Arnoldshain Seminar XV
“The EU and Latin America Facing Globalization”
September 4 – 6, 2017

Florensa, Luis Marcelo, Universidad Nacional de Córdoba, and María Luisa Recalde, Interdependence of Preferential Trade Agreements in Latin America.
Although the concept of PTA domino effect or interdependence exist since Baldwin (1995), the empirical literature that have quantified its importance is rather scarce. Egger and Larch (2008), hereafter EL, examine the determinants of bilateral PTA formation emphasizing on the impact of pre-existing PTAs. Also they consider whether this impact is larger when the members of pre-existing PTAs are geographically close to the pair of countries. Baldwin and Jaimovich (2012), hereafter BJ, proposed to test empirically the domino effect using a theory-based measure of contagion by computing an index that considers the bilateral exports related to total exports, capturing the spatial dependence in a different way. Finally, Baier, Bergstrand and Mariutto (2014), hereafter BBM, search deeply into the sources of interdependence by distinguishing two sources of PTA interdependence: an “own FTA” effect and a “cross FTA” effect. The former refers to the impact of a PTA between two countries owing to either already having other PTAs. The latter measures the impact of a PTA between the pair of countries owing to other PTAs existing in the rest of the world (third country-pair effect). The aim of our paper is to identify the main sources of interdependence that have been in place in Latin America during the last 25 years and that may have influenced the signing of a high number of PTAs in this period. That is, we study empirically whether the signing of a new preferential trade agreement or the expansion of existing ones creates incentives for other countries or pairs of countries in the region to join or form new PTAs. To achieve this goal we are going to apply two empirical strategies: on the one hand, the one used by EL and BJ and on the other side the one applied by BBM. In both strategies a logit is estimated, as the dependent variable reflects the existence or absence of an agreement between a given pair of countries. To predict both, PTA membership levels and changes, the two methodologies use panel data where the explanatory variables are lagged to avoid a bias associated with feedback effects. Besides, we also follow Chamberlain (1980) and Wooldridge (2002) to eliminate a possible correlation of time-variant regressors with the time-invariant component of the error term. Finally, both methodologies control for several economic, institutional and geographic determinants of PTAs formation as in Baier and Bergstrand (2004). However, these methodologies differ in the way of detecting the existence of interdependencies in the formation of trade agreements. In the present paper we apply these methodologies exclusively to Latin America to test three main hyphotesis associated with the different measures of interdependence: a) if the results are different when a region as Latin America is taken into account; b) if the results are robust to the different ways of measuring interdependence and c) if in LA the own effect is as important as at the aggregate level. Regarding the economic fundamentals, we expect countries of similar size will obtain greater welfare gains by forming a PTA; likewise, countries with different relative factor endowments will experience greater welfare gains relative to countries with similar factor endowments. Similarly, a pair of countries that are far from the rest of the world will have greater welfare gains in forming a PTA than those pairs of countries that have lower trade costs with the rest of the world (considering distance as a proxy for trade cost). Finally, PTA memberships will result in higher welfare gains if bilateral non-tariff trade costs for a pair of countries are low.