where the Breusch-Godfrey Lagrange multiplier test is unable to reject the null of no
autocorrelation with lag order 2 at the 5% significance level. The results of the shortrun
Granger causality test are reported in Table 6. This causality test is conducted
with the F-test based on the null hypothesis of all lagged first differences of
independent variable are jointly insignificant. These results suggest that there is a
unidirectional short-run Granger causality running from GDPP to OFDI.
The results of long-run Granger causality test are reported in Table 7. This
causality test is conducted with the t-test based on the null hypothesis of the
coefficient of error correction term is equal to zero. The coefficient of error
correction term of equation (5) or equation (6) is statistically different from zero at
5% significance level. They also have the correct negative sign. This suggests that
there is a bidirectional long-run Granger causality between GDPP and OFDI.
In this study, I also use the approaches of Ram (1988) and Zhang (2001) to
determine the sign of short-run Granger causality. The sign of short-run Granger
causality from an independent variable to the dependent variable is determined by
adding up the coefficients of all lagged first differences of the independent variable. If
the summation of these coefficients provides a positive value, the independent
variable has positive effects on the dependent variable in the short-run. The selected
value of n is 1 for equation (6). Therefore, only one lagged first difference of GDPP is
introduced in equation (6). The estimated coefficient of DGDPt1 in this equation is
0.000154 suggesting that the sign of this unidirectional short-run Granger causality
running from GDPP to OFDI is positive.