2.1 Introduction
In this chapter, we provide a general overview of Latin American (LA) countries’ experience with and exposure to preferential trade agreements (PTAs), with a specific emphasis on empirically situating the current landscape. This is an important stock-taking exercise as it helps contextualize the more focused contributions contained in the subsequent chapters of this volume. Our primary goal in this chapter, therefore, is not to lay out a precise argument, which we then test empirically, but rather to uncover important patterns that can help serve as a starting point for additional theoretical and empirical work.
We turn our focus first to understanding the macro-level drivers underlying countries’ motivations to negotiate and use PTAs in the region. We argue that currently there are three distinct clusters of countries with different underlying motivations behind their engagement with PTAs, which we label the liberal globalists, reluctant globalists and anti-globalists. The liberal globalists use PTAs as part of a broader commitment to a relatively open, liberal economic strategy, which entails signing deep PTAs with a variety of partners both within and outside the region that contain significant commitments to market liberalism. The reluctant globalists, in contrast, tend to pursue shallower PTAs aimed at furthering more statist developmental goals and are more oriented towards the region, whereas the anti-globalists either opt out of PTAs altogether or sign agreements that are more explicitly political in an effort to counter or prevent outside influence in the region.
After this theoretical stock-taking exercise, we turn to empirical examinations at the regional, PTA and country levels. First, we explore whether major differences exist when we compare LA to other world regions. Second, we seek to understand differences within the region in terms of how many PTAs countries sign as well as their design. We find that several economic and political factors help explain the content and ambition of PTAs. The results suggest that richer, more democratic countries not only sign more PTAs but that they also sign deeper PTAs, which typically include issues that go beyond just trade. Third, we employ text-as-data methods to explore whether there are models from which LA countries regularly adopt language when designing their PTAs. Our empirical analyses show some evidence that agreements involving the US have diffused within the region, but we fail to uncover strong evidence of a single template or templates that LA countries disproportionally emulate.
Overall, our results have important implications for understanding the design and diffusion of PTAs in LA. They paint a picture of a region where political and economic differences from country to country produce profound differences with respect to PTA signing patterns and PTA design. We conclude by pointing out how our mapping exercise helps contextualize the results in subsequent chapters of the book, as well as suggesting some promising avenues for future research.
2.2 Latin American PTAs Today
Currently, the underlying motivations for the use of PTAs in Latin America vary greatly from country to country. Nevertheless, in order to understand broad patterns of PTA signing in the region, we argue that it is useful to classify countries into three groups, which we label liberal globalists, reluctant globalists and anti-globalists.Footnote 1 Each group varies with respect to the underlying rationale – in terms of both economic and foreign policy goals – for why the country in question is motivated to sign PTAs. We acknowledge at the outset that these categories are broad, and hence likely conceal fine-grained differences between countries, but we nonetheless wager that these classifications can help us understand more nuanced differences in terms of both partner choice and treaty design.
Our first group we dub the liberal globalists, which is a group of countries that, by and large, use PTAs as a way to engage in relatively open trade with countries both within and outside Latin America.Footnote 2 These countries, which include Chile, Peru, Colombia, Mexico and most of the Central American countries, have signed PTAs that liberalize many sectors of the economy and typically include a variety of non-trade or trade-related provisions that aim to reform domestic economic and political systems (Sanahuja Reference Sanahuja, Alfonso, Peña and Vazquez2009; Quiliconi Reference Quiliconi2014; Carranza Reference Carranza, Briceño-Ruiz and Morales2017; Laursen Reference Laursen2018; Bohigues and Rivas Reference Bohigues and Rivas2019). To varying degrees, this reflects a domestic political consensus in many of these countries that such agreements advance important interests in society, including generating economic growth and enhancing gains from trade, as long as they are designed to exclude and protect some important domestic economic actors, and are supported by strategic sectors and transnational firms (Hicks et al. Reference Hicks, Milner and Tingley2014; Bohigues and Rivas Reference Bohigues and Rivas2019).Footnote 3 Moreover, several of these countries are thought to have used trade agreements as a vehicle to ‘lock-in’ domestic reforms that would be difficult to make in the absence of international legal commitments (see Fernandez and Portes Reference Fernandez and Portes1998; Baccini and Urpelainen Reference Baccini and Urpelainen2014). Many of these countries have signed PTAs with the US and EU, have participated in relatively open regional efforts such as the (aborted) Free Trade Area of the Americas (FTAA) and the Trans-Pacific Partnership/Comprehensive and Progressive Agreement for Trans-Pacific Partnership (TPP/CPTPP), and have signed bilateral and regional trade agreements with other liberal globalists in and outside the region.
In contrast, another group of countries, which we call the reluctant globalists, are still committed to engaging in international trade but have generally been more sceptical of a fully open or liberal model of doing so. Thus, these states have attempted to foster a more developmental or protection-based form of regionalism and have viewed PTAs with extra-regional partners as potentially costly in both economic and sovereignty terms. Argentina, Brazil, Paraguay and Uruguay fit this mould.Footnote 4 These countries still engage in considerable international trade but do so to promote statist industrial development goals more in line with ‘new South American regionalism’, which evolved after the switch away from import-substitution industrialization (ISI) strategies (Briceño-Ruiz Reference Briceño-Ruiz2013; Vieira Reference Vieira2014; Bohigues and Rivas Reference Bohigues and Rivas2019). These countries have sought to channel their cooperation through agreements such as the Southern Common Market (MERCOSUR) in an effort to consolidate regional integration along with developmental and social goals. Domestically, these countries have economies that contain significant, and politically influential, export-oriented sectors, but they often benefit from state industrial policy (Brooks and Kurtz Reference Brooks and Kurtz2012).Footnote 5 These countries also, on average, promote a larger role for the state in the economy than the liberal globalists.Footnote 6 At the same time, however, it should be noted that these countries have often vacillated between elements of an open and closed economy depending on political developments (Briceño-Ruiz Reference Briceño-Ruiz2018). Regardless, the primary difference between this group and the liberal globalists is that, in general, the reluctant globalists are more sceptical of the benefits of across-the-board openness and seek to sign economic agreements that promote state developmental goals even if some of those come at the potential expense of economic growth and trade expansion.
Finally, there are the anti-globalists, which have generally sought to limit their exposure to trade and investment, especially if it is outside the region. Moreover, to the extent that these countries have engaged in regional and bilateral PTAs, they have tended to be oriented towards the pursuit of social goals such as group rights, redistribution and social inclusion, while also pushing back against agreements and policies that are perceived as neo-colonial or imperialist (Bohigues and Rivas Reference Bohigues and Rivas2019, p. 6). This group includes Venezuela, Cuba, Bolivia, Nicaragua and several Caribbean countries. When countries in the anti-globalist group sign PTAs, they are often less oriented towards trade and economic integration, instead pursuing political goals such as the need to counter perceived US or foreign dominance in the region. One of the most notable attempts here is the creation in 2002 by Venezuela and Cuba of the Bolivarian Alliance for the Peoples of Our America or ALBA agreement, which was conceived of as an alternative to the US-proposed FTAA (Serbin and Serbin Pont Reference Serbin and Serbin Pont2017). Since its creation, ALBA has added ten additional members – including Bolivia, Nicaragua, Dominica, Antigua and Barbuda, St. Vincent and the Grenadines, Saint Lucia, Grenada, and Saint Kitts and Nevis – but has seen the withdrawal of Honduras and Ecuador.
These groupings suggest several important observable implications that we address in the following sections. First, it suggests that we should see considerable variation at the country level in the region. In other words, we should expect that countries in the different groupings will sign notably different types of economic agreements. Thus, we should expect to see some degree of fragmentation across the PTA universe in the region. Moreover, and as we discuss further in the conclusion, changes in the domestic political dynamics within each country also affect the stability of country preferences over time. At this point it is unclear, however, which state-level factors are likely to systematically explain this variation. One possibility is that economic factors such as reliance on trade or market structures at the country level predict variation in PTA design. Another possibility is that political factors, such as type of government, left–right orientation or geopolitical interests, affect PTA design. Second, it suggests that we are likely to see competing models of agreements in the region. As noted above, we expect broadly that the liberal globalists will prefer deeper agreements and will be more willing to work across regional divides, whereas the members of the other groups are more likely to sign shallower agreements with regional partners who share similar interests.
To address these issues, we proceed in three steps. First, we compare LA to other regions to identify broad differences at the regional level. Second, we look at variation across several economic and political dimensions at the country and PTA levels. Third, we use quantitative text analysis to look for patterns of text diffusion in the region. Our aim here is to better understand whether there are clear PTA templates from which LA countries routinely draw.
2.3 Comparing Latin America to Other World Regions
As a first exercise, we compare Latin America to six other world regions. We rely on the World Bank’s definition of region, which, for Latin America, includes forty-two countries spanning Central America, the Caribbean and South America.Footnote 7 We compare regions not only on their overall relationship to PTAs but also along relevant political and economic dimensions. Although this method of comparison is crude, masking considerable intra-regional variation, we view it as important to help establish if there are any unique regional patterns.
Table 2.1 compares Latin America to six other world regions with respect to the average number of PTA partners, PTA design and across several political and economic dimensions.Footnote 8 For our primary design variable, we include depth from the Design of Trade Agreements (DESTA) database, which measures the degree to which a PTA requires the state to undertake significant economic liberalization across areas including trade in goods, services, investment, standards, public procurement, competition and intellectual property rights (IPRs) (Dür et al. Reference Dür, Baccini and Elsig2014). The variable is constructed as an additive index which varies from 0 (shallow) to 7 (deep). We also include the degree of democracy for each country as measured by the commonly used Polity2 score which ranges from −10 to +10 (Marshall and Gurr Reference Marshall and Gurr2020). Gross domestic product (GDP) per capita, as well as exports, mining, merchandise exports, and natural resources rents as a percentage of GDP are all from the World Bank’s World Development Indicators.Footnote 9
Table 2.1 Comparing PTAs and political and economic characteristics of Latin America to other world regions.
Note: Data on PTAs taken from the DESTA database; Polity2 scores from Polity5 data; all additional variables from the World Bank’s World Development Indicators. All indicators are averages computed over the years corresponding to the signature of PTAs.
Several interesting trends stand out. First, column 2 shows that LA countries have the second-highest number of average PTAs signed of all regions at nearly twenty-three. At first glance, this might appear surprising as international trade is less important as a percentage of the economy in LA than in several other regions (column 6), including East Asia and the Pacific, Europe and Central Asia, and the Middle East and North Africa (MENA). For example, exports comprise about 30 per cent of the average LA country economy compared to 61 per cent for East Asia. What also stands out is the relatively high importance of mining for many economies (column 7), providing a potential additional incentive for negotiating PTAs from the perspective of the export market.
One potential driver for the relatively high number of PTAs relative to other regions is that LA is, on average, much more democratic than other regions with similar economic profiles. Numerous studies have argued for a positive effect of democracy on PTA signing (Mansfield et al. Reference Mansfield, Milner and Rosendorff2000; Milner and Kubtoa Reference Milner and Kubota2005; Manger and Pickup 2016). Mansfield and Milner (Reference Mansfield and Milner2012) argue that democratic governments tend to sign more PTAs on average than other countries as this can help leaders communicate to domestic audiences that the government is a good steward of the economy. Column 4 shows that the average Polity2 score for the LA region is 2 points higher than East Asia and the Pacific and South Asia, and approximately 8 points and 10 points higher than sub-Saharan Africa and MENA, respectively.
Figure 2.1 visualizes the aggregate relationship between democracy and the total number of PTAs across the seven regions. This is an admittedly broad comparison, as there is significant intra-regional variation, which we explore below, but it is nonetheless illustrative. The graph shows that the general trend is for regions with more democracies to have more PTAs on average. Aside from Europe and Central Asia, which is an extreme outlier due to its high values on both variables, the LA region stands out. It is the region having signed on average the second-highest number of PTAs while scoring third in terms of level of democracy.Footnote 10 Thus, the fact that the average LA country has more PTA partners than sub-Saharan Africa, MENA, East Asia and the Pacific and South Asia might be related to higher levels of democracy. These data alone, however, do not tell us the full story of the relationship between PTAs and democracy, which may potentially be endogenous.
A second interesting trend illustrated by the regional comparisons concerns the design of PTAs, which we assess with the depth measure from DESTA. The data indicate that LA countries are clearly in the middle of the pack, with neither overly deep nor shallow PTAs. The average depth of LA PTAs is 2.54, which is well below regions such as Europe and Central Asia, North America, and East Asia and the Pacific, which have average depth values of 3.99, 4.91, 3.8, respectively, but higher than MENA, sub-Saharan Africa and South Asia with averages of 2.09, 1.93 and 1.26, respectively. One potential explanation for this middle position might be due to the fact that the region is comprised primarily of middle-income countries. Figure 2.2 shows that there is a strong relationship between GDP per capita across the different regions and the average depth of PTAs. This relationship is explored at the country level later, but here it is useful because it shows that even though LA countries sign many PTAs, due to their relatively more democratic governments, on average, their PTAs tend to be less ambitious.
This also corroborates our country classification, as this middle position is also a function of the fact that there is considerable heterogeneity in the type of agreements pursued, with the liberal globalists preferring more open agreements relative to the reluctant globalists and anti-globalist groups. For example, the liberal globalists began signing more open agreements in the 1990s and continue to sign deep agreements today with a variety of countries inside and outside the region. Whereas the other groups have tended to sign shallower agreements, often with regional partners. Thus, the fact that the average depth of the region is lower than Europe, North America and East Asia but higher than regions such as Africa and South Asia makes sense, given the more varied agreements.Footnote 11
2.4 Within LA Region Variation
In this section, we examine variation in PTA design at the country and PTA levels. Our goal is both empirical and conceptual. Empirically, we hope to illustrate important differences between countries in the region with respect to the political and economic determinants of PTA design. Conceptually, we hope to better understand how these associations are related to the broad country groupings we laid out in the opening section. We also hope to improve our understanding of how country-specific factors lead to different treaty outcomes.
In Table 2.2, we summarize key descriptive statistics on the number of PTAs and their design, which we average at the country level. As above, we are particularly interested in the depth of PTAs, but we also include flexibility, also from DESTA, which measures the degree to which countries can opt out of core provisions, and DSM strength, which is the strength of dispute settlement provisions in each PTA (Allee and Elsig Reference Allee and Elsig2016).Footnote 12 The data show that seven countries have signed more than thirty total PTAs, including Chile, Mexico, Argentina, Peru, Brazil, Uruguay and Colombia. Again, this suggests that many countries in the region, aside from the anti-globalists, have made PTAs central to their foreign economic policy. It is, however, striking to observe the considerable variation in design. Chilian and Peruvian PTAs have an average depth of over 4, which is much higher than the average depth of Argentinian, Brazilian and Uruguayan PTAs, which average between 1 and 1.5. Mexico and Colombia stand in the middle with an average depth of 3. This pattern is relatively similar when looking at other design aspects such as flexibility and DSM strength.
| Country | # PTAs | Depth | Flexibility | DS strength |
|---|---|---|---|---|
| Chile | 46 | 4.21 | 3.1 | 5.97 |
| Mexico | 43 | 3.02 | 2.63 | 3.57 |
| Argentina | 40 | 1.31 | 2.1 | 2.24 |
| Peru | 37 | 4.09 | 2.94 | 5.48 |
| Brazil | 35 | 1.45 | 2.15 | 2.75 |
| Uruguay | 34 | 1.48 | 2.11 | 2.9 |
| Colombia | 33 | 3.00 | 2.83 | 4.03 |
| Venezuela | 27 | 0.88 | 2.14 | 1.61 |
| Cuba | 27 | 1.11 | 2.33 | 1.5 |
| Guatemala | 26 | 2.96 | 2.64 | 5.00 |
| Costa Rica | 25 | 4.18 | 3.05 | 5.95 |
| El Salvador | 25 | 2.91 | 2.55 | 5.18 |
| Panama | 25 | 3.27 | 2.59 | 5.14 |
| Paraguay | 24 | 1.71 | 2.33 | 3.74 |
| Honduras | 23 | 3.52 | 2.9 | 5.62 |
| Guyana | 21 | 1.84 | 2.47 | 2.53 |
| Nicaragua | 21 | 3.05 | 2.65 | 4.6 |
| Ecuador | 20 | 1.29 | 2.57 | 2.73 |
| Trinidad & Tobago | 20 | 2.00 | 2.63 | 3.05 |
| Grenada | 18 | 2.18 | 2.41 | 3.12 |
| Antigua & Barbuda | 17 | 2.38 | 2.62 | 2.81 |
| Barbados | 17 | 2.19 | 2.56 | 3.00 |
| Dominica | 17 | 2.31 | 2.44 | 3.00 |
| St. Lucia | 17 | 2.31 | 2.44 | 3.00 |
| Jamaica | 16 | 2.27 | 2.47 | 3.2 |
| St. Kitts & Nevis | 16 | 2.47 | 2.53 | 3.00 |
| St. Vincent & Grenadines | 16 | 2.47 | 2.53 | 3.00 |
| Bahamas | 14 | 2.38 | 2.38 | 3.31 |
| Belize | 14 | 2.69 | 2.62 | 3.69 |
| Bolivia | 13 | 2.33 | 2.67 | 2.89 |
| Suriname | 12 | 2.64 | 1.91 | 3.45 |
| Dominican Republic | 10 | 3.62 | 2.25 | 4.75 |
| Haiti | 6 | 2.40 | 2.20 | 3.20 |
We know that deeper agreements tend to include on average more flexibility provisions (Baccini et al. Reference Baccini, Dür and Elsig2015). Chile, Costa Rica and Peru demonstrate this tendency, as these countries sign PTAs characterized by a high average depth and flexibility. In contrast, Argentina, Brazil, Uruguay and Paraguay show an inverse pattern on average, signing shallower PTAs but with few flexibility provisions. This makes sense, given that shallower agreements have fewer commitments and thus less need for flexibility. This contrasts with countries such as Bolivia, and to some extent Ecuador, which have signed fewer agreements but have tended to sign shallower agreements with more flexible provisions.
Taken together, this illustrates that there are different logics that underlay the rationale for PTA design. The liberal and reluctant globalist groups both negotiate and sign many PTAs, but the design of their agreements is quite different. The liberal globalists sign newer generation PTAs that include extensive liberalization, cover a variety of new trade issues and include dispute settlement and flexibility provisions. In contrast, the reluctant globalists sign shallower agreements that lack the same level of ambition and, as a consequence, do not need the same degree of flexibility. For example, the average depth of PTAs signed by the liberal globalists is 3.46 compared to an average of 1.96 for the reluctant globalists.
Other examples of differences in design include Panama and Nicaragua who have signed on average much deeper and more comprehensive agreements relative to Paraguay and Ecuador, even though they have signed a similar number of agreements (20–25). Costa Rica and Venezuela also follow a similar pattern. Dominican Republic is another interesting case. Although the country has signed a relatively small number of agreements (10) compared to other LA countries, they seem to be very comprehensive with high scores across PTA design metrics.
We now focus on the degree to which PTAs in the region include provisions beyond trade in goods. Figure 2.3 summarizes the percentage of each country’s PTAs that include provisions on trade-related issues (e.g., investment, services, IPR) and non-trade issues (NTIs) (e.g., environment and labour). These issues are among the most important issues that indicate the overall ambition of agreements. Three groups of countries emerge. The first group is composed of countries in the top one-third of the chart, from Chile to El Salvador, that tend to regularly include trade-related and NTIs in their agreements. The second group, roughly the middle third of the figure from Suriname through Bolivia, tend to include investment and services but do not regularly include IPR, environment and labour provisions. Finally, the last third, from Uruguay down, tend to only sparingly include provisions on investment, and nearly never include provisions on the other NTIs.
What this exercise demonstrates is that the liberal globalists, including Chile, Peru, Colombia and many of the Central American countries, tend to include a variety of trade-related and NTIs in their PTAs. This reinforces the patterns above, which demonstrate that this group also signs deeper agreements on average. In contrast, the reluctant globalists and anti-globalists often include investment and services but rarely include newer generation issues such as IPR, the environment and labour.
These broad comparisons demonstrate that the liberal globalists and reluctant globalists both sign many trade agreements on average but that there are major differences in design, with the liberal globalists signing deeper, more ambitious agreements on average. Moreover, they include more clauses related to the environment as well as investment, IPR and labour.Footnote 13 In contrast, the reluctant globalists are also active signers of PTAs but sign agreements that are both shallower and narrower in scope. Finally, the anti-globalists sign fewer agreements on average, and those that are signed tend to be even shallower and narrower.
2.5 Empirical Analysis – What Factors Determine PTA Design?
The evidence suggests that our typology is helpful as a heuristic. Therefore, in this section, we engage in a more systematic empirical analysis of the determinants of PTA design in the region. We focus on three sets of explanations for PTA design. First, we look at the economic determinants of PTA design, including level of development and the export orientation of the parties. Second, we look at the political determinants of design, including regime type and government ideology. Finally, we examine whether design is impacted by partner choice.Footnote 14 We focus both on intra-LA PTAs (i.e., a PTA signed by LA countries only) and extra-LA PTA (i.e., a PTA signed by at least one LA country but that also includes members from a country or countries outside the region).
The three main dependent variables used in the analyses are PTA depth, flexibility and dispute settlement strength from the DESTA database.Footnote 15 We use the World Bank’s World Development Indicators and the UN’s COMTRADE data to measure the economic characteristics of the PTA members. We measure level of development with Per Capita GDP, which is members’ average GDP per capita. Exports measure the members’ average exports.Footnote 16 Ceteris paribus, we expect that countries with a higher GDP per capita as well as more reliance on exports should favour deeper agreements. We also expect this to go hand in hand with more flexible agreements, as these provide assurance against shocks as well as provide safeguards for groups who stand to lose from liberalization (Rosendorff and Milner Reference Rosendorff and Milner2001; Baccini et al., Reference Baccini, Dür and Elsig2015). Finally, we do not have strong expectations about whether economic factors are directly related to the strength of dispute settlement as other research indicates that dispute settlement design is related most strongly to other design features, such as depth (Allee and Elsig Reference Allee and Elsig2016).
For political determinants, we use data from both the Polity Project and Global Leader Ideology (Herre Reference Herre2023). Polity is the members’ average Polity 2 score and has been rescaled to a 0–20 scale for regression purposes. Ideology leader captures the leaders’ ideology of countries, classifying them as ‘leftist’, ‘centrist’ or ‘rightist’. We recoded this variable as 1 (leftist), 2 (centrist) and 3 (rightist) and computed the average ideology at the PTA level across countries at the time of signing the agreement. Although this measurement is very crude, it still enables us to evaluate whether leader ideology matters for PTA design. Overall, we expect that more democratic countries will sign PTAs that are deeper as democratization has been associated with economic reform in LA (Murillo Reference Murillo2001; Weyland Reference Weyland2002) as well as globally (Milner and Mukherjee Reference Milner and Mukherjee2009; Baccini Reference Baccini2012). At the same time, we expect that democratic polities will demand more flexibility as they face on average domestically more veto players (Allee and Elsig Reference Allee and Elsig2017), while they tend to be more open to legalized dispute settlement, as democracies support the rule of law and implementation of commitments. We also expect that as the membership of a PTA becomes more rightist in orientation that PTA design should become deeper as rightist governments are more insulated from social groups and therefore able to make more widespread economic reforms, such as those entailed by deeper PTAs (O’Donnell Reference O’Donnell1978; Nelson Reference Nelson1990). Similarly, we expect that left-wing governments are more likely to include flexibility provisions due to the influence of social groups who fear losses; however, this is a weak expectation due to the fact that depth and flexibility are often empirically related (Baccini et al., Reference Baccini, Dür and Elsig2015). We do not have clear expectations for the influence of leader ideology on the strength of dispute settlement.
The remaining three independent variables capture other aspects of PTA membership: EU is a dummy variable to identify whether the EU is part of the agreement; we include a similar dummy variable for the US. In both of these cases, we expect that the presence of the US or EU as a PTA partner should lead to deeper, more flexible and more legalized agreements.Footnote 17 Intra − LA is a dummy variable, where 1 identifies an agreement composed by only LA countries and 0 an agreement signed by at least one LA country but also includes members from outside the region. Initially, we are agnostic about the sign of this variable, as it is not a priori clear whether intra-regional agreements will be driven more by the preferences of the liberal globalists or one of the other groupings. PTAs signed among members from other regions are not included in this analysis.
Lastly, PTAs including more flexibility clauses also tend to be deeper, as adding such clauses can ease the negotiations of deeper provisions (Baccini et al. Reference Baccini, Dür and Elsig2015). Similarly, stronger DSM are more likely to be included in deeper agreements (Allee and Elsig Reference Allee and Elsig2016). We therefore also include the PTA depth in our flexibility and DSM strength regressions as these design provisions are often substantially related.
As our dependent variables are all ordered indices, we use ordered logistic regression to estimate the relationship between political, economic and membership patterns on PTA design.Footnote 18 Our sample includes all PTAs signed by at least one LA country and includes a total of 231 PTAs over the period from 1951 to 2021.Footnote 19
2.5.1 Discussion of Results
The results of the regression in Table 2.3 highlight several important dynamics. First, as expected, PTA depth is a strong predictor of the level of flexibility and the strength of dispute settlement provisions in PTAs. Second, the average members’ GDP per capita has a positive and statistically significant effect (p < .01) on depth but, surprisingly, a negative and significant effect on DSM. This contradictory effect could be explained by the fact that countries with higher level of development are generally more economically powerful and might sign deep agreements to boost their economy but see less need to rely on a legalized dispute settlement system. Third, the average members’ export level has a positive and significant effect on depth but neither on flexibility nor on DSM strength. These results are in line with the findings of Allee and Elsig (Reference Allee and Elsig2016) on the determinants of dispute settlement provisions in PTAs. In particular, the authors find that trade volumes do not seem to have an important impact.
(1) Depth | (2) Flexibility | (3) DSM | |
|---|---|---|---|
| Depth index | 0.64*** (0.09) | 0.78*** (0.10) | |
| Log mean GDPPC | 1.05*** (0.28) | −0.53 (0.32) | −1.03*** (0.34) |
| Log mean exports | 0.23* (0.09) | 0.01 (0.09) | 0.13 (0.11) |
Mean polity score (0–20) | 0.14** (0.04) | 0.01 (0.04) | 0.22*** (0.06) |
| Mean ideology leader | 0.27 (0.23) | 0.07 (0.25) | −0.33 (0.28) |
| Intra-LA | −1.08** (0.37) | −1.20*** (0.45) | −1.92*** (0.49) |
| EU | −0.05 (0.71) | −0.49 (0.76) | 0.64 (0.77) |
| US | 0.26 (0.80) | −1.72** (0.77) | 3.28*** (1.20) |
| Observations | 181 | 168 | 171 |
| McFadden R-squared | 0.24 | 0.32 | 0.46 |
Note: As we estimate an ordered logistic regression, there are several intercepts for all the different intervals in the dependent variable. The full set of results are available by request from the authors. Significance levels are 0.01 (***), 0.05(**), 0.1(*).
Turning to our political determinants, we find a positive and significant impact of the members’ average polity score on two of the three design variables. The coefficients for depth and DSM are positive and significant at the 99 per cent level. The coefficient for flexibility is positive but not significant. This demonstrates that PTAs with more democratic members are deeper and contain stronger DSM provisions on average.Footnote 20 By contrast, we find no significant relationship between leader ideology and the main PTA design characteristics. There are several potential explanations for this. First, contained within the ‘right’, ‘left’ and ‘centre’ categories, there are potentially very different types of regimes (which would also have an effect as seen previously). Second, the variable we use only captures the ideology of the leader but does not account for the ideology of other parts of the government (parliament, ministers, etc) which may be different. Lastly, it is also possible that trade policy is less sensitive to changes in government than to overall regime type. Moreover, as has been shown recently, in many countries there is institutional inertia in the bureaucracy which can be reinforced by interest groups domestically. For example, Calvert and Tienhaara (2022) show that many bureaucrats and interest groups in Peru continued to support investment provisions in PTAs, even after Peru experienced costly arbitration, due to embedded ideas about the positives of liberal, market-based policies.
Lastly, we also uncover interesting patterns when it comes to PTA membership. An important finding is the negative and significant impact of intra-LA on all our dependent variables. Most importantly, PTAs signed among LA countries tend to be on average shallower than those signed by LA countries and partners outside the region. One interpretation is that this is the result of distributional concerns: LA countries are willing to engage in deeper liberalization with developed partners outside the region where there are likely to be significant gains from trade due to complementary factor endowments (capital-rich and labour-rich, respectively) but sign shallower agreements with countries with similar factor endowments. This result tracks with theoretical thrust of Rodrik’s argument (Reference Rodrik1992) as well as the empirical pattern shown in Baccini (Reference Baccini2012). Moreover, this indicates that the design of intra-LA PTAs is driven by reluctant globalists.
Surprisingly, the EU and US variables are not systematically significant across our models, except for a negative impact of US membership on flexibility and a positive impact of US membership on DSM strength. These results may seem surprising as the EU and the US are known for being influential trade partners with strong preferences for deeper, more legalized agreements. Although the EU and US have signed deep agreements with countries from LA, they represent a relatively small share in comparison to the overall number of deep PTAs signed by at least one LA country.Footnote 21 The significant negative impact of US membership on flexibility is driven by the low representation of the US among the most flexible agreements.Footnote 22 By contrast, the US is a member of 40 per cent of the agreements, including the most far-reaching DSM provisions.Footnote 23 To test for the robustness of this result, we re-estimated the model adding the PTA overall depth as an independent variable, and we still find a strong correlation.Footnote 24 This seems to signal the US’ willingness to include strong DSM in trade agreements.
Overall, the regression results reveal several important patterns. First, we see that the economic characteristics of countries help explain PTA design outcomes. There is strong evidence, for example, that richer countries and those who have more export-oriented economies have a preference for deeper PTAs. Second, we also see that political factors matter. Our analyses reveal that PTAs with more democratic memberships favour agreements that are deeper and have more legalized dispute settlement provisions. Third, we show that there appears to be stark differences with respect to geographic scope, and therefore partner selection logic of LA PTAs. Intra-regional PTAs tend to be much shallower on average. Surprisingly, however, we did not find that these design differences are overwhelmingly influenced by economically powerful partners from outside LA. The results did not show, for example, that PTAs signed with the US or EU were deeper or more flexible than agreements signed with other outside the region partners. We did, however, find that the agreements with the US tended to have more legalized dispute settlement. Taken together we interpret this as evidence that the liberal globalists sign deeper agreements with a wide range of countries outside of LA, whereas agreements within the region tend to be driven by preferences of reluctant globalists or anti-globalist states (or at least reconciled with their interests).
2.6 Text Analysis of LA PTAs
In this section, we look at the text of negotiated PTAs to determine whether there are distinct models or templates around which LA PTAs converge. There exist several different theoretical perspectives for why states may choose to reuse language in PTAs. Allee and Elsig (Reference Allee and Elsig2019) show that many countries directly copy PTA text from earlier agreements. This phenomenon is driven by several dynamics. First, states with low bureaucratic capacity may rely on copy-pasting as a way to bypass their lower capacity. Second, powerful states may strategically reuse language in order to push their preferences (see Allee and Lugg Reference Allee and Lugg2016), which they can deploy via strategic sequencing (Castle Reference Castle2023). Moreover, states may choose to adopt language from different agreements, including from other forums such as the World Trade Organization (WTO) or bilateral investment treaties (BITs) in order to incorporate their preferred language (see Allee et al., Reference Allee and Elsig2017; Chaisse et al., Reference Chaisse, Elsig, Jusoh and Lugg2022). Third, Peacock et al. (Reference Peacock, Milewicz and Snidal2019) posit that copy-pasting creates consistency that can benefit firms, consumers and the economy by promoting innovation, reducing costs and increasing competition. Fourth, recent research suggests that countries may reuse language in order to enhance the legitimacy of cooperation (Clark and Pratt Reference Clark and Pratt2024). In contrast, Kim (Reference Kim2015) finds that commonalities in text among PTAs is low on average, which suggests the existence of different templates. Given the current debate (and potential mechanisms), we conduct text analyses on PTAs signed by LA countries to discern what determines levels of similarity, if they do exist. The main question we are interested in is whether there are distinct clusters or models around which countries’ PTA practices converge.
To determine similarity between PTAs, we use automated text analysis to examine the text of 185 trade agreements signed by at least one LA country.Footnote 25 To facilitate analysis, we employed basic pre-processing on the text by removing numbers and transforming all text to lower case, while still retaining word order. To identify identical text, we require that the text between two agreements matches word-for-word (with no deviations) in sequences of five words or more. We then compute Jaccard scores, which tell us the total number of words shared between two documents (Alschner and Skougarevskiy Reference Alschner and Skougarevskiy2016).Footnote 26 The resulting numbers range from 0 to 1, with higher numbers indicating greater similarity between two PTAs.
Text analysis of 185 PTAs in our corpus of texts does not indicate that there are distinct clusters or models that dominate in the region. At the same time, however, there is evidence that some of the countries do copy-paste on occasion. Below we present four heatmaps to show country networks. Darker cells indicate areas of increased overlap. For illustrative purposes we chose Chile, Mexico, Brazil and Venezuela as these countries represent the different groupings of liberal globalists, reluctant globalists and anti-globalists. What the results suggest is that similarity is not uniformly distributed throughout each country’s network. In other words, we do not observe countries adopting a single unique model which they then use repeatedly or expand upon. Instead, they seem to opportunistically reuse language from one treaty to the next, which is likely influenced by the preferences and power of the other country or countries involved (Figure 2.4).
This variation poses a unique possibility: perhaps there are specific agreements that are copy-pasted regularly, even if there is no evidence that countries are engaging in widespread copy-pasting in their own networks. Is there evidence that LA countries draw much of their language from their treaties with outside countries, or if they do so primarily within the region? For example, do we observe an overwhelming influence of US or EU-led PTAs, or do LA countries draw from within the region, or, perhaps, from their own networks? To investigate this, we took four distinctive PTAs of the different models we are interested in, namely MERCOSUR for an intra-regional template, Chile-EC for the EU template, and both Central American Free Trade Agreement (CAFTA) and Peru-US for a US template. We then compared the similarity scores between these agreements and all other agreements signed by at least one LA country.
These PTA templates were chosen for a number of reasons. First, there has been a considerable time span since they were signed (MERCOSUR in 1999, EC Chile in 2002, CAFTA in 2004 and Peru-US in 2006) which increases the possibility that they might be used as model templates for later agreements. We then looked at the characteristics of the most important PTAs signed to date and found that Chile-EC, CAFTA and Peru-US all have comparable features, such as depth and average polity scores of member countries.Footnote 27 MERCOSUR was chosen as the Latin America PTA template since it was created with the intention of regional integration and the establishment of a Southern Common Market (Manzetti Reference Manzetti1993). Therefore, despite the characteristics of MERCOSUR being different to the other PTA templates, it still stands out as the most relevant treaty template for an intra-regional model (Table 2.4).
The results suggest the US PTAs have had an impact in the region, albeit not as extensive as one might expect. The highest similarity percentage across all four PTAs is between Peru-US and Colombia-US, with an overlap of 75 per cent. Moreover, there is some evidence that language included in both CAFTA and Peru-US has diffused to PTAs where the US is not a partner, such as Nicaragua-Taiwan, Australia-Chile and Australia-Peru. That said, there is not extensive evidence that these models have been adopted broadly in the region. Similarly, there is some evidence suggesting that European agreements, both EC and EFTA, have been copied, especially when the EC is a common partner. However, these percentages are typically around or below 20 per cent of the total language re-used, indicating, as in the case of US agreements, that there are limits to their diffusion. In contrast, we find little evidence that text from MERCOSUR has widely diffused in the region.
Overall, there does not appear to be a model or models that LA countries copy text from exclusively, either within or outside the region. There is some evidence that countries draw text from US and European PTAs, but the degree to which this is occurring is not as high as one might initially suspect. Our intuition is that this is due to the fact that countries within the liberal globalist group sign many deep agreements but, given the nature of their treaty networks, tend to draw language from a variety of sources or craft new provision on an ad hoc basis. Thus, the most likely story is that LA countries – principally the liberal globalists – tend to draw from their own past treaty practice but that the actual content is likely to be an amalgam of several of their different past treaties which can be with a diverse slate of countries geographically and politically. We see this in the treaty networks of countries such as Chile, which have signed relatively liberalizing agreements with partners across the Pacific Rim and elsewhere, including the US, Singapore, Australia, Indonesia, Vietnam, Malaysia, Japan, the UK and Turkey.
2.7 Conclusions
The analyses in this chapter paint a picture of a heterogeneous region with respect to exposure to, experience with and preferences towards PTAs. One central conclusion, which is borne out strongly by the analyses, is that governments in the region sign different types of agreements, with different partners, depending on their underlying preferences with respect to economic development. The economically liberal states in the region, principally Chile, Peru, Colombia, Mexico, Costa Rica and Panama, have tended to seek out partners both within and outside Latin America with whom they can negotiate relatively deep PTAs. Importantly, these deep PTAs contain provisions that go beyond trade liberalization, to include investment and non-tariff barriers as well as a variety of other NTIs, including environmental standards, and labour provisions.
In contrast, the reluctant globalists – including Brazil, Argentina, Paraguay and Uruguay – sign shallower agreements which skew towards other, like-minded countries in the region. There are notable exceptions to this pattern, of course, but in general these countries’ scepticism of liberalization and the political obstacles towards the dismantling of a strong developmental state, mean that their agreements tend to be shallower and narrower in design.
Finally, the anti-globalist states – including Venezuela, Cuba and Bolivia – have a distinct pattern of PTA signing altogether. They have, on average, signed fewer agreements than the other groups and signed much shallower agreements on average, which often have underlying geopolitical motivations such as the ALBA initiative. As such, these countries have not been exposed to market forces or international trade and investment flows to the same degree as the countries in the other groups. Instead, PTAs, to the extent that they are pursued at all, often reflect the more idiosyncratic priorities of individual regimes.
Interestingly, however, these divergent underlying preferences towards PTAs have not led to a coalescing of treaty practice whereby countries anchor their PTA design around one of several distinct PTA templates. For the liberal globalists, major US-influenced initiatives in the region such as CAFTA and Peru-US have clearly had some impact on design choices. Moreover, we see that several of the more ambitious countries in this group have tended to sign deep agreements by seeking out like-minded states both within and outside the region. But, that said, there was no clear single template around which these states engaged with, indicating that their own idiosyncratic preferences are embedded in their treaty practice and that the final provisions are influenced in part by negotiating dynamics in any given case.
Ultimately, it is useful to highlight here several important larger dynamics which help contextualize the chapters to come and should help serve as a guide for future research. First, many LA countries clearly have domestic political incentives to sign PTAs, and most countries have heeded this call. The aggregate data suggest that this may be a function of LA’s high level of democracy relative to other regions, as current research suggests that leaders in democratic countries are incentivized to sign PTAs (Mansfield and Milner Reference Mansfield and Milner2012). Moreover, this is borne out by looking within the region; the countries with weaker democratic elements have tended to sign fewer agreements on average, and their design often reflects the idiosyncratic preferences of their leaders or important domestic actors. There are important differences within the democratic societies of the region, however. Rather than signing similar PTAs, countries in the region have chosen to sign qualitatively different PTAs, likely in response to divergent domestic political factors unique to each country.
This core dynamic is taken up in important ways in several of the other chapters in this volume. Dür and Huber (Chapter 3), for example, show that preferences for PTA partners are driven by relevant domestic political cleavages, including citizen and legislator political ideology. Moreover, the chapter by Calvert (Chapter 11) shows how preferences for reform of investment provisions in PTAs hinge on the influence of various domestic political actors. Similarly, how popular PTAs are, and hence the ability of governments to sign future agreements, also depends greatly on domestic politics, as highlighted by Campello and Urdínez (Chapter 13) and López and Bórquez (Chapter 8). Future studies should continue to probe this important research vein to better understand how democratic politics within LA – and variation in the nature of democratic institutions and societal preferences across LA countries – affects the supply and demand for PTAs. Importantly, these factors, by impacting PTA partner selection and design, are likely to condition the economic impact of a given PTA.
Second, we see that the legacy of the development state and ongoing domestic political debates about how to reconcile a developmental state with the exigencies of an increasingly globalized world clearly influence treaty design, which may have important downstream effects. The less statist, more liberal states in the region have tended to be the most ambitious in signing deep PTAs. Whereas countries with a more statist legacy have tended to sign shallower agreements that favour protecting important domestic economic and political interests. This may show up under the auspices of special and differential treatment, as shown by Zelicovich et al. (Chapter 5), but also likely impacts partner selection and design through multiple channels. How this has impacted countries’ political economy is unclear, however. For example, we see that per capita wealth and treaty practice are correlated in the region – with richer countries signing more and deeper PTAs on average – but it is unclear whether PTAs are a consequence of greater levels of development or, potentially, a cause. Future studies should aim to unravel this relationship in more detail. Moreover, the developmental state and backlash against globalization in places like Brazil and Mexico can also lead to ‘murky protectionism’, as highlighted by Albertoni and Wise (Chapter 12). This is something that we also observe in formerly liberal globalist states where public opinion is shifting as shown in the case of Chile by Lopez and Borquez (Chapter 8).
Third, we see LA is also a fertile ground for experimentation in the design of PTAs, which may also explain the high degree of heterogeneity in terms of design outcomes (see, e.g., the chapter on gender provisions by Bahri (Chapter 4), as well as on digital provisions by Klotz and Ugarte (Chapter 6)). The analyses in this chapter do not yield determinative predictions with respect to when policy experimentation will occur, but they do strongly suggest that the dynamics of democratic politics in the region as well as the fact that LA countries are more than willing to sign agreements with a variety of countries both within and outside the region likely plays a role. This experimentation may lead to inconsistency in countries’ PTA networks, which can complicate the rules required for domestic economic actors and provide openings for disguised protectionism, but it also indicates a certain level or responsivity of PTA design to citizen concerns. Moreover, as is the case with the gendered economic impact of service provisions in Chilean PTAs, as shown by Munoz and Caceres (Chapter 9), experimentation may be particularly important as an avenue for countries to help foster sustainable development in line with local preferences. Importantly, as is shown in the chapter by Álvarez Zarate et al. (Chapter 7), while bound in many cases by legal obligations at the WTO and in PTAs, countries often still retain policy space to seek domestic goals. Finally, the contribution by Saco (Chapter 10) shows how public policies ‘travel’ across Latin America and how some PTAs can act as catalysts of policy diffusion, a role the past research has not sufficiently accounted for.
Open access