Page 222 - Contributed Paper Session (CPS) - Volume 8

CPS2256 Norzarita Samsudin
variable) in the VAR when a shock is applied to the error term. Variance
Decomposition technique separates the variation in the endogenous variable
into the component shocks to the VAR. Thus, the variance decomposition
provides information about the relative importance of each random shock in
affecting the variables in VAR (Ogungbenle, Olawumi and Obasuyi, 2013).

4. Discussion and Conclusion
Correlation analysis is conducted between Malaysia real GDP and sales
value of retail trade. The studied variables has very strong positive relationship
(r = +0.96). Unit root test is conducted to ascertain level of integration of the
studied variables. It is found that both the real GDP and sales value of retail
trade are stationary at first difference. The summary is in Table 1.

Table 1: Augmented Dickey-Fuller Test Statistic
Variable Stationarity t-Statistic Prob.
Gross Domestic Product First difference 7.15 0.0000
Sales value of retail trade First difference -6.34 0.0000
Source: Author’s computation

Akaike Information Criterion (AIC) is used to identify optimal lag. The
optimal lag is Lag one. The summary result is shown in Table 2 below.

Table 2: Optimal Lag: Akaike Information Criterion (AIC)

43.59738 Lag 0
36.74791* *Lag 1
36.91179 Lag 2
Source: Author’s computation

The next step is to find out whether the real GDP and sales value of retail trade
are integrated in the long run.

4.1 Johansen Cointegration Test
Johansen Cointegration Test is used to determine the long run
cointegration of the studied varibles. Trace test and Max-eigenvalue test
indicated that there are no cointegration between real GDP and sales value
of retail trade at the 0.05 level. The result of Johansen co-integration test
is presented in Table 3.

211 | I S I W S C 2 0 1 9