In this section, I will estimate the gravity models above, including the benchmark gravity, three-way fixed effects, two-way fixed effects, and generalize fixed effects models. Table 13 shows the main estimation results. Column one presents the result of the PCS model, columns two and three present the results of three-way effect model with time-invariant, and without time-invariant respectively, and column four shows for
provides the adjusted R2 (commonly as the empirical power), and the information criteria AIC and SIC (to indicate an appropriate model with the smallest value).
According to the result in the first column of Table 13, a 10% rise in exporter’s GDP should be associated with a 11.7% rise in exports. If an importer’s GDP rises by 10%, exports should rise by 7.95%. Holding the other condition to be the same, if an exporter’s population rises by 10%, exports should decrease by 0.66%, while an importer’s population increases by 10%, its exports should increase 1.04%. Also as shown in column one of Table 13, a country will export 88.6% less to a market that is double as distant as another identical market. Exports would be higher by 74.8% than the other trade flows if trading partners use the same language, while exports should be higher by 28.7% if trading partners share a common border. Trade bloc dummy variables are used to evaluate the effect of free trade agreements. In column one of Table 13, the coefficient on AFTAijt and MERCOSURijt are positive and significant statistically, indicating that bilateral trade in the AFTAijt and MERCOSURijt countries is more intense than the other countries. In contrast, for the EUijt, the PCS model predicts a -60.8% lower of bilateral trade if both exporting and importing countries are the EUijt
members. The PCS model also indicates that intra-EURO trade is about 11.6% higher than the average trade flow, but the effect is statistically insignificant. Intra-NAFTA trade is about -13.2% lower than the average trade flow, and statistically significant.
The second column of Table 13 reports the estimation result for the three-way fixed model with time-invariant variables. According the result, a 10% rise in exporter’s GDP should be associated with a 13.16% rise in exports. If an importer’s GDP rises by 10%, exports should rise by 12.78%. Holding the other condition to be the same, if an exporter’s population rises by 10%, exports should increase by 0.9%, while an importer’s population increases by 10%, its exports should increase 0.9%. Also as shown in column two of Table 13, a country will export 130% less to a market that is
double as distant as another identical market. Exports would be higher by 76.5% than the other trade flows if trading partners use the same language, while exports should be lower by 28% if trading partners share a common border.
Column two of Table 13 indicates that the coefficient on NAFTAijt and
MERCOSUR
ijt are positive and significant statistically, i.e., bilateral trade flows in theNAFTA
ijt and MERCOSURijt countries are more intense than the other countries. In contrast, for the EUijt, the three-way fixed model predicts a -43% lower of bilateral trade if both exporting and importing countries are the EUijt members. The three-way fixed model also indicates that intra-EURO trade is about 12.8% higher than the average trade flow, and intra-AFTA trade is about 6.6% higher than the average trade flow, and statistically. However, both of their effects are statistically insignificant.The third column of Table 13 reports the estimation result for three-way fixed effects model without time-invariant. Specifically, it suggests that acting in EUijt tends to an increase in trade of 100.1%, and that NAFTAijt and AFTAijt led to increase in trade of greater than 200%, and that MERCOSURijt
causes members to trade greater than
400% than an average trade. The estimated effect of the NAFTAijt using the three-way fixed effects model without time-invariant indicates that there is something special about the relationship between Mexico, the United States and Canada that makes them trade relatively less with each other than the gravity variables would predict.The fourth column of Table 13 reports the estimation result for model FE2. As shown in the result, a 10% rise in exporter’s GDP should be associated with a 12.9%
rise in exports. If an importer’s GDP rises by 10%, exports should rise by 12.48%.
Holding the other conditions to be the same, if an exporter’s population rises by 10%, exports should increase by 1.59%, while an importer’s population increases by 10%, and its exports should increase 3.54%. With regarding the REI effect on trade flows, the
contrast, the coefficients on EUijt and AFTAijt are statistically insignificant.
The last column of Table 13 reports the estimation results for the FEG model.
According to the result, a 10% rise in exporter’s GDP should be associated with a 12.9% rise in exports. If an importer’s GDP rises by 10%, exports should rise by 12.48%. Holding the other condition to be the same, if an exporter’s population rises by 10%, exports should increase by 1.59%, while an importer’s population increases by 10%, its exports should increase 3.54%. The REA coefficients on bilateral trade are the same as those of model FE2. According to the estimation result, models FE 2 and FEG have high explanatory power of the adjusted R2. Also, from the estimation result, models FE2 and FEG are identical, except for the dummy variables which indicate specific trading pairs and trading countries. Based on information criterion BIC, model FE2 is preferred.