Trade Linkages between the PRC and Latin America: A Gravity Analysis

In empirical trade studies, the gravity model has been the workhorse for analyzing the determinants of international trade flows. In its most basic form, the gravity model posits that trade increases with the economic size of the trading partners and decreases with the distance between them. The basic gravity equation is often augmented with a number of country-specific variables to capture the trade effects of geographic, cultural, and economic proximity. The theoretical underpinnings of the gravity model have been validated in recent literature.⁴

For the purpose of our analysis, we adopted the following specification of gravity equations:

where X_ijt is the exports of country i to country j in year t; GDP_it is the real GDP of country i in year t; GDPPC_it is per capita real GDP of country i in year t; D_ij is the physical distance between two countries, i and j; and Z_ij is a vector of a number of other time invariant countrypair dummy variables that previous studies have found significant in explaining trade flows— such as shared borders, common languages, colonial ties, and common free trade areas.

As we were concerned with the trade performance of the PRC and Latin America, we introduced a set of regional and region-pair dummies to represent the different components of the two regions' trade. CHN_E is a dummy for capturing the PRC as an exporter, while CHN_M is a dummy for the PRC as an importer. Similarly, LAC_E and LAC_M are regional dummy variables for Latin America and the Caribbean as an exporter and an importer, respectively. Because of the heterogeneity among LAC countries in their economic structures and factor endowments, we have distinguished two subregions of Latin America by using several dummy variables to capture differential export relations of the LAC subregions vis-à-vis the PRC. Specifically, CAC_CHN is the bilateral dummy for the exports of Central America (including Mexico) and the Caribbean to the PRC, and SA_CHN is the dummy for South America's exports to the PRC.⁵ The bilateral dummy CHN_LAC denotes exports from the PRC to Latin America.⁶

Equation (1) was estimated on a panel of 161 countries from 2000 to 2006. The data sources and variables are described in Appendix II [ PDF 12.3KB | 1 page ]. We used pooled ordinary least squares, including fixed time effects to account for common shocks affecting all countries over time. The estimation results reported in the first column of Table 2 [ PDF 14.2KB | 1 page ] indicate that the standard gravity model in Equation (1) can explain nearly 70% of the variation in bilateral trade flows. As expected, economic size has a highly significant and positive impact on bilateral trade. The coefficients on exporters' and importers' GDP are close to 1, consistent with the unitary income elasticities for both importers and exporters implied by theoretical gravity models such as Anderson and van Wincoop (2003). Per capita income has a positive impact on trade volume, suggesting that high-income countries are more dependent on trade than lowincome countries. A 1% increase in bilateral distance decreases bilateral trade by 1%. Shared borders, common languages, colonial ties, and free trade agreements all have a significant and positive impact on trade.

With respect to the trade performance of the PRC and Latin America, the estimation results show that the PRC's trade over-performed the world average in terms of its overall trade, and its export performance was stronger than its import performance. In contrast, Latin America under-traded in terms of its exports, compared with the world average. Its import performance was also slightly less than the world average, but the coefficient estimate is not statistically significant. Even controlling for the PRC as an over-performer in trade, the coefficient for its exports to Latin America is still positive, suggesting the PRC's strong export performance in the LAC region. For Latin America's exports to the PRC, the performance is mixed. Although on the whole Latin America and the Caribbean was an under-performer in exports, South America had strong export performance in the PRC, while Central America had weak export performance there. The coefficient suggests that the export of Central America and the Caribbean to the PRC was about 60% lower than the LAC average.

A number of other factors, which are not fully captured by the standard gravity model specification of Equation (1), may further explain the divergent trade performance of the PRC and Latin America. Recent literature has emphasized the roles of institutions and infrastructure in determining trade performance. Levchenko (2007) suggested that the difference in institutional quality can be a source of comparative advantage, creating more trade flows. Using a gravity model, Anderson and Marcoullier (2002) found that bilateral trade volumes are positively affected by the trading countries' institutional quality. Helpman, Melitz, and Rubinstein (2008) found that regulation costs are important determinants of trade between two countries.

Table 3 [ PDF 16KB | 2 page ] summarizes the global ranking of major East Asian and LAC economies in terms of the ease of doing business as an indicator of the business environment. The ranking is intended to capture the extent to which a government creates a regulatory environment that is conducive to operating a business. The table shows that some middle-income economies in East Asia—such as Thailand and Malaysia—performed much better than any middleincome country in Latin America. The largest emerging economy in the world, the PRC, performed better than many middle-income LAC countries, including Brazil, which is the largest in Latin America. Even Viet Nam, a low-income East Asian country, performed better than many middle-income LAC countries. Although some East Asian economies—Lao People's Democratic Republic, Philippines, Cambodia, and Indonesia—are ranked low as in the case of several LAC countries, the table suggests that overall LAC economies lag behind East Asian economies in terms of business environment.

Because transport costs represent a significant barrier to international trade, infrastructure is also likely to have a significant effect on trade. By studying Sub-Saharan African trade, both internally and with the rest of the world, Limão and Venables (2001) found that infrastructure problems largely explain the relatively low levels of African trade. Nordås and Piermartini (2004) arrived at similar conclusions using a global data set. By incorporating institutional and infrastructure variables into a gravity model, Francois and Manchin (2007) found that institutional quality and infrastructure are significant determinants not only of export volumes, but also of the possibility of exports taking place at all.

To capture the effects of institutional and infrastructure factors on bilateral trade flows, we augmented the standard gravity model in Equation (1) with a set of institutional and infrastructure variables. The percentage of paved roads out of total roads and the number of mobile phones per 1,000 people are included as measures of infrastructure. We also included the number of legal procedures needed for an entrepreneur to legally start a business to reflect the regulation costs of firm entry.⁷ Cross-border trading costs are also included in the augmented gravity model. These trading costs are official costs incurred for exporting or importing a container of goods, excluding the costs of ocean transit and trade policy barriers. The data for both regulation costs and cross-border trading costs were obtained from the World Bank's Doing Business Index database. All these infrastructure and institutional variables are in natural logarithm in the gravity equation.

As the data for regulation costs and infrastructure were not available for 10 of the 161 countries examined in the standard gravity model, our sample for the augmented model was 151 countries. Column 2 of Table 2 shows the estimation results of the augmented gravity model. They suggest that the coefficients of all the gravity variables are largely unaffected, with the exception of per capita GDP. The estimate of the per capita GDP coefficient decreases from positive in the standard gravity model to negative, suggesting that this variable partly captured cross-country differences in infrastructure development and institutional quality in the standard gravity model.

With the exception of the percentage of paved roads out of total roads, all the coefficients of infrastructure, regulation cost, and cross-border trading cost variables are statistically significant and present the correct sign. Transportation and communication infrastructure in both exporting and importing countries have a positive effect on trade volumes. The effect of regulation and trading costs on trade seems even greater, with an estimated elasticity of 0.2–0.3.

Once the differences in infrastructure, regulation costs, and trading costs at the border are controlled for, the coefficients of the PRC as an exporter and as an importer (CHN_Exporter and CHN_Importer) are reduced by around half. Thus, the PRC's relatively high infrastructure quality, light regulation, and low trade costs were likely explanations for its superior trade performance in the standard gravity model. The institutional and infrastructure factors also partly accounted for the weak trade performance of Latin America, albeit to a lesser extent compared with the PRC. However, the coefficients on the bilateral trade dummy between the PRC and Latin America remain largely unaffected by the introduction of institutional and infrastructure variables, suggesting the role of other factors in the determination of their bilateral trade flows.

As discussed in the previous section, FDI has played an important role in the formation of the regional production network and, consequently, the rapid growth of trade in Asia. To assess the impact of FDI on trade in Asia and Latin America, we further augmented the gravity model with FDI variables. Following Kawai and Urata (1998), we used one period lagged inward aggregate FDI flows of exporting countries to remedy the possible endogeneity. To distinguish the varied roles of FDI in different regions, we interacted the dummies for four regional exporters (East Asia, Central America and Caribbean, South America, and the rest of the world) with the FDI variable, denoted in column 3 of Table 2 by EastAsia_FDI, CAC_FDI, SA_FDI, and ROW_FDI, respectively. The results point toward a strong and significant complementary relationship between FDI and exports in East Asia. For South America and the rest of the world, the interaction between FDI and exports is positive and significant, but of much smaller magnitude. The coefficient estimate for Central America is small and only marginally significant at the 10% level. These results confirm that the export-enhancing effects of inward FDI are pronounced in East Asia, present in South America, and virtually absent in Central America and the Caribbean.

With the interaction dummies of FDI inflows and exporting regions included, the coefficient estimate of the PRC dummy as an exporter changes from positive to negative, indicating that massive FDI inflows to the PRC may fully account for its exceptional export performance. On the contrary, the coefficient of the LAC dummy as an exporter changes from significantly negative to insignificantly positive, reflecting that the weak growth of FDI flows to LAC countries in the past decade may have been responsible for their weak export performance. The coefficient estimates of bilateral trade dummies are not much affected by the introduction of FDI variables. We postulate that some bilateral factors, such as strong complementarity in comparative advantage between the PRC and South America, which is not reflected in the gravity framework, may lead to their above average bilateral trade linkages.

Download this Paper [ PDF 128.5KB| 32 pages ].

Comment(s)

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