The optimal lags of each ARDL model are selected based on SIC. Because of
relatively small sample size and annual data used in this study, the maximum possible
values of p and q in equation (3) and of r and s in equation (4) are set at 4. The
selected values of p, q, r and s are 1, 0, 2 and 2, respectively. Then, each of
these ARDL models are reparameterized to derive the long-run coefficients. The
long-run coefficients after the reparameterization are reported in Table 5. The
statistically significant positive long-run coefficient of independent variable indicates
GDPP has positive effects on OFDI. Similarly, OFDI has positive impacts on GDPP.
The existence of co-integrating relations between OFDI and GDPP suggests the
inclusion of an error correction term into a VAR for the testing of Granger causality.
The Granger causality tests are implemented based on the error correction models
shown here:
The optimal lags of each ARDL model are selected based on SIC. Because ofrelatively small sample size and annual data used in this study, the maximum possiblevalues of p and q in equation (3) and of r and s in equation (4) are set at 4. Theselected values of p, q, r and s are 1, 0, 2 and 2, respectively. Then, each ofthese ARDL models are reparameterized to derive the long-run coefficients. Thelong-run coefficients after the reparameterization are reported in Table 5. Thestatistically significant positive long-run coefficient of independent variable indicatesGDPP has positive effects on OFDI. Similarly, OFDI has positive impacts on GDPP.The existence of co-integrating relations between OFDI and GDPP suggests theinclusion of an error correction term into a VAR for the testing of Granger causality.The Granger causality tests are implemented based on the error correction modelsshown here:
การแปล กรุณารอสักครู่..
