1.
Test the stationarity for each variable
Based on Figure 1.0, Figure 1.1, Figure 1.2, Figure 1.3 and Figure 1.4, LNKLCI, LNIPI,
MYR/USD, MYR/BAHT and LNM1 are stationary at first difference stage because the values
are fluctuate around a constant mean or fluctuate with constant variation.
2. Based on appropriate hypothesis, how many cointegrating relationship exists based on JJ test?
Null Hypothesis Alternative Hypothesis Result Trace
Ho : r = 0 Ho : r > 0 Trace > Critical Value
73.04 > 68.52
Ho : r ≤ 1 Ho : r > 1 Trace < Critical Value
40.19 < 47.21
Ho : r ≤ 2 Ho : r > 2 Trace < Critical Value
12.07 < 29.68
Ho : r ≤ 3 Ho : r > 3 Trace < Critical Value
3.11 <15.41
Ho : r ≤ 4 Ho : r > 4 Trace < Critical Value
0.03 < 3.76
Table 1.0
We choose Trace test because of the Figure 2.0. After we choose Trace, we run the
cointegration test again and the result for Trace test are included in Table 1.0 above. Based on
the Table 1.0, we can see that the trace value is smaller than critical value at 5% level at r ≤ 1.
Ho : r = 0 , we reject Ho because there is at least one cointegrating vector.
Ho : r ≤ 1, we failed to reject Ho indicates that there is at least one cointegrating vector.
3. Determine the optimal lag for VAR
Based on the Final Prediction Error (FPE) in Figure 3.0, the optimal lag is at 3.
4. Estimate VAR based on optimal lag. Explain your VAR results
We have to estimate VAR by using the optimal lag that we have choose before which is lag 3.
We get the coefficient for the variables as shown below:
LNKLCI as dependent variable :
1
𝐿𝑁𝐾𝐿𝐶𝐼 = 0.96236 + 1.200329𝐿𝑁𝐾𝐿𝐶𝐼𝑡 − 1 – 0.390113𝐿𝑁𝐾𝐿𝐶𝐼𝑡 − 2
+ 0.102223 𝐿𝑁𝐾𝐿𝐶𝐼𝑡 − 3 + 0.134424𝐿𝑁𝐼𝑃𝐼𝑡 − 1 – 0.128393𝐿𝑁𝐼𝑃𝐼𝑡
− 2 – 0.057452 𝐿𝑁𝐼𝑃𝐼𝑡 − 3 + 0.228702 𝐿𝑁𝑀1𝑡 − 1 + 0.008180 𝐿𝑁𝑀1𝑡
− 2 – 0.155120 𝐿𝑁𝑀1𝑡 − 3 + 0.001182 𝑀𝑌𝑅/𝐵𝐴𝐻𝑇𝑡
− 1 – 0.012378 𝑀𝑌𝑅/𝐵𝐴𝐻𝑇𝑡 − 2 – 0.000215 𝑀𝑌𝑅/𝐵𝐴𝐻𝑇𝑡 − 3
+ 0.261526 𝑀𝑌𝑅/𝑈𝑆𝐷𝑡 − 1 + 0.811553𝑀𝑌𝑅/𝑈𝑆𝐷𝑡
− 2 – 1.447785 𝑀𝑌𝑅/𝑈𝑆𝐷𝑡 − 3
Based on the t-stat of each coefficient, it shows that for LNKLCIt-1, and LNKLCIt-2 are
significant because the t-stat value which is 14.1998 and 3.12776 are greater than critical value
at 5% which is 1.96. However, for the remaining independent variables the coefficients are not
significant because the t-values are smaller than the critical value at 5% level.
LNIPI as dependent variable :
𝐿𝑁𝐼𝑃𝐼 = 0.182201 + 0.171117 𝐿𝑁𝐾𝐿𝐶𝐼𝑡 − 1 − 0.026605 𝐿𝑁𝐾𝐿𝐶𝐼𝑡
− 2 – 0.079721 𝐿𝑁𝐾𝐿𝐶𝐼𝑡 − 3 + 0.152476 𝐿𝑁𝐼𝑃𝐼𝑡 − 1
+ 0.423466 𝐿𝑁𝐼𝑃𝐼𝑡 − 2 + 0.316645 𝐿𝑁𝐼𝑃𝐼𝑡 − 3 − 0.577168 𝐿𝑁𝑀1𝑡
− 1 + 0.340191𝐿𝑁𝑀1𝑡 − 2 + 0.240714 𝐿𝑁𝑀1𝑡 − 3
+ 0.016975 𝑀𝑌𝑅/𝐵𝐴𝐻𝑇𝑡 − 1 – 0.036373 𝑀𝑌𝑅/𝐵𝐴𝐻𝑇𝑡 − 2
+ 0.017339𝑀𝑌𝑅/𝐵𝐴𝐻𝑇𝑡 − 3 − 0.989340𝑀𝑌𝑅/𝑈𝑆𝐷𝑡 − 1
+ 1.852874𝑀𝑌𝑅/𝑈𝑆𝐷𝑡 − 2 – 1.407334 𝑀𝑌𝑅/𝑈𝑆𝐷𝑡 − 3
Based on the t-stat of each coefficient, it shows that for LNKLCIt-1, LNIPIt-1, LNIPIt-2, LNIPIt-
3, LNM1t-1, LNM1t-2 and MYR/BAHTt-1are significant because the t-stat values are greater than
critical value at 5% which is 1.96. However, for the remaining independent variables the
coefficients are not significant because the t-values are smaller than the critical value at 5%
level.
LNM1 as dependent variable :
𝐿𝑁𝑀1 = −0.112694 + 0.054801𝐿𝑁𝐾𝐿𝐶𝐼𝑡 − 1 + 0.003005𝐿𝑁𝐾𝐿𝐶𝐼𝑡
− 2 – 0.081793 𝐿𝑁𝐾𝐿𝐶𝐼𝑡 − 3 + 0.053228𝐿𝑁𝐼𝑃𝐼𝑡 − 1 + 0.018718𝐿𝑁𝐼𝑃𝐼𝑡
− 2 + 0.034379 𝐿𝑁𝐼𝑃𝐼𝑡 − 3 + 0.871543 𝐿𝑁𝑀1𝑡 − 1 – 0.035111𝐿𝑁𝑀1𝑡
− 2 + 0.135626 𝐿𝑁𝑀1𝑡 − 3 + 0.010026 𝑀𝑌𝑅/𝐵𝐴𝐻𝑇𝑡 − 1
+ 0.007152 𝑀𝑌𝑅/𝐵𝐴𝐻𝑇𝑡 − 2 – 0.014835 𝑀𝑌𝑅/𝐵𝐴𝐻𝑇𝑡
− 3 – 0.513678𝑀𝑌𝑅/𝑈𝑆𝐷𝑡 − 1 + 0.587988𝑀𝑌𝑅/𝑈𝑆𝐷𝑡 − 2
+ 0.326393 𝑀𝑌𝑅/𝑈𝑆𝐷𝑡 − 3
2
Based on the t-stat of each coefficient, it shows that for LNM1t-1 is significant because the t-
stat value is greater than critical value at 5% which is 1.96. However, for the remaining
independent variables the coefficients are not significant because the t-values are smaller than
the critical value at 5% level.
MYR/BAHT as dependent variable :
𝑀𝑌𝑅/𝐵𝐴𝐻𝑇 = 4.121282 – 0.657970𝐿𝑁𝐾𝐿𝐶𝐼𝑡 − 1 + 0.048143𝐿𝑁𝐾𝐿𝐶𝐼𝑡 − 2
+ 0.861663 𝐿𝑁𝐾𝐿𝐶𝐼𝑡 − 3 – 0.137911𝐿𝑁𝐼𝑃𝐼𝑡 − 1 – 0.188912𝐿𝑁𝐼𝑃𝐼𝑡
− 2 – 0.401106 𝐿𝑁𝐼𝑃𝐼𝑡 − 3 – 0.664899 𝐿𝑁𝑀1𝑡 − 1 – 0.016300𝐿𝑁𝑀1𝑡
− 2 + 0.640081𝐿𝑁𝑀1𝑡 − 3 + 0.782655𝑀𝑌𝑅/𝐵𝐴𝐻𝑇𝑡 − 1
+ 0.176962𝑀𝑌𝑅/𝐵𝐴𝐻𝑇𝑡 − 2 – 0.103948 𝑀𝑌𝑅/𝐵𝐴𝐻𝑇𝑡 − 3
+ 1.531083𝑀𝑌𝑅/𝑈𝑆𝐷𝑡 − 1 – 9.530965𝑀𝑌𝑅/𝑈𝑆𝐷𝑡 − 2
+ 6.038574𝑀𝑌𝑅/𝑈𝑆𝐷𝑡 − 3
Based on the t-stat of each coefficient, it shows that for MYR/BAHTt-1 is significant because
the t-stat value is greater than critical value at 5% which is 1.96. However, for the remaining
independent variables the coefficients are not significant because the t-values are smaller than
the critical value at 5% level.
MYR/USD as dependent variable :
𝑀𝑌𝑅/𝑈𝑆𝐷 = 0.020756 + 0.013440𝐿𝑁𝐾𝐿𝐶𝐼𝑡 − 1 – 0.015001𝐿𝑁𝐾𝐿𝐶𝐼𝑡 − 2
+ 0.005595𝐿𝑁𝐾𝐿𝐶𝐼𝑡 − 3 – 0.006808𝐿𝑁𝐼𝑃𝐼𝑡 − 1 – 0.004885𝐿𝑁𝐼𝑃𝐼𝑡 − 2
+ 0.004176𝐿𝑁𝐼𝑃𝐼𝑡 − 3 + 0.010930 𝐿𝑁𝑀1𝑡 − 1 – 0.009903 𝐿𝑁𝑀1𝑡 − 2
+ 0.000248 𝐿𝑁𝑀1𝑡 − 3 + 0.000922𝑀𝑌𝑅/𝐵𝐴𝐻𝑇𝑡
− 1 – 0.000778𝑀𝑌𝑅/𝐵𝐴𝐻𝑇𝑡 − 2 – 0.001036 𝑀𝑌𝑅/𝐵𝐴𝐻𝑇𝑡 − 3
+ 1.215038 𝑀𝑌𝑅/𝑈𝑆𝐷𝑡 − 1 – 0.404947𝑀𝑌𝑅/𝑈𝑆𝐷𝑡 − 2
+ 0.1185124𝑀𝑌𝑅/𝑈𝑆𝐷𝑡 − 3
Based on the t-stat of each coefficient, it shows that for MYR/USDt-1 and MYR/USDt-2 are
significant because the t-stat value is greater than critical value at 5% which is 1.96. However,
for the remaining independent variables the coefficients are not significant because the t-values
are smaller than the critical value at 5% level.
5. Test the stability of your VAR results
Based on Figure 5.0, the graph shows that there is no root lies outside the unit circle, hence the VAR
satisfy the stability condition.
6. Estimate VECM based on optimum lag. Explain your VECM results.
3
By choosing the optimal lag of VAR as lag 3, therefore, our VECM lag is lag 2. Hence, we regress
LKLCI as our dependent variable and the rest are considered as independent variable. Based on the
Figure 6.0, the equation of VECM (2) for LKLCI is shown as below:
VAR (3) -> VECM (2)
∆𝐿𝑁𝐾𝐿𝐶𝐼 = −0.000690 + 0.300448∆𝐿𝑁𝐾𝐿𝐶𝐼𝑡 − 1 – 0.148911∆𝐿𝑁𝐾𝐿𝐶𝐼𝑡 − 2
+ 0.177544∆𝐿𝑁𝐼𝑃𝐼𝑡 − 1 + 0.030081∆𝐿𝑁𝐼𝑃𝐼𝑡 − 2 + 0.181879 ∆𝐿𝑁𝑀1𝑡 − 1
+ 0.161264∆ 𝐿𝑁𝑀1𝑡 − 2 + 0.005469 ∆𝑀𝑌𝑅/𝐵𝐴𝐻𝑇𝑡 − 1
− 0.005237 ∆𝑀𝑌𝑅/𝐵𝐴𝐻𝑇𝑡 − 2 + 0.480626 ∆𝑀𝑌𝑅/𝑈𝑆𝐷𝑡 − 1
+ 1.071149 ∆𝑀𝑌𝑅/𝑈𝑆𝐷𝑡 − 2 + 0.007486 𝑒𝐿𝑁𝐾𝐿𝐶𝐼, 𝑡 − 1
In the short run, based on the |t-stat| of each coefficient, it shows that for ∆LN𝐾𝐿𝐶𝐼t-1, ∆LN𝐾𝐿𝐶𝐼t-2,
∆LNIPIt-1 are significant because the |t-stat|, 3.61996, 1.7450 and 2.16740 respectively are greater than
critical value at 5% level which is 1.96. However, for the remaining independent variables, the
coefficients are not significant because the |t-stat| are smaller than the critical value at 5% level.
∆𝑳𝑵𝑰𝑷𝑰 as dependent variable:
∆𝐿𝑁𝐼𝑃𝐼 = +0.011788 + 0.112014∆𝐿𝑁𝐾𝐿𝐶𝐼𝑡 − 1 + 0.109559∆𝐿𝑁𝐾𝐿𝐶𝐼𝑡 − 2
− 0.743547∆𝐿𝑁𝐼𝑃𝐼𝑡 − 1 − 0.305791∆𝐿𝑁𝐼𝑃𝐼𝑡 − 2 − 0.598563 ∆𝐿𝑁𝑀1𝑡 − 1
− 0.242917∆ 𝐿𝑁𝑀1𝑡 − 2 + 0.0025822 ∆𝑀𝑌𝑅/𝐵𝐴𝐻𝑇𝑡 − 1
− 0.012108 ∆𝑀𝑌𝑅/𝐵𝐴𝐻𝑇𝑡 − 2 − 0.490652 ∆𝑀𝑌𝑅/𝑈𝑆𝐷𝑡 − 1
+ 1.457166 ∆𝑀𝑌𝑅/𝑈𝑆𝐷𝑡 − 2 + 0.006415 𝑒𝐿𝑁𝐼𝑃𝐼, 𝑡 − 1
Based on the t-values for this equation above, only coefficient for ∆LNIPIt-1, ∆LNIPIt-2 and ∆LNM1t-
1are considered as significant because its |t-stat|, 9.55433, 4.35288, 4.95285 are greater than the critical
value at 5% level. In contrast, the coefficient for other independent variables are not significant as the
|t-stat| are smaller than critical value at 5% level.
∆𝑳𝑵𝑴𝟏 as dependent variable:
∆𝐿𝑁𝑀1 = +0.010761 + 0.069008∆𝐿𝑁𝐾𝐿𝐶𝐼𝑡 − 1 + 0.072085∆𝐿𝑁𝐾𝐿𝐶𝐼𝑡 − 2
− 0.032197∆𝐿𝑁𝐼𝑃𝐼𝑡 − 1 − 0.026489∆𝐿𝑁𝐼𝑃𝐼𝑡 − 2 − 0.093179 ∆𝐿𝑁𝑀1𝑡 − 1
− 0.133892∆ 𝐿𝑁𝑀1𝑡 − 2 + 0.003759∆𝑀𝑌𝑅/𝐵𝐴𝐻𝑇𝑡 − 1
+ 0.011818∆𝑀𝑌𝑅/𝐵𝐴𝐻𝑇𝑡 − 2 − 0.845527 ∆𝑀𝑌𝑅/𝑈𝑆𝐷𝑡 − 1
− 0.285239 ∆𝑀𝑌𝑅/𝑈𝑆𝐷𝑡 − 2 − 0.006187 𝑒𝐿𝑁𝑀1, 𝑡 − 1
Based on this equation, we found that all coefficients are not significant as the |t-stat| are smaller than
the critical level at 5% level.
∆𝑴𝒀𝑹/𝑩𝑨𝑯𝑻 as dependent variable:
4
∆𝑀𝑌𝑅/𝐵𝐴𝐻𝑇 = +0.012650 − 1.065520∆𝐿𝑁𝐾𝐿𝐶𝐼𝑡 − 1 − 0.783948∆𝐿𝑁𝐾𝐿𝐶𝐼𝑡 − 2
+ 0.786943∆𝐿𝑁𝐼𝑃𝐼𝑡 − 1 + 0.563205∆𝐿𝑁𝐼𝑃𝐼𝑡 − 2 − 0.701384 ∆𝐿𝑁𝑀1𝑡 − 1
− 0.655181∆ 𝐿𝑁𝑀1𝑡 − 2 − 0.085242∆𝑀𝑌𝑅/𝐵𝐴𝐻𝑇𝑡 − 1
+ 0.094502∆𝑀𝑌𝑅/𝐵𝐴𝐻𝑇𝑡 − 2 + 4.960383 ∆𝑀𝑌𝑅/𝑈𝑆𝐷𝑡 − 1
− 3.920100 ∆𝑀𝑌𝑅/𝑈𝑆𝐷𝑡 − 2 + 0.068208 𝑒𝐿𝑁𝑀1, 𝑡 − 1
Based on the t-values for this equation above, only coefficient for ∆LNKLCIt-1 is considered as
significant because the |t-stat|, 2.46397 is greater than the critical value at 5% level. In contrast, the
coefficient for other independent variables are significant as the t-values are greater than critical value
at 5% level.
∆𝑴𝒀𝑹/𝑼𝑺𝑫 as dependent variable:
∆𝑀𝑌𝑅/𝑈𝑆𝐷 = −0.123932 + 0.009828∆𝐿𝑁𝐾𝐿𝐶𝐼𝑡 − 1 − 0.003751∆𝐿𝑁𝐾𝐿𝐶𝐼𝑡 − 2
+ 0.003536∆𝐿𝑁𝐼𝑃𝐼𝑡 − 1 − 0.002078∆𝐿𝑁𝐼𝑃𝐼𝑡 − 2 + 0.009675 ∆𝐿𝑁𝑀1𝑡 − 1
+ 0.000171∆ 𝐿𝑁𝑀1𝑡 − 2 + 0.002087∆𝑀𝑌𝑅/𝐵𝐴𝐻𝑇𝑡 − 1
+ 0.001236∆𝑀𝑌𝑅/𝐵𝐴𝐻𝑇𝑡 − 2 + 0.278582 ∆𝑀𝑌𝑅/𝑈𝑆𝐷𝑡 − 1
− 0.123932 ∆𝑀𝑌𝑅/𝑈𝑆𝐷𝑡 − 2 + 0.000812 𝑒𝑀𝑌𝑅/𝑈𝑆𝐷1, 𝑡 − 1
Based on the t-values for this equation above, only coefficient for ∆MYR/USDt-1 is considered as
significant because the |t-stat|, 3.27501 is greater than the critical value at 5% level. In contrast, the
coefficient for other independent variables are significant as the t-values are greater than critical value
at 5% level.
7. Test long run Granger Causality using VAR. Explain
We regress LNKLCI (yt) on lagged LNKLCIterms and obtain the RSS of this regression and
label it as [Link] weRegress LNKLCI (yt)on lagged LNKLCIterms plus laggedLNIPI(x) and
obtain the RSS and label it as [Link] 2 are repeated with other lagged x variables which is
lagged MYR/USD, MYR/BAHT and LNM1. Then, we calculate the F statistic for the normal
Wald test on coefficient restrictions using the formula F-stat:
(RSSr−RSSu)/m
𝐹= Where k= m+n+1
m RSSu/(n−k)
All the steps are repeated with LNIPI, MYR/USD, MYR/BAHT and LNM1 as dependent
variables and the results are tabulate as follows:
Variables RSSr RSSu F-Stat
LNKLCI 0.223681 - -
LNKLCI_LNIPI - 0.231769 2.007507
LNKLCI_MYR/USD - 0.002022 6285.088691
LNKLCI_MYR/BAHT - 6.112955 22.235432
LNKLCI_LNM1 - 0.083688 95.90700379
5
LNIPI 0.246479 - -
LNIPI_LNKLCI - 0.216509 7.936298
LNIPI_MYR/USD - 0.002097 6681.561519
LNIPI_MYR/BAHT - 6.140839 55.032107
LNIPI_LNM1 - 0.081208 116.682314
MYR/USD 0.002108 - -
MYR/USD_LNKLCI - 0.223213 56.791883
MYR/USD_LNIPI - 0.241023 56.831892
MYR/USD_MYR/BAHT - 6.388073 57.314414
MYR/USD_LNM1 - 0.083708 55.889521
MYR/BAHT 6.791112 - -
MYR/BAHT_LNKLCI - 0.220540 1708.138182
MYR/BAHT_LNIPI - 0.235963 1592.743535
MYR/BAHT_MYR/USD - 0.002048 190058.432300
MYR/BAHT_LNM1 - 0.083066 4629.988652
LNM1 0.084172 - -
LNM1_LNKLCI - 0.207518 34.078187
LNM1_LNIPI - 0.211527 34.520076
LNM1_MYR/USD - 0.002070 2274.006440
LNM1__MYR/BAHT - 6.132156 56.546357
Table 2.0
LNKLCI as dependent variable:
𝐻𝑜 : 𝜑1 𝐿𝑁𝐼𝑃𝐼1 = 𝜑2 𝐿𝑁𝐼𝑃𝐼2 = 𝜑3 𝐿𝑁𝐼𝑃𝐼3 = 𝜑1 𝑀𝑌𝑅/𝑈𝑆𝐷1 = 𝜑2 𝑀𝑌𝑅/𝑈𝑆𝐷2 =
𝜑3 𝑀𝑌𝑅/𝑈𝑆𝐷3 = 𝜑1 𝑀𝑌𝑅/𝐵𝐴𝐻𝑇1 = 𝜑2 𝑀𝑌𝑅/𝐵𝐴𝐻𝑇2 = 𝜑3 𝑀𝑌𝑅/𝐵𝐴𝐻𝑇3 =
𝜑1 𝐿𝑁𝑀11 = 𝜑2 𝐿𝑁𝑀12 = 𝜑3 𝐿𝑁𝑀13 = 0
𝐻1 : 𝜑1 𝐿𝑁𝐼𝑃𝐼1 = 𝜑2 𝐿𝑁𝐼𝑃𝐼2 = 𝜑3 𝐿𝑁𝐼𝑃𝐼3 = 𝜑1 𝑀𝑌𝑅/𝑈𝑆𝐷1 = 𝜑2 𝑀𝑌𝑅/𝑈𝑆𝐷2 =
𝜑3 𝑀𝑌𝑅/𝑈𝑆𝐷3 = 𝜑1 𝑀𝑌𝑅/𝐵𝐴𝐻𝑇1 = 𝜑2 𝑀𝑌𝑅/𝐵𝐴𝐻𝑇2 = 𝜑3 𝑀𝑌𝑅/𝐵𝐴𝐻𝑇3 =
𝜑1 𝐿𝑁𝑀11 = 𝜑2 𝐿𝑁𝑀12 = 𝜑3 𝐿𝑁𝑀13 ≠ 0
HO: the coefficients of the lagged terms of LNIPI, MYR/USD, MYR/BAHT, LNM1 are equal
to zero
H1:the coefficients of the lagged terms ofLNIPI, MYR/USD, MYR/BAHT, LNM1are not
equal to zero
6
F-critical = F3,175, at 5% significant level is equal to 2.60
Since the value of computed f-stat,6285.088691,22.235432 and 95.90700379 are greater than
f-critical, then MYR/USD, MYR/BAHT and LNM1 granger cause LNKLCI. However, for
LNIPI the value of f-stat is lower than f-critical,2.007507<2.60. Thus, LNIPI does not granger
cause LNKLCI.
LNIPI as dependent variable:
HO: the coefficients of the lagged terms of LNKLCI, MYR/USD, MYR/BAHT, LNM1 are
equal tozero
H1: the coefficients of the lagged terms of LNKLCI, MYR/USD, MYR/BAHT, LNM1are not
equal to zero
F-critical = F3,175, at 5% significant level is equal to 2.60
Based on table above, the value of computed f-statare greater than f-critical, then LNKLCI,
MYR/USD, MYR/BAHT and LNM1 granger cause LNIPI.
MYR/USD as dependent variable:
HO: the coefficients of the lagged terms of LNKLCI, LNIPI, MYR/BAHT, LNM1 are equal
tozero
H1: the coefficients of the lagged terms of LNKLCI, LNIPI, MYR/BAHT, LNM1are not equal
to zero
F-critical = F3,175, at 5% significant level is equal to 2.60
Based on table above, the value of computed f-statare greater than f-critical, then LNKLCI,
LNIPI, MYR/BAHT and LNM1 granger cause MYR/USD.
MYR/BAHT as dependent variable:
HO: the coefficients of the lagged terms of LNKLCI, LNIPI, MYR/USD, LNM1 are equal
tozero
H1: the coefficients of the lagged terms of LNKLCI, LNIPI, MYR/USD, LNM1are not equal
to zero
F-critical = F3,175, at 5% significant level is equal to 2.60
Based on table above, the value of computed f-statare greater than f-critical, then LNKLCI,
LNIPI, MYR/USD and LNM1 granger cause MYR/BAHT.
LNM1 as dependent variable:
HO: the coefficients of the lagged terms of LNKLCI, LNIPI, MYR/USD, MYR/BAHTare
equal tozero
H1: the coefficients of the lagged terms of LNKLCI, LNIPI, MYR/USD, MYR/BAHTare not
equal to zero
7
F-critical = F3,175, at 5% significant level is equal to 2.60
Based on table above, the value of computed f-stat are greater than f-critical, then LNKLCI,
LNIPI,MYR/USD and MYR/BAHT granger cause LNM1.
8. Test long run and Short Run Granger Causality using VECM. Explain
1
Using Formula: 𝑆𝑝𝑒𝑒𝑑 𝑎𝑑𝑆𝑝𝑒𝑒𝑑 𝑎𝑑𝑗𝑢𝑠𝑡𝑚𝑒𝑛𝑡= |α|
From Figure 8.1 we focus on |t-stat|, the ECT as table below:-
INDEPENDENT LONG RUN 𝑺𝑷𝑬𝑬𝑫
VARIABLE |𝒕𝒔𝒕𝒂𝒕| 𝑬𝑪𝑻,|𝜶| 𝑨𝑫𝑱𝑼𝑺𝑻𝑴𝑬𝑵𝑻,𝟏|𝜶|
LNKLCI 2.13357 0.007486 133.583
LNIPI 1.92447 0.006415 155.885
LNM1 2.91204 0.006187 161.629
MYR_BAHT 3.73120 0.068208 14.6610
MYR_USD 2.37931 0.000812 1231.52
Table.3.0
𝐻0: 𝑇ℎ𝑒 𝑑𝑒𝑝𝑒𝑛𝑑𝑒𝑛𝑡 𝑑𝑜𝑒𝑠 𝑛𝑜𝑡 𝑐𝑜𝑛𝑖𝑛𝑡𝑒𝑔𝑟𝑎𝑡𝑒𝑑
𝐻1: 𝑇ℎ𝑒 𝑑𝑒𝑝𝑒𝑛𝑑𝑒𝑛𝑡 𝑖𝑠 𝑐𝑜𝑛𝑖𝑛𝑡𝑒𝑔𝑟𝑎𝑡𝑒𝑑
For the long run, we concluded for the t-value as below:-
For LNKLCI as dependent, the t-value for long run is greater than 1.96 and it is significant.
Hence the dependent reject 𝐻0. Thus, LNKLCI as dependent is conintegrated.
For LNIPI as dependent, the t-value for long run is less than 1.96 and it is significant.
Hence the dependent failed to reject 𝐻0. Thus, LNIPI as dependent is not conintegrated.
For LNM1 as dependent, the t-value for long run is greater than 1.96 and it is significant. Hence
the dependent reject 𝐻0. Thus, LNIPI as dependent is conintegrated.
For MYR_BAHT as dependent, the t-value for long run is greater than 1.96 and it is significant.
Hence the dependent reject 𝐻0. Thus, MYR_BAHT as dependent is conintegrated.
For MYR_USD as dependent, the t-value for long run is greater than 1.96 and it is significant.
Hence the dependent reject 𝐻0. Thus, MYR_USD as dependent is conintegrated.
Below we concluded the speed adjustment as in table 4 for every variable:-
• • For the speed adjustment of LNKLCI as dependent, there is short run disturbance, it will take
about 133 months or 11 years for short run disturbance to go back to long run equation.
• • For the speed adjustment of LNIPI as dependent, there is short run disturbance, it will take
about 155 months or 12 and quarter years for short run disturbance to go back to long run equation.
8
• • For the speed adjustment of LNM1 as dependent, there is short run disturbance, it will take
about 161 months or 13 years for short run disturbance to go back to long run equation.
• • For the speed adjustment of MYR_BAHT as dependent, there is short run disturbance, it will
take about 14 months for short run disturbance to go back to long run equation.
• • For the speed adjustment of MYR_USD as dependent, there is short run disturbance, it will
take about 1231 months or 102 years for short run disturbance to go back to long run equation.
Short Run Causality Test
For LKLCI :
𝐻01 : 𝑎15 = 𝛽15 = 𝛾15 = 𝜋15 = 𝜔15 = 0
𝐻11 : 𝑎𝑡 𝑙𝑒𝑎𝑠𝑡 𝑜𝑛𝑒 𝑜𝑓 𝑐𝑜𝑒𝑓𝑓𝑖𝑐𝑖𝑒𝑛𝑡𝑠 ≠ 0
From figure 8.1, P-value for LNKLCI is greater than 5% and it is fail to reject the null
hypothesis. Hence LNKLCI does not granger cause of LNIPI, LNM1, MYR/BAHT,
MYR/USD.
For LNIPI:
𝐻20 : 𝑎25=𝛽25=𝛾25=𝜋25=𝜔25=0
𝐻21 :𝑡 𝑙𝑒𝑎𝑠𝑡 𝑜𝑛𝑒 𝑜𝑓 𝑐𝑜𝑒𝑓𝑓𝑖𝑐𝑖𝑒𝑛𝑡𝑠 ≠0
From figure 8.1, P-value for LNIPI is less than 5% and it is reject the null hypothesis. LNIPI
have granger cause of LNKLCI, LNM1, MYR/BAHT. MYR/USD.
For LNM1:
𝐻30 : 𝑎35=𝛽35=𝛾35=𝜋35=𝜔35=0
𝐻31 :𝑡 𝑙𝑒𝑎𝑠𝑡 𝑜𝑛𝑒 𝑜𝑓 𝑐𝑜𝑒𝑓𝑓𝑖𝑐𝑖𝑒𝑛𝑡𝑠 ≠0
From figure 8.1, P-value for LNM1 is greater than 5% and it is fail to reject the null hypothesis.
Hence LNM1 does not granger cause of LKLCI, LNIPI, MYR/BAHT, MYR/USD.
For MYR/BAHT:
𝐻40 : 𝑎45=𝛽45=𝛾45=𝜋45=𝜔45=0
𝐻41 :𝑡 𝑙𝑒𝑎𝑠𝑡 𝑜𝑛𝑒 𝑜𝑓 𝑐𝑜𝑒𝑓𝑓𝑖𝑐𝑖𝑒𝑛𝑡𝑠 ≠0
From figure 8.1, P-value for MYR/BAHT is less than 5% and it is reject the null hypothesis.
MYR/BAHT have granger cause of LNKLCI, LNIPI,LNM1, MYR/USD.
For MYR/USD:
𝐻50 : 𝑎55=𝛽55=𝛾55=𝜋55=𝜔55=0
𝐻15 :𝑡 𝑙𝑒𝑎𝑠𝑡 𝑜𝑛𝑒 𝑜𝑓 𝑐𝑜𝑒𝑓𝑓𝑖𝑐𝑖𝑒𝑛𝑡𝑠 ≠0
From figure 8.1, P-value for MYR/USD is greater than 5% and it is fail to reject the null
hypothesis. MYR/USD does not granger cause of LNKLCI, LNIPI, LNM1, MYR/BAHT,
MYR/USD.
9
From figure 8.1 and above can be summarized as all null hypothesis if both null hypotheses
are failed to reject except for LNIPI and MYR/BAHT as variable. Hence there is no
bidirectional short-run causality among all variables.
9. Provide short run and long run causality table. Explain
Independent Dependent Variable
Variable LNKLCI LNIPI MYR/USD MYR/BAHT LNM1
LNKLCI - 4.731351 1.546201 11.16663 4.720320
LNIPI 6.111431 - 0.712087 3.566692 0.479779
Short MYR/USD 2.064155 3.152952 - 1.660722 3.197453
Run MYR/BAHT 0.300781 4.540834 2.607068 - 1.901353
LNM1 3.121947 26.41460 0.616312 1.784496 -
Long- ECT 0.007486 0.006415 0.000812 0.068208 -
run 0.006187
Table 4.0
The positive ECT term signifies the leader in system are the LNKLCI, LNIPI, MYR/USD,
and BAHT. While LNM1 in the system is negative ECT term reflects follower to the system
or in other word, LNM1 following a short-term deviation and is suggest that these variables
are responsible for adjustment to long run equilibrium.
Question 10
Based on the results, it shows that all the variables are stationary at the first difference stage.
For the long run using VAR results it shows that, MYR/USD, MYR/BAHT and LNMI granger cause
LNKLCI but LNIPI does not granger causes LNKLCI. If LNIPI is the dependent, there is granger cause
LNKLCI. This result was supported by Nwaolisa and Chijindu (2016). This means that there exists
unidirectional relationship between LNIPI and LNKLCI using Granger Causality Test on VAR. The
authors suggest that some of the strict market entry requirements should be relaxed to ensure that more
businesses are listed as the largest user of external funds and grow faster in countries with a high level
of stock market development and an efficient legal system. Furthermore, it should be given some
relaxation to small and medium-sized and provide them with the opportunity to expand their operations.
In this case, the market capitalization are further enhance automatically will contribute positively to
industrial [Link] another variables which are MYR/BAHT, MYR/USD and LNMI are
bidirectional causality with LNKLCI and LNIPI. This was supported by (Hussain and Sen Liew, 2004;
Seong, 2013). According to the study of (Husin et. al, 2012), there is a long term causality of stock
market in Malaysia with IPI an MYR. This was supported by our result which LNKLCI variable in the
equation bears the burden of dispersed error correction of short term balance to achieve long term
balance as 133 months and demonstrates endogenues of towards the formed [Link], our
10
result indicates that there is no bidirectional causality among the variables in short term. This was
supported by (Ahmad et. al, 2009).
11