation of the correlation coefficient in Table 4.6, the correlation coefficient of 0.937 indicates a very strong relationship between the independent variable (X) with the LCR (Y).

4.4.6 Multiple Linear Regression Model From the output in Table 4.7, known that constanta and regression coefficient values can be set up so that multiple linear regression equation as follows: Y = 0.135868 + 0.004542 X3 + 0.004088 X5 - 0.01773 X6 + 0.002966 X7 - 0.00411 X8 + e…………………...……………………………………………………..(Equation 4.2)

Above equation can be interpreted as follows: b0 = 0.135868 means that if the variables X1, X3, X3, X4, X5, X6, X7, X8 is zero, then the Liquidity Coverage Ratio will be worth 0.135868. b3 = 0.004542 indicates that for every 1 % increase in time deposit growth (X3) will cause a increase by 0.004542 % in Liquidity Coverage Ratio. b5 = 0.004088 indicates that for every 1 % increase in cash growth (X5) will cause a increase by 0.004088 % in Liquidity Coverage Ratio. b6 = - 0.01773 indicates that for every 1 % increase in Reserves growth (X6) will cause a decrease of 0.01773 % in Liquidity Coverage Ratio. b7 = 0.002966 indicates that for every 1 % increase in securities growth (X7) will cause a increase by 0.002966 % in Liquidity Coverage Ratio. b8 = -0.00411 indicates that for every 1 % increase in capital growth (X8) will cause a decrease by 0.00411 % in Liquidity Coverage Ratio.

34

4.4.7 Coefficient of Determination Analysis Reffer to Table 4.7 known that the R-Squared values is 0.878. Thus, the coefficient of determination value obtained at 87.8 % which shows the sense that the Time deposits

(X3), Cash (X5), Reserves (X6), Securities (X7), and Capital (X8) give effect simultaneously (together) for 87.8 % to the LCR (Y). While the rest of

12.2 %

influenced by other factors that are not observed.

4.4.8 Partial Hypothesis Test Based on the significant level ! of 5 %, the hypothesis of partial test for independend variables as follow. ! < 0.05 H0

: Independend variables have no significant effect on the LCR

H1

: Independend variables have significant effect on the LCR

From Table 4.7, it can be conclude that there are five independent variables that have significant effect to LCR. It contain of Time Deposit, Cash, Reserves, Securities, and Capital.

4.5 Discussion 4.5.1 Interpretation of The Multiple Regression Calculation Once the model is determined and carried out tests of the model, the next step is interpreting the results.

4.5.1.1 Current account The management cannot predict flow of funds of Current accounts. The characteristic of this fund that can be withdrawn at any time make this third-parties fund tends to fluctuate. This has become an important point for management in making decisions to put the fund still liquid. Bank management chose to keep this fund and do not utilize it

35

to generate profit to avoid the liquidity risk. Thus, H1 rejected. Can be concluded that the growing number of deposit has not a significant impact on liquidity coverage ratio.

4.5.1.2 Saving Deposit Saving deposit is one of the third party funds that have fluctuates character. Based on the purpose of saving deposit itself is to ensure the safety of customer funds deposits, bank must prioritize it first. After fulfilling the primary purpose of saving deposits, then the management is carefully allocated to support the bank's operations to make profits. The decision is made by analyzing character and trend of each customer to see the probability of which fund can be allocated in banking operation activities. With system management that has been well controlled, then the liquidity problem can be avoided. By that explanation, H2 is rejected. Can be concluded that the increase in the amount of saving deposits will not give significant effect to the liquidity coverage ratio. The positive growth of time deposits would increase liquidity coverage ratio that means liquidity risk decreased.

4.5.1.3 Time Deposit The more the amount of time deposits owned bank, the growth of time deposits would be increased. Increasing the growth of time deposits indicates that funds deposits held by banks increased. By than increase, the bank can utilize those funds to generate profits for the bank. Besides that, the fund can also used to cover liquidity obligation bank owned. As with the Current accounts and saving deposits fluctuated, time deposits tend to be more stable in the period of withdrawals. With maturities of time deposits, bank management can make the allocation plan with the goal of profit maximization but still consider liquidity risk. Due to the nature of time deposits, which tend to be stable, then the liquidity risk can be minimized. Meanwhile, the allocation of time deposits contain a risk that bank cannot return the fund and do not able to fulfill the withdrawal to customer. By that reasons, it can be summarized that H3 is accepted. The positive growth of time deposits would increase liquidity coverage ratio that means liquidity risk decreased.

36

4.5.1.4 Loan The growth of loan has an effect on liquidity risk. The more loan issued by bank, the greater probability of bank to default. Based on analysis result, loan does not have significant effect towards liquidity coverage ratio. Therefore H4 is rejected. It happen because, when the amount of loan increase, the availability of fund owned by bank to cover is decreased.

4.5.1.5 Cash Increasing the number of cash bank-owned from month to month will reduce the liquidity risk. Cash owned by banks are liquid assets, so it can be used at any time to meet short-term bank liabilities. With the availability of cash, the bank can meet shortterm obligations with ease. Therefore, it can be concluded that the cash could reduce the liquidity risk that are in the bank. The availability of fund to cover owned would decrease liquidity risk, so H5 is accepted. The positive growth of cash would increase liquidity coverage ratio that means liquidity risk decreased.

4.5.1.6 Reserves Growth in the number of bank-owned reserves would reduce the number of Liquidity Coverage Ratio owned bank. This indicates that the increase in reserves will raise the amount of liquidity risk the bank owned. Reserves consist of Current accounts in Bank Indonesia and certificate of Bank Indonesia. The portion of Current accounts is set by the central bank in a form of reserve requirement (GWM). Bank management prefers to put the funds in certificate of Bank Indonesia because it gives high interest rate and can be used anytime without causing loss. Even so, the storage in the form of reserve should be balanced. If the bank excess deposits their funds in the form of reserves, banks cannot use these excess funds to generate profit. Banks also must be smart in dividing the allocation portion of the funds in these types of reserves. Because if the storage portion in the form of Current accounts in BI is larger than certificate of BI, the bank would not get high return. Refer to the definition and function, the reserves should have a direct and mutually supportive relationship with the LCR. However, in this study found that the growth in reserves will reduce the liquidity ratio, which means more

37

liquidity risk. This can be caused by the imbalance in the form of deposit funds in Current account and certificate of Bank Indonesia. By that explanation, it can summarize that H6 is accepted.

4.5.1.7 Securities The growth of bank-owned securities indicates that the availability of liquid funds owned by banks in good condition. Of the various types of securities that consists of the Money Market Securities, Securities, Commercial Papers, Certificate of Mutual Funds and Medium Term Notes bank can get a high enough return and reduce potential losses due to save their money in different types of securities. By having sufficient number of securities, banks can mitigate liquidity risk. Banks can sell the securities are held to meet the pressing obligations. Therefore, the increase amount of securities would reduce liquidity risk. The positive growth of securities would increase liquidity coverage ratio that means liquidity risk decreased, so H7 is accepted.

4.5.1.8 Capital The number of capital owned is used as a reserve to absorb losses occurs before or during liquidation. Positive capital growth indicates that the bank has sufficient funds available to overcome liquidity problems. However, the research results obtained in this study stated that the increase in the number of bank-owned capital is not always good. It shows from the declining of liquidity coverage ratio while capital growth increasing. Increased capital indicates that the shareholder funds sediment too much as a reserve bank. Meanwhile, the purpose of shareholder invest is to get a high return. By insufficient utilities of shareholder fund, bank would not operate well to generate profit. It will cause reduced return. The reduced return earned by shareholders, can cause a sense of disappointment from the shareholder's own. Declining sense of confidence that could affect decreasing interest investors to invest and lead them to withdraw the funds already invested. These conditions can be detrimental to the bank, as a reserve fund from investors will be reduced. This will increase liquidity risk that indicated by the decrease in liquidity coverage ratio. Therefore, H8 is accepted.

38