Interdependent preferential trade agreement memberships: An empirical analysis

Abstract

The establishment of a new preferential trade agreement (PTA) or the expansion of an existing one alters the incentives of non-members to participate in a PTA. This can lead to a domino effect whereby non-members join an existing PTA. Or it can lead a pair of countries to establish a new PTA. We examine the determinants of why a pair of countries enters a bilateral PTA. Our emphasis is on (a) the impact of pre-existing PTAs and (b) whether this impact is larger when the members of pre-existing PTAs are on average geographically close to the pair of countries. Using data for 145 countries during 1955–2005, we find evidence that pre-existing PTAs increase the probability that a country-pair will enter a bilateral PTA and that this effect diminishes with distance. The analysis makes use of techniques drawn from spatial econometrics.

Introduction

If everything in the universe depends on everything in a fundamental way, it might be impossible to get close to a full solution by investigating parts of the problem in isolation. S. Hawking and L. Mlodinov (2005), A Briefer History of Time, Bantam Dell, New York, p. 15

The continued integration of the European Union (EU), the formation of the North American Free Trade Agreement (NAFTA), as well as the political discussion about a possible preferential trade agreement (PTA) between the Americas have been major sources for the renewed interest in PTAs in the last two decades. The increasing integration of the world economy brings about a rising concern about the global consequences of trade regionalism (see Krugman, 1991a, Bond and Syropoulos, 1996, Bagwell and Staiger, 1997, Bagwell and Staiger, 2005, Riezman, 1999, Bond et al., 2004; or Baldwin, 2006). With the worries about its consequences, economic research has described the spread of regionalism and focused on explaining its reasons. In particular, the latter line of research analyzed the welfare effects of PTA formation and countries' willingness to participate in a PTA (see Baldwin, 1995, Baldwin, 1997, Frankel et al., 1995, Frankel et al., 1998, Grossman and Helpman, 1995, Yi, 1996, Yi, 2000, Maggi and Rodríguez-Clare, 1998, Maggi and Rodríguez-Clare, in press, Baier and Bergstrand, 2004, Trefler, 2004).

A key feature of theoretical models of PTA formation is that countries or country-pairs do not decide about PTA membership in isolation. Rather, their decisions and behavior depend on other countries' actions. The formation of PTAs changes an outsider country's willingness to participate in a PTA. This is explicitly addressed in the domino theory of regionalism introduced and defined by Baldwin, 1995, Baldwin, 1997. As Baldwin (1997, p. 877) puts it, idiosyncratic incidents of regionalism trigger a multiplier effect that knocks down “… bilateral import barriers like a row of dominos.” The establishment of both NAFTA and the European Single Market created tremendous asymmetries among firms with versus to those without access to these huge markets. Market access is particularly important if firms are mobile across borders and multinationals control goods trade to a large extent. Then, market integration through PTA formation creates an incentive for multinational plant location within the PTA and stimulates a capital influx from abroad. In turn, the threat of capital flight into PTAs exerts pressure on outsiders to join existing PTAs. Baldwin, 1995, Baldwin, 1997, Baldwin, 2006 is a rich source of examples of PTA memberships that are consistent with the domino theory of regionalism. There, an initial shock associated with the formation of some PTA “… is amplified by the way in which enlargement makes nonmembership even more costly” (Baldwin, 1995, p. 45). If accession to an existing PTA is not feasible for political reasons, countries might prefer engaging in a new PTA with other outsiders for similar reasons. Game theoretical models even consider interdependence in the foundation of new PTAs through simultaneous rather than sequential membership decisions (see Yi, 1996, Yi, 2000).

This paper lays out an empirical analysis of PTA memberships. When new PTAs are founded or existing ones expanded, this paper asks about the impact of existing PTAs on the incentives of a pair of countries to entering a PTA.¹ In contrast to previous work, we allow PTAs in the past to affect a country-pair's empirical probability to join an existing PTA or found a new one. The former captures a domino effect of PTA membership as described by Baldwin, 1995, Baldwin, 1997. The latter reflects other forms of a spread of PTA membership. Ignoring such effects in empirical work may be harmful. For instance, one may mis-attribute the impact of PTA memberships in the past on subsequent PTA memberships to other factors.

We provide two pieces of evidence for the role of interdependence in PTA formation. One is based on panel data for 10,585 unique country-pairs in ten five-year intervals (or eleven years) between 1955 and 2005. With this data-set we explore the short-to-medium-run response in PTA membership probabilities to PTA formations in the past. The second piece of evidence relies on a cross-sectional data-set for the year 2005. In the latter, we cover a much larger set of 15,753 country-pairs for reasons of data availability. The cross-sectional approach explains PTA membership status rather than mainly the change thereof and, hence, provides inference about the importance of interdependence in the long run.

The empirical findings support the existence of interdependent decisions about PTA membership. Indeed, interdependence declines in distance and increases in ‘natural’ trade² among country-pairs. This is consistent with the view that countries participate in PTAs to avoid a welfare loss from trade diversion. The latter would be particularly strong if a country were an outsider to a PTA among its natural trading partners.

As hypothesized by Baldwin, 1995, Baldwin, 1997 in his domino theory of regionalism, the formation of PTAs in the past creates an incentive for outsiders to become a member of an existing PTA. This hypothesis is well supported by the data. We also find that PTA formation in the past triggers the formation of new PTAs subsequently. But the latter effect is relatively small as compared to the domino effect on membership in existing PTAs. Also, there is evidence that PTA memberships are interdependent in the very long run.

In two case studies focusing on the EU and NAFTA, we illustrate that the results are useful to ‘predict’ the process of regional integration both in Europe and North America. The contribution of the domino effect of regionalism to predicting PTA memberships is quite important for many country groups and periods. Ceteris paribus, our model typically would not have predicted actual PTA events without accounting for domino effects.

The remainder of the paper is organized as follows. In the next section we reconcile hypotheses regarding the interdependence of country-pairs in their decisions about PTA membership. Section 3 lays out the empirical model for interdependent observations with limited dependent variables in cross-sectional as well as panel data-sets. Section 4 summarizes the empirical results, and the last section concludes with a short summary of the most important findings.