Page 15 - AEI Insights 2018 Vol 4 Issue 1

AEI Insights, Vol 4, Issue 1, 2018

variables at first difference could be considered as nonstationary process. This means that these
variables are integrated of order one, I(1), implying that standard statistical methods, such as
linear regression, can be used efficiently to examine the relationship among these variables.

In the second stage of empirical analysis, the Johansen cointegration test was used to examine
the long-run relationship between exports and economic development in Singapore. Empirical
findings from the Johansen cointegration test for the analysis of exports to China (EXPC) and
economic development (GDP) are reported in Table 2. As shown in the table, the maximum
eigenvalue statistic for the first null hypothesis of no cointegrating equation is greater than the
five percent critical value. This means that the Johansen test rejected the first null hypothesis
and indicated the presence of one cointegrating equation. Furthermore, the maximum
eigenvalue statistic for the second null hypothesis of one cointegrating equation is also more
than the five percent critical values. This indicates that the Johansen test failed to reject the
second null hypothesis and confirm the presence of two cointegrating equations. In other
words, the results of the Johansen cointegration analysis suggested that there would be no
significant cointegrating relationship between exports to China and economic development in
Singapore.

Table 2: Results of Johansen test (EXPC and GDP)

Number of cointegrating Eigenvalue Maximum 5 percent critical
equation Eigenvalue value
None 0.097 17.146* 14.264
At most 1 0.021 4.239* 3.841
Notes: ** indicates the significant level at the 5 percent

The results of the Johansen cointegration test for the analysis of exports to EU (EXPEU) and
economic development (GDP) are reported in Table 3. The maximum eigenvalue statistic for
the first null hypothesis of no cointegrating equation is smaller than the five percent critical
value. This means that the Johansen test rejected the first null hypothesis and indicated the non-
existence of one cointegrating equation. Furthermore, the maximum eigenvalue statistic for the
second null hypothesis of one cointegrating equation is also more than the five percent critical
values. This indicates that the Johansen test failed to reject the second null hypothesis and show
the presence of two cointegrating equations. In other words, the results of the Johansen
cointegration analysis indicated that there would be no significant cointegrating relationship
between exports to EU and economic development in Singapore.

Table 3: Results of Johansen test (EXPEU and GDP)

Number of cointegrating Eigenvalue Maximum 5 percent critical
equation Eigenvalue value
None 0.044 7.654 14.264
At most 1 0.025 4.409* 3.841
Notes: ** indicates the significant level at the 5 percent

In the final stage of empirical analysis, the Granger causality test was used to examine the
causal relationship between exports to China (EXPC) and economic development (GDP). The
empirical findings from the Granger causality test are reported in Table 4. The results of the
test rejected the null hypothesis of no causality between exports to China (EXPC) and economic

15