Results

Table 4 [ PDF 44.8KB | 1 page ] presents our estimation results of equation (16), which are ordinary least squares (OLS) coefficient estimates with robust standard errors in columns 1 to 5 and instrumental variables (IV) estimates in column 6. Column 1 includes the independent variables that are generally used in gravity equations. Column 2 adds import distance lnImDist_i as an independent variable. Column 3 includes the dummy variable Eastj and the interaction terms. Columns 4 and 5 show the estimation results for the subsamples of East Asia and OECD countries, respectively; whereas column 6 includes the full sample for the IV estimation.

The results generally provide support for Hypotheses 1 and 2. Let us first examine the evidence for Hypothesis 1. Specifically, in column 2, both coefficients on import distance and export distance are negative and statistically significant. In column 3, the coefficient on import distance remains negative and statistically significant, but the coefficient on export distance is positive. Comparing columns 4 and 5, the results suggest that import and export distance both have a negative impact on processing exports to East Asian economies; import distance has a negative impact on processing exports to non-Asian OECD countries, but export distance is an insignificant determinant of processing exports to non-Asian OECD countries.

We also find strong evidence in favor of Hypothesis 2. Specifically, in column 3 the coefficient on East_j∗lnExDist_i is negative and statistically significant, while the coefficient on East_j∗lnExDist_it is positive and statistically significant. In line with Hypothesis 2, this suggests that processing exports destined for East Asian economies are more sensitive to export distance and less sensitive to import distance than processing exports destined for non- Asian OECD countries. The coefficients on import and export distance in columns 4 and 5 are also in line with Hypothesis 2. Namely, the absolute value of the coefficient on lnExDist is larger for processing exports to East Asian economies than to non-Asian OECD countries; whereas the absolute value of the coefficient on lnImDist is smaller for processing exports to East Asian economies than to non-Asian OECD countries.

The other coefficients in columns 2 to 5 all take on the expected signs. Processing exports are larger for more populated provinces and those with higher GDP per capita, higher wages, and lower internal distance. In addition, processing exports are greater for destinations that are more populated and have a higher GDP per capita. Finally, the positive coefficient on the dummy variable Eastj suggests that there are extra trade costs related to interregional trade.

By comparing the other coefficients in columns 4 and 5, we are able to identify additional differences between processing exports to East Asian versus non-Asian OECD economies. The results show that the coefficients on lncGDPpcit, lnGDPpcjt, lncPopit, lnPopjt, lnWageit, and lnIDisti are in absolute value all larger for the subsample of non-Asian OECD countries than for the subsample of East Asian economies.¹⁴ This implies that processing exports to non- Asian OECD countries tend to concentrate in more populated provinces with higher GDP per capita, higher wages, and smaller internal distance than processing exports to East Asian economies. Moreover, processing exports to non-Asian OECD countries are mainly shipped to richer and larger destination markets.

Download this Paper [ PDF 340.3KB| 28 pages ].

Comment(s)

There are [0] comment(s) for this entry. Post a comment.