Roles Conceptualization, Funding acquisition, Investigation, Methodology, Writing – original draft * E-mail: sgil@uv.es Affiliation Department of Applied Economics II, University of Valencia, Valencia, Spain
Roles Data curation, Funding acquisition, Investigation Affiliation Department of Applied Economics II, University of Valencia, Valencia, Spain ⨯
Roles Conceptualization, Investigation Affiliation Department of Applied Economics II, University of Valencia, Valencia, Spain ⨯
The Doha Development Agenda recognizes the central role that international trade can play in the promotion of economic development. In fact, the increase of exports from developing countries to developed nations' markets has been considered a key element for developing countries to realize the potential benefits of globalization. Over the last decades, developed countries have provided preferential access to their markets to developing countries through nonreciprocal trade agreements. Moreover, developing countries have also participated in reciprocal trade agreements. This paper re-examines comparatively the effect of both kinds of trade agreements on exports from developing countries but also from the developed world. In line with other studies, our results across specifications are unstable. However, the results of our preferred specification give additional support to the argument raised by critics of nonreciprocal preference regimes who consider that developing countries should abandon their reliance on one-way trade preferences in favor of reciprocal agreements.
Citation: Gil-Pareja S, Llorca-Vivero R, Martínez-Serrano JA (2019) Reciprocal vs nonreciprocal trade agreements: Which have been best to promote exports? PLoS ONE 14(2): e0210446. https://doi.org/10.1371/journal.pone.0210446
Editor: María Carmen Díaz Roldán, Universidad de Castilla-La Mancha, SPAIN
Received: January 11, 2018; Accepted: December 21, 2018; Published: February 4, 2019
Copyright: © 2019 Gil-Pareja et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: All relevant data is available at http://www.uv.es/sgil/Base_Plos_ONE_2019.zip.
Funding: The authors acknowledge financial support provided by Ministerio de Economía y Competitividad (project ECO2015-68057-R, in part-financed by the European Regional Development Fund) and Generalitat Valenciana (PROMETEOII/2014-053) to SG-P, RL-V and JAM-S and Generalitat Valenciana (PROMETEO/2018/102) to SG-P and RL-V.
Competing interests: The authors have declared that no competing interests exist.
One of the most important challenges in the Doha’s round of trade negotiations is the promotion of development. The Doha Development Agenda recognizes the central role that international trade can play in the promotion of economic development. In fact, the rise of exports from developing countries to developed nations' markets has been for a long time viewed as a key element for developing countries to realize the potential benefits of globalization. The usual approach to accomplish this aim has implied that the developed world give support to the integration of developing countries into their economies through a series of unilateral concessional measures towards developing countries in the form of nonreciprocal trade preferences. In addition to all developed countries, since 2000s a number of emerging economies (Morocco, Turkey, India, China, Russia, Chile or Thailand, among others) also offer their own programs focused on Least Developed Countries (LDCs), although they are typically limited in scope. [1] provides the list of existing programs including their starting year, the current number of beneficiaries, their key features and the main online source of information for each program of preferential access.
The main instrument for the so-called “special and differential treatment” for developing countries has been the Generalized System of Preferences (GSP), but there have been many other nonreciprocal preferential trade agreements (NRPTAs). Examples of these others nonreciprocal programs are several EU and USA initiatives (the GSP+, ACP-EU partnership agreement, the Everything but Arms arrangement, the Caribbean Basin Initiative, the Andean Trade Preference Act and the African Growth and Opportunity Act), or the duty-free treatment granted by many other countries for least developed countries or for developing countries from a specific region. The intellectual underpinnings for the “special and differential treatment” arrangements in GATT for developing countries go back to the 1950s. They were based on balance of payments problems for developing countries, protection on infant-industry grounds and the Singer-Prebisch thesis about the secular decline in developing countries terms of trade ([2]).
As is well known, trade arrangements for developing countries have not been confined to one-way trade preferences. On the one hand, a large number of developing countries are members of the GATT/WTO system, which is based on the reciprocity and the most-favored nation principles. On the other hand, developing countries have also made a conscious effort to forge reciprocal preferential trade agreements, involving only developing countries (known as South-South agreements) or implicating both developing and developed countries (known as North-South agreements). The Common Market of Eastern and Southern Africa (COMESA), the Association of Southeast Asian Nations (ASEAN), and the Mercado Común del Sur (MERCOSUR) are examples of South-South agreements. In contrast, the North American Free Trade Agreement (NAFTA), the agreement between Canada and Chile and that between the European Union and Mexico are examples of North-South agreements.
Critics of nonreciprocal preference schemes have traditionally argued that developing countries should abandon their reliance on one-way trade preferences in favor of reciprocal agreements, since the latter implies a stronger, credible and lasting commitment (see, for example, [3]; [4] and [5]). This approach is also advocated by those who believe that the infant-industry argument, often used to justify unilateral concessions, is a fallacious argument.
The gravity equation has become the main econometric approach for examining ex post the “partial” (or direct) effects of economic integration agreements on aggregate bilateral trade flows. After accounting for multilateral resistance terms with time-varying fixed-effects and controlling for endogeneity bias using panel data techniques, [6] find that free trade agreements do increase countries’ bilateral trade flows significantly using data at five-year intervals from 1960 to 2000 for 96 countries. Following this empirical strategy and the same data set, [7] go a step further by comparing the impacts of North-South and South-South trade agreements on bilateral trade and show that free trade agreements lead to an increase in bilateral trade regardless of whether the signatories are developing or developed countries. In particular, they find that the percentage rise in bilateral trade is higher for South-South agreements that for North-South agreements.
Moreover, other recent studies focus on investigating the impact of nonreciprocal trade agreements on bilateral trade including also controls for regional trade agreements. [8] analyse the GSP with annual data for 184 countries over the period 1953–2006 and find that the GSP tends to foster exports in both directions in the short-run, but hampers them in the long-run. In contrast, [9] and [10], with data for 177 countries over the period 1960–2008, provide an in-depth analysis on the issue finding robust evidence that NRPTAs positively affect developing countries’ exports to developed countries but also find a positive effect in the opposite direction, that is, from benefactor countries to beneficiary countries. [11] examine the effects of various types of economic integration agreements (including one-way preferential trade agreements) on trade flows and trade margins with data at five-year intervals from 1965 to 2000, concluding that deeper integration agreements have larger impacts on aggregate trade flows, extensive margins and intensive margins than shallower agreements. [12] study the extent to which developing countries export more as a result of being in the official LDCs list and the effect of unilateral preference regimes. Using data over the period 1970–2013 they find that the inclusion in the LCDs list is associated with substantially higher exports. However, unilateral preference regimes are, on average, not always beneficial in terms of exports for beneficiary developing countries but do have some impact in some sectors. Finally, [13], with data between 1960 and 2015, find that whereas the average trade effect of the nonreciprocal preferences on the exports of beneficiaries is highly unstable across specifications the effect is strong and robust when these countries are members of the WTO and very poor. In contrast, non-LDCs beneficiaries only expand exports if they are not WTO members.
This paper re-examines comparatively the impact on exports from developing countries to developed countries of both reciprocal and nonreciprocal trade agreements using the latest available data and techniques of structural gravity estimation and taking into account the direction of export flows of the reciprocal agreements. The use of data up to 2016 allows us to analyze whether the nonreciprocal preferences have been eroded in recent years due to the numerous reciprocal trade agreements signed between developed and developing countries over the last decade. The sample period also incorporates the revision of the European GSP in 2014, when several middle income countries lost their preferential access to the European Union. Moreover, in line with [10] results, we also examine the potential differential impact of both types of agreements on exports from developed countries to beneficiary countries. From an econometric point of view, we control simultaneously for several sources of bias (multilateral resistance terms, unobserved bilateral heterogeneity, heteroskedastic residuals, and zero trade flows), which is in line with current “state of the art” in the literature on the gravity equation.
To preview our results, we find that the evidence reported varies across specifications. However, according to our preferred specification, PPML estimation with the dependent variable in export shares and allowing for correlation in the error term within all possible cluster dimensions, for the whole sample period only reciprocal agreements have had a positive effect in trade flows between developed and developing countries and only when the exporter is the developing country. For data up to 2008, there is evidence of a positive impact of the NRPTA on exports from beneficiary countries that vanishes in the last years. Finally, for exports from developed countries to beneficiary countries our preferred specification does not provide evidence of a significant effect in any case.
The paper is organised as follows. Section 2 presents the methodology. Section 3 describes the data. Section 4 discusses the estimation results. Finally, section 5 concludes the paper.
Since it was independently developed by [14]and [15] more than five decades ago, the gravity model has become the key econometric framework for estimating ex post the “partial” (or direct) effects of different kinds of economic integration agreements on bilateral trade flows. Our estimation strategy follows that of [6] and [16]. In particular, we control for multilateral resistance terms by including exporter-time and importer-time fixed effects. [17] emphasized that the gravity model theory implies that it is not just bilateral trade costs (the bilateral resistance to trade), but also the trade costs relative to the rest of the world (the multilateral resistance to trade) that are relevant for predicting bilateral trade flows. Moreover, we control for endogeneity by means of bilateral fixed effects. This issue has received a great deal of attention in the empirical gravity-equation literature since [6] noted that trade agreements are not exogenous. They showed that ex post estimation of the partial effects of free trade agreements (FTA) suffered from endogeneity bias, mainly due to self-selection of country-pairs into agreements (as a result of pre-existing trade levels), and find that this self-selection bias may be substantially reduced when employing pair-specific fixed effects or using first-differenced data.
As noted by [6] a few papers attempted to deal with the potential bias caused by endogenous PTAs in cross-section gravity equations with instrumental variables providing mixed and unstable results (see, for instance, [18] or [19]). Moreover, it is worth noting that finding appropriate instruments is not an easy task. The approach of [6] assumes that the main source of bilateral bias is time invariant and so they argue that panel regression techniques (estimation with country-pair fixed effects or first differencing) are more suitable to account for endogeneity and yield more stable results. An alternative way to deal with endogeneity is to use matching techniques. Obviously, country-pair fixed effects do not fully eliminate the concern about potential selection bias (endogeneity) since countries could adopt a PTA after a surge in trade within the sample period.
Since the late 2000s, empirical work on the determinants of trade flows has increasingly relied on a theoretically motivated gravity equation that controls for both multilateral resistance terms and unobserved bilateral heterogeneity with three sets of fixed effects: exporter-time, importer-time and country-pair (see, for example, [20], [21], [22], [23], [24],[25]-, [26], [7], [27] or [28]).
Our benchmark specification is the gravity Eq (1), which comprehensively accounts for multilateral resistance terms by including time-varying fixed effects and for self-selection endogeneity bias with country-pair fixed effects: (1) where i and j denote trading partners, t is time, and the variables are defined as follows: Xijt are the bilateral export flows from i to j in year t, CU, PTA, GATT/WTO are binary variables for common membership in currency unions, preferential trade agreements and General Agreement on Tariffs and Trade/World Trade Organization, NRPTAXbenMdevijt (NRPTAXdevMbenijt) is a binary variable which is unity if i is a beneficiary (benefactor) of a nonreciprocal preferential trade agreement and j is the corresponding preference-giving (beneficiary) country, ηij are country-pair fixed effects, χit and λjt are exporter-year and importer-year fixed effects, respectively, and uijt is the standard classical error term.
The benchmark specification includes only one dummy for all (reciprocal) preferential trade agreements (PTA). In order to investigate the issues addressed in this paper, we disaggregate the dummy variable PTA in three different ways by interacting the PTA dummy with dummies for whether the exporter and/or the importer countries are or not beneficiaries of nonreciprocal preferential trade agreements. Firstly, we split this dummy into two dummies depending on whether the exporter is a beneficiary country (PTAXben) or a developed country (PTAXdev). Secondly, we split the PTA dummy into two from the importer’s perspectives (denoted by PTAMben and PTAMdev). Finally, and most importantly, we split the PTA dummy into four dummies taking into account the group which each trading partner in the pair belong to: Exporter beneficiary and importer developed (PTAXbenMdev), exporter developed and importer beneficiary (PTAXdevMben), exporter and importer beneficiary (PTAXbenMben) and exporter and importer developed (PTAXdevMdev). For further clarification, let consider the first of these dummies as an example. PTAXbenMdev is a dummy that is unity if country i is a beneficiary country of a nonreciprocal trade agreement and country j is a developed country and they share membership in a reciprocal preferential trade agreement and zero otherwise. Comparing the estimated coefficient for this variable with that obtained for NRPTAXbenMdev allows us to test whether reciprocal or nonreciprocal trade agreements have been best to promote exports from developing countries. In a similar way, we can compare the export performance of developed countries participating in reciprocal (PTAXdevMben) and nonreciprocal (NRPTAXdevMben) trade agreements in trade with beneficiary countries.
Moreover, in line with other more recent works, we go further than [6] because we additionally account for heteroskedastic residuals and zero trade flows by estimating the gravity equation in levels rather than in logs with the Poisson Pseudo-Maximum Likelihood (PPML) estimator. This estimator was popularized by [29] to deal with a potential bias related to heteroscedasticity arising through log-linearization of gravity equations, but it is also useful by its ability to handle zeros in bilateral trade flows. After [29], a large number of recent papers deals with both econometric problems using PPML (see, for example, [30], [31], [32], [33], [34], [9], [28], [10]). However, the large datasets used in most of these studies or difficulties to achieve convergence have precluded to estimate the Poisson gravity equations including, at the same time, the three types of high dimensional fixed effects that are required to account for both unobserved bilateral heterogeneity and multilateral resistance terms (country-pair fixed effects, exporter-time and importer-time fixed effects). Fortunately, [16] have recently implemented an iterative PPML estimator that allows researchers to control simultaneously for both issues in a large dataset like the one used in this paper. Taking advantage of this technical development, we estimate the following gravity equation with the PPML estimator: (2)
Data on the dependent variable (bilateral export flows) come from the Direction of Trade dataset (IMF). The sample covers 182 countries and territories over 15 years of the period 1960–2016 at four-year intervals. Data for currency unions are taken from the CIA's World Factbook. The indicators of preferential trade agreements and GATT/WTO have been built using data from the World Trade Organization. In this study, we use the expression “preferential trade agreement” to refer also to other agreements involving a higher degree of economic integration. In fact, most economic integration agreements considered in the sample are free trade agreements. Data on the one-way trade preferences come from different sources. Data on the African Growth and Opportunity Act and Everything but Arms initiative come from the corresponding websites. The list of beneficiaries of the Caribbean Basin Initiative and the Andean Trade Preference Act come from the Office of the United States Trade Representative. The listing of beneficiaries of the Cotonou Agreement (ACP-EU Partnership Agreement) comes from its website and[35]. The list of countries that are beneficiaries of the standard GSP programs are taken from the United Nations Conference on Trade and Development (UNCTAD). For years before 2000, we use data from UNCTAD kindly provided by Bernard Herz and Marco Wagner. Finally, the information on developing countries that provide their own programs for LDCs (Morocco, Turkey, India, China, etc.) comes from the WTO database on Preferential Trade Arrangements and from the websites of the individual programs. Tables A and B in S1 Appendix show some descriptive statistics.
The list of developed countries includes those countries that have never been beneficiaries of one-way trade preferences Australia, Austria, Belgium-Luxembourg, Canada, Denmark, Finland, France, Germany, Greece, Iceland, Ireland, Israel, Italy, Japan, the Netherlands, New Zealand, Norway, Portugal, South Korea, Spain, Switzerland, Sweden, the United Kingdom and the United States, plus Cyprus, Czech Republic, Estonia, Hungary, Latvia, Lithuania, Malta, Poland, Slovak Republic and Slovenia (after their accession to the European Union in 2004), and Bulgaria and Romania (after their accession to the European Union in 2007).
The estimation results for the structural gravity Eq (1) by OLS appear in column 1 of Table 1. As it is usual in the literature, we report standard errors clustered by country-pairs (in parentheses). In this specification, all time-invariant standard regressors of the gravity equation (such as the bilateral distance between countries or the use of a common language) are absorbed into the pair-specific fixed effects. Moreover, all exporter-specific and importer-specific time-variant variables (such as the GDPs or the theoretical multilateral resistance terms) are controlled for the time-varying country-year fixed effects for both exporters and importers. Before discussing the results, note that this is a linear (in logs) regression and, therefore, do not permit the inclusion of zeros and, most importantly, it may provide inconsistent parameter estimates due to the likely presence of heteroscedasticity in trade data ([29] and [36]). As it is usual, the gravity equation works well explaining a large percentage of the total variation of bilateral exports flows. Moreover, with the exception of the point estimate of the dummy for currency union, all estimated coefficients for the economic integration agreements are positive and statistically significant at conventional levels. The GATT/WTO dummy shows the largest point estimate (0.543) and the difference with respect the other estimated coefficients is statistically significant at conventional levels.