Essays in International Trade

Abstract

The gravity model proved to to be one of the most successful framework for analyzing international trade flows, being referred to as the “workhorse” in the international trade literature (Head and Mayer (2014)). Microfoundations to this model has been provided in Anderson (1979) and it has often been employed to estimate the effects of a variety of trade policies (see Cipollina and Salvatici (2010) for a meta-analysis on reciprocal trade agreements, Rose (2000) for the effects of currency unions). The two chapters of this dissertation, which are independent empirical pieces, both make use of gravity equations for the estimation of trade flows, although with different purposes. The first chapter focuses on the specification of the gravity equation. In the second chapter, instead, gravity equations are employed for assessing the relationship between trade and growth: in fact, their estimation represents the first step for the creation of an instrumental variable for export flows. In the first chapter, a solo-authored work titled Scale Economies in European Trade, I show that European data support the existence of economies of scale in trade flows. The impact of trade costs on trade flows, in fact, is assumed to be constant by almost all empirical studies employing the gravity framework. Anderson et al. (2016) are the first to depart from this assumption, allowing trade costs to vary as a function of trade volumes. Their model nests the more traditional one and hence can be used to test for the existence of these scale economies, which are shown to be in place for trade between US and Canada. For my analysis I construct a comprehensive dataset for European trade in manufacturing over a long time span (from 1980 to 2013), on which I employ the same methodology. My results show that scale economies in trade costs are indeed a strong empirical fact outside of the American continent, and this holds for all the 26 manufacturing sectors considered, with an estimated average of 0.64% decrease in trade costs given by a 10% increase in trade volume. The focus on Europe allows me to test whether the EU expansion affected these economies of scale. While this is not true on average, it seems to be the case for some industries: trade with a EU partner entails scale elasticities 50% lower than trade with a non-EU member for 11 sectors out of the 26 considered. I also investigate whether scale elasticities can be rationalized by the existence of informational asymmetries. Using detailed product-level data, I do not find evidence that the degree of product homogeneity can account for the observed cross-sectoral variation. The scale coefficients are instead linked to country-specific institutional variables, such as the level of corruption: exporting to the country whose level of corruption is the lowest in the sample entails half the scale elasticity than exporting to the most corrupted one. In other words, corruption depresses trade to an higher extent on longer distances. In the second chapter, joint with Carlo Altomonte and Italo Colantone and titled Trade and Growth in the Age of Global Value Chains, we revisit the relationship between trade and income, taking into account the recent surge of global value chains (GVCs). First, we develop a new geography-based, time-varying instrument for export, exploiting the sharp increase (almost tripling) in the maximum size of container ships between 1995 and 2007. This global shock has an asymmetric impact on bilateral trade flows across countries, affecting disproportionately more countries endowed with a larger number of deep-water ports, which are needed to accommodate the new, much larger ships. We exploit this heterogeneity for identification, building up the instrument for export in a gravity framework. Our result show that export has a positive effect on GDP per capita, with a 0.6 elasticity. Evidence at the country-level shows that this effect works through capital accumulation. Exploiting the decomposition methodology by Wang et al. (2013), we show that differences in the value added composition of exports matter for trade-growth nexus. We find evidence in favor of an income premium for countries that upgrade their positioning in GVCs, whereas the degree of participation to GVCs does not seem to play a role. Consistent with this finding, we show that countries whose average level of upstreamness (a’ la Antras and Chor (2013)) increases the most over time exhibit a higher trade elasticity of income. Both papers indirectly deal with the effect of geographical distance on international trade flows. One of the strongest regularities in economics is certainly the negative role played on trade flows by the distance between origin and destination. Disdier and Head (2004), comparing 1,467 different studies, compute an average distance elasticity of trade of about -0.9. Hummels (2007) shows that the distance elasticity of trade does not seem to diminish over time, as it would do should distance be capturing only transportation costs, thanks to the technological developments witnessed in the transportation sector. Distance seems then to refer to trade costs in general, including institutional, policy and regulatory barriers that, also for historical reasons, often increase the further away countries are located. In the first paper, I show that the impact of distance on trade flows is not constant but varies with trade volumes. This corresponds with having a component of the composite friction described before, hidden in the distance term, being fixed and is consistent with micro-evidence on the export behavior obtained from firm-level data (Roberts and Tybout (1997)). It seems natural, then, to test whether some characteristics, either at the product-level or at the country-level, have a prominent role in explaining the non-linear effect that distance has on trade. My results find in level of corruption of the destination country an important determinant. In the second paper, we test whether the distance elasticity of trade varies as a function of the number of deep water ports on both the importer’s and the exporter’s shores, capturing the extent to which countries can trade via container vessels. The data support this claim for all the manufacturing sectors considered, showing that geographical distance, even though non-exclusively, captures the incidence of transportation costs on export flows.