Investment Treaty Arbitration in Latin America

This study analyzes variations in the incidence of state involvement in investment treaty arbitration in Latin America and the Caribbean over the 1987–2014 period. Its main contributions are fourfold. First, by focusing on the balance of incentives and opportunities facing political leaders and foreign investors, the study establishes a new basis for understanding the reasons actors opt for the risks and uncertainties of international arbitration instead of resolving international investment disputes via alternative means. Second, by restricting the focus of research to the Latin American region, the study is able to move beyond the analysis of relatively time-invariant structural and institutional conditions and crude indicators to address the role played by the preferences of political actors. Third, by disaggregating disputes by sector of investment, the analysis documents the variable role of ideology and electoral incentives in investment treaty arbitration. Finally, by underlining the distinctiveness of the Latin American experience with investment treaty arbitration, the study offers new insights into the global backlash against investment treaty arbitration, which has gone farther in the region than anywhere else. Overall, the results underline the potential payoffs of breaking down the international experience with investment treaty arbitration into regional patterns to obtain a deeper and more granular understanding of the political economy of investor-state dispute settlement.

This study analyzes variations in the incidence of state involvement in investment treaty arbitration in Latin America and the Caribbean over the 1987-2014 period. Its main contributions are fourfold. First, by focusing on the balance of incentives and opportunities facing political leaders and foreign investors, the study establishes a new basis for understanding the reasons actors opt for the risks and uncertainties of international arbitration instead of resolving international investment disputes via alternative means. Second, by restricting the focus of research to the Latin American region, the study is able to move beyond the analysis of relatively time-invariant structural and institutional conditions and crude indicators to address the role played by the preferences of political actors. Third, by disaggregating disputes by sector of investment, the analysis documents the variable role of ideology and electoral incentives in investment treaty arbitration. Finally, by underlining the distinctiveness of the Latin American experience with investment treaty arbitration, the study offers new insights into the global backlash against investment treaty arbitration, which has gone farther in the region than anywhere else. Overall, the results underline the potential payoffs of breaking down the international experience with investment treaty arbitration into regional patterns to obtain a deeper and more granular understanding of the political economy of investor-state dispute settlement.
Over the past three decades, the dramatic expansion of overseas investment coupled with the proliferation of international investment agreements has fueled a major surge in treaty-based investment disputes. Under the provisions incorporated in most investment treaties, foreign investors gain the right to pursue claims of unfair or unequal treatment against states at privately contracted arbitral tribunals rather than being forced to rely on host-country courts for property rights protection. The benefits of investment treaties for states are more ambiguous. Although there is some mixed evidence that investment treaties promote foreign investment (Egger and Pfaffermayr 2004;Salacuse and Sullivan 2005;Kerner 2009;Vis-Dunbar and Nikiema 2009;Yackee 2011;Colen, Persyn, and Guariso 2016), their dispute settlement provisions also create new sets of risks for states, including costly settlement proceedings, unpredictable arbitral decisions, potentially crippling financial awards, and constraints on national sovereignty.
Unlike the rules governing world trade, which provide for nation-states to take disputes to the World Trade Organization, investment treaty arbitration can be initiated only by investors. Moreover, states cannot legally decline to participate in arbitration. The idea that states and investors jointly decide whether to initiate arbitral proceedings is therefore somewhat misleading, as is the presumption that treaty arbitration typically results from host states pursuing illegal policies. Investors turn to investment arbitration to challenge government measures that threaten to undermine their expected profits, often with a view toward preempting policy changes and/or discouraging other states from adopting similar policies (Weiler 2010). 1 Although the total number of investor-state disputes brewing at any time is unknown, only a small subset result in investment treaty arbitration. Disputes often simmer for years and come to light only after investors file a formal notice of arbitration. At that point, state actors face a choice of negotiating a settlement with investors or incurring the costs and uncertainties of treaty arbitration.
Under which conditions are the interests of state actors and foreign investors most likely to collide in ways that give rise to investment treaty arbitration? This article attempts to shed new light on this question by analyzing the factors that have shaped the relative probability of state involvement in investment treaty disputes in the Latin American and Caribbean (LAC) region. Although prior research has addressed the origins of investment disputes on the basis of global data, several issues warrant more systematic analysis.
First, to date spatial variance in investment treaty arbitration has received limited attention. Researchers have assumed that the risk of dispute involvement is relatively uniform across the less industrialized world and responsive to a common set of causal conditions. This study calls this assumption into question by placing the LAC experience with dispute settlement into comparative perspective. Second, prior efforts to explain the outbreak of investor-state disputes have relied heavily on structural and institutional variables that leave us very far from the realm of actors and agency. By disaggregating the global experience with investment treaty arbitration to focus more narrowly on the Latin American region, this article is able to marshal both qualitative and quantitative data on the political preferences of leaders and mass publics to enhance the understanding of variations in dispute participation. Third, the analysis establishes a new basis for explaining the global backlash against investment treaty arbitration, which is rooted in its systematic bias in favor of upper-income states, asymmetric treatment of states and investors, as well as acute tensions between the international investment regime and democratic accountability.
The first section of this article outlines the LAC experience with investment treaty arbitration in relationship to the broader global context, emphasizing its distinctiveness with respect to frequency of disputes, relative risk of adverse outcomes, and high level of public support for policies that potentially bring governments into conflict with foreign investors. The second section focuses on the political economy of variation within Latin America. Placing the balance of incentives and opportunities facing political leaders and foreign investors at the center of analysis, it develops a theoretical argument about the factors that shape dispute incidence and outlines a set of hypotheses to guide the empirical inquiry. The third section presents time-series negative-binomial panel regression results to explore variations in the incidence of investment treaty arbitration across Latin America over the 1987-2014 period. The final section summarizes the major findings of the analysis and their implications for the ongoing debate over investor-state dispute settlement.

Overview of Investor-State Dispute Settlement in Latin America
Although investor-state dispute proceedings may be based on investment contracts or host country law, most claims arise under international investment agreements (IIAs), which create broad and enforceable rights for foreign investors and typically empower them with the right to initiate claims against states at international arbitration tribunals. Since the early 1990s, a complex array of these agreements have come 1 A case in point is the recent dispute between Colombia and the Swiss pharmaceutical giant, Novartis. Responding to domestic political pressures, the Colombian Minister of Health began pressuring the company to negotiate a major price cut in a cancer chemotherapy drug. The company immediately issued a formal notice of arbitration-twenty-two months before the government decided to impose price controls on the drug (Williams 2016; "Emiten resolución que declara al medicamento Glivec de interés público," 2016).
to link the countries of the LAC region, including Cuba, with one another as well as with the rest of the world. NAFTA, CAFTA, CARICOM, and other regional agreements form part of this web, along with other free-trade agreements and more than four hundred bilateral investment treaties (BITs). Because of this growing pattern of legal integration, LAC nations, particularly those that attract a high volume of foreign investment, have been exposed to increased risk of dispute involvement. Whereas only 13 regional arbitral claims were registered before 2000, by January 2015 that number had soared to 194. The most important outlier is Brazil, which has yet to ratify a single investment treaty, even though failure to do so has exposed Brazil's own investments abroad to the risk of unfair treatment or hostile takeover (Campello and Lemos 2015). Although all world regions have been exposed to a rising risk of investor-state dispute settlement (ISDS), Latin America stands out on three counts. First, along with Eastern Europe, the region has been disproportionately targeted by arbitral claims. As of mid-2015, 31.0 percent of the universe of 626 known treaty-based investment disputes involved countries in the region (UN Conference on Trade and Development [UNCTAD] 2015; International Center for the Settlement of Investment Disputes [ICSID] 2015). This percentage far exceeds the LAC's 13 percent share of global foreign-direct-investment flows over the 1980-2014 period (UNCTAD 2016a), or its relative level of participation in bilateral or other international investment treaties. Out of a universe of 2,562 international investment agreements in force as of early 2015, only 490, or 19.1 percent, involved LAC nations (UNCTAD 2016b). In comparison, East Asia, with a lower rate of participation in international investment treaties (13.2 percent) but a larger share of global foreign direct investment (20.4 percent), accounts for only 1.1 percent of known treaty-based investor-state disputes (UNCTAD 2015a(UNCTAD , 2016a(UNCTAD , 2016bICSID 2016).
Second, the outcomes of treaty arbitration in Latin America have been significantly less favorable to states than they have been in other regions. As Table 1 shows, investors have won a higher percentage of awards than states in the region, whereas states have prevailed heavily over investors in other world regions. 2 These differences, which are statistically significant at the .01 level, underline the distinctiveness of the LAC experience and the importance of disaggregating data on global outcomes to achieve a more complete understanding of investment treaty arbitration.
Third, a study of awards issued and made publically available by ICSID, the predominant venue for international investor-state dispute resolution, also shows that the LAC region stands out in terms of the magnitude of the financial damages awarded to investors by arbitral tribunals. As of June 30, 2013, 60 percent of the thirty largest awards for damages involved states in the region, including the single largest 2 For an analysis that explores the variance in dispute outcomes, see Remmer (2018).  (Hodgson 2014;Sinclair, Fisher, and Macrory 2009) and costing an estimated $8 million (Gaukrodger and Gordon 2012) to $10.6 million (Commission 2016) in legal fees and expenses plus tribunal costs, such prolonged proceedings have added to the costs of investment treaty arbitration for states in the region. The high profile of LAC in ISDS reflects the nature of foreign investments in the region, which have been channeled heavily into the extraction of natural resources, particularly oil, gas, and mining, as well as public utilities and other infrastructural investments, in which the financial stakes of conflict are typically high. In Latin America, these are precisely the economic sectors characterized by a long tradition of state involvement, if not outright public ownership. Although this pattern was partially reversed in the 1990s via privatizations, the interests of investors and states continue to collide over public regulation in the former areas of state economic involvement, giving rise to the majority of the region's international investment disputes. Of the 145 treaty-based arbitral claims filed against LAC nations between 1987 and 2014, 29.0 percent have involved oil, gas, and mining, as compared to 17.2 percent for other countries. Infrastructural investments, including electricity and power, transportation, water, and sanitation, account for another 32.5 percent of the regional dispute total, as compared to 23.7 percent for rest of the world (UNCTAD 2015a; ICSID 2016). Not surprisingly, regions such as East Asia, in which foreign investment has been directed mainly into manufacturing production for global value chains, have experienced far more harmonious relations between foreign investors and state actors.
Equally distinctive are political conditions in the region, which has been governed almost entirely by democracies since the end of the 1980s. Whereas most prior research has linked investor risk with authoritarian rule (Jensen 2008;Li 2009) or discounted the relevance of regime type entirely (Franck and Wylie, 2015), the Latin American experience suggests an entirely different possibility: namely, that the attitudes of mass publics in the region create incentives for elected officials to adopt policies that are opposed by foreign investors and resist backing down, even when threatened with investment treaty arbitration. This proclivity appears particularly marked in countries characterized by high levels of popular protest against foreign investors. Bolivia is a case in point. In 1999-2000 violent protest against the privatization of water and sewage services in Cochabamba led to a head-on collision between the Bolivian government and a consortium of foreign investors, which filed a $50 million investment treaty claim against Bolivia for rescinding its privatization contract (Schultz 2000). Similarly, a $301 million arbitral claim against El Salvador-a sum equivalent to more than 40 percent of the country's total budget, was filed in the wake of the 2009 election after the newly elected leftist government failed to approve a permit for a gold mine to avoid potential ground water contamination ("El Salvador" 2015, 15;"El Salvador Government Spending 1990. Mining was a significant issue in the 2009 electoral contest, not least because several antimining activists were killed in the run-up to the election. Tensions over energy and mining companies have likewise fueled protests across Peru, many of them violent and highly disruptive ("Peru's Humala Faces Threat of Spiralling Social Conflicts" 2015; Arce 2014).
The possibility of achieving electoral gains by challenging private investors is documented by public opinion data. In Argentina, a poll taken in 2012 showed that between 62 percent and 74 percent of the population approved of the nationalization of the Spanish energy giant Repsol's controlling interest in YPF S.A., the country's leading energy producer ("Más del 70% de los argentinos, a favor del a expropiación de YPF" 2012). Similarly, a 2013 survey revealed that 64 percent of the Mexican public opposed foreign investment in the state-owned oil company ("La encuesta sobre la privatización de Pemex" 2013). Bolivian opposition to foreign investment is even higher. In the referendum of 2004, 84.6 percent of the electorate voted to repeal the law opening the energy sector to foreign investment (Political Database of the Americas 2016), establishing the basis for President Morales's takeover of foreign investments in the hydrocarbons sector. The popularity of El Salvador's stand against foreign gold mining is likewise consistent with the results of a 2007 survey by the University of Central America in San Salvador, in which more than two-thirds of respondents agreed that the mining of precious metals should be banned in the country (Achtenberg 2011). Even in Chile, where market-oriented policies have prevailed since the era of the Pinochet dictatorship, public opposition to foreign investors is strong. According to a 2013 survey, 67 percent of the electorate supports the nationalization of foreign owned copper mines ("CERC: 67% de chilenos es partidario de nacionalizar grandes mineras privadas" 2013).
Strong political sensitivities to foreign direct investment (FDI), combined with the prevalence of democracy, comparatively high levels of dispute involvement, and weak success rates in dispute settlement, have all contributed to the vigor of the regional reaction against ISDS. Interviews with government officials across the less industrialized world indicate that investment treaties were signed with hopes that they might foster greater FDI; however, they were mainly viewed as diplomatic gestures involving a "token of goodwill," with few significant implications for host state sovereignty (Poulsen 2015;Poulsen and Aisbett 2013, 280-283). Poulsen and Aisbett (2013, 285-296) have also shown that after being targeted with an investment treaty claim, a state's enthusiasm for further BIT participation declines significantly. That reaction has been particularly strong in Latin America. After arbitration losses, a number of states have openly challenged the legitimacy of arbitral awards, including Argentina, which has adopted an aggressive legal strategy involving appeals against adverse awards before courts at the seat of arbitration as well as multiple rounds of annulment, supplementation, and rectification proceedings at ICSID (Goodman 2007). The country has also drawn on the "Rosatti doctrine" to insist on national court review of arbitral awards on the grounds that BITs do not supersede the national constitution (Alfaro 2004;Alfaro and Lorenti 2005). As of mid-2015, the country had yet to comply fully with a single adverse international arbitral ruling ("Argentina Settles" 2013).
Ecuador, Bolivia, and Venezuela have gone even farther (Gaillard 2008) by denouncing the convention establishing the ICSID and formally withdrawing from the institution. The three countries also stand out with respect to the variety of legal steps they have taken to insulate themselves from adverse awards, including the termination of BIT agreements and the adoption of constitutional amendments designed to protect them against the threat of arbitral claims (Gómez 2011;Gillman 2008). All three states are enthusiastic proponents of the regionalization of dispute settlement under the auspices of the recently established Union of South American Nations (UNASUR), which will operate under a new set of rules designed to allow for the appeal of arbitral decisions, limit challenges to state sovereignty, and establish a system of precedent to ensure greater consistency and predictability in arbitral awards (Leathley 2011).
The strength of backlash against ISDS, however, varies considerably even within UNASUR. Some of its members have responded to the challenges of ISDS by creating specialized agencies capable of identifying possible risks and coordinating effective responses to investor challenges. Peru, for example, has been singled out for its "good practice in the prevention and management of investor-State disputes" (UNCTAD 2011, 1). Colombia, likewise, has been described as a "model" for its development of an arbitration prevention and mitigation program (Gomezperalta and Dansby 2014). Despite its extended web of treaty commitments and success in attracting FDI, Colombia did not face a single investment dispute until 2016.
Although these variations have important implications for investment flows and relations with the advanced capitalist world, they have yet to be systematically explored in the literature on Latin American politics, foreign investment, or related topics. In an attempt to shed new light on this intraregional diversity as well as to enhance the broader understanding of the political economy of investment treaty arbitration, the analysis turns to the study of shifting patterns of investment treaty arbitration across the LAC region since 1987, when the first treaty-based investor-state dispute was registered at an international tribunal.

Explaining Variations in Dispute Involvement across the LAC Region
What accounts for the widely varied experience of countries in the LAC region with the emergent international investment regime? Why have some countries become involved in a long series of investorstate disputes while others have avoided investment treaty arbitration? Which factors have shaped the relative willingness of regional governments to risk formal international arbitration proceedings rather than resolve conflicts via negotiation? What strategic considerations guide foreign investors and state leaders?
Unfortunately, prior research offers limited insight into these questions. Most of it is limited to the presentation of descriptive data on issues such as investor nationality, the development status of respondent states, sector of investment, type of legal claim, and firm size (e.g., Franck 2007Franck , 2009Gallagher and Shrestha 2011;Behn 2015;Schultz and Dupont 2014). The main exceptions are Freeman (2013) and Kim (2017), who have conducted statistical analyses explaining the incidence of investment disputes in terms of the relative strength of democracy and/or host country legal institutions. The contributions of both studies, however, are limited in several important respects. First, both draw causal inferences from a set of cases that are not representative of the broader universe of known disputes. Freeman's research covers 277 arbitrations registered in the 1987-2007 period, but the overwhelming majority of disputes, which currently total more than 800, were registered after his study was completed. This is an issue because major changes have occurred in investment treaty arbitration over time (Poulsen and Aisbett 2013;Remmer 2018). Kim's sample covers a broader period but includes only 84 out of a total of 139 lower-and middle-income countries and-as a result of missing data records-results for only 1,255 country-year observations out of a potential universe of 4,170. His analysis also begins in 1981, before the first investor-state dispute, meaning that the risk of arbitral involvement was zero for 40 percent of his observations. Second, the key explanatory variables utilized by both Freeman and Kim, such as the rule of law, are both highly autocorrelated over time and potentially endogenous to investment disputes. Third, the methodology utilized in these studies is open to question because it does not fully address the panel structure of the available data. Kim's interpretation of his statistical findings is also misleading inasmuch as he claims to uncover an inverted U-shaped relationship between democracy and disputes, but his coefficients for polity and polity squared are both negative and statistically significant; that is, his data actually show increased democracy reduces disputes across the full range of the Polity spectrum. Finally, and most important, the theoretical arguments presented in these prior studies closely track research on expropriation (e.g., Henisz 2004;Jensen 2008;Li 2009) with its traditional emphasis on the importance of institutional constraints on state actors. As Pelc (2017) has persuasively argued, however, the literature on political risk and expropriation offers limited insight into the strategic calculus of parties participating in investment treaty arbitration.
Taking this latter critique as its central point of theoretical departure, this analysis begins by assuming that investment disputes are not analogous to expropriations. First, unlike expropriations, disputes cannot be equated with breaches of international law and are therefore not useful indicators of political risk. Considerable legal ambiguity characterizes arbitral claims, which mainly address indirect rather than direct expropriation and tend to target democracies rather than rent-seeking autocracies. Second, investment disputes may yield sizable monetary awards of damages to investors, but states do not ordinarily stand to make comparable gains. Even when they win, states merely reduce their potential material and reputational losses without realizing any benefits relative to the status quo ante. Cases involving outright expropriation are a partial exception, but they account for only 16 percent of cases filed over the past three decades (Pelc 2017, 562). In the contemporary legal setting, the costs of expropriating foreign investments are also high. Even Latin American states that have vigorously challenged the legitimacy of the international investment regime have been compelled to resolve awards for expropriation, often via the transfer of funds to thirdparty firms specializing in the purchase of rights to collect unpaid awards. The expected utility of dispute involvement for states is therefore distinctly limited, along with the availability of third-party funding, which has become an established feature of international treaty arbitration (Cremades 2011).
The difficulties are particularly pronounced for governments of small states facing off against large multinational corporations. A case in point is Chevron v. Ecuador (2006), in which the host state's total government budget amounted to 3 percent of the annual revenues of the claimant (World Bank 2015; Statista 2017). Even low-quality claims may yield material gains for claimants by deterring states from imposing regulatory, tax, or other burdens on foreign investments (Pelc 2017). Given these asymmetries, investors have incentives to be aggressive, especially if they can secure third-party funding for their legal expenses. Thus, even if they expect to lose at arbitration, firms may initiate proceedings for strategic reasons, particularly against states that do not possess the financial and legal resources or "rule of law" reputations that help insulate wealthy states from investor claims. Taking into account cases resulting in discontinuances or settlements, investors obtain compensation in 28.9 percent of all arbitral claims against middle-and lower-income states as compared to a figure of only 15.1 percent for upper-income respondents.
The second major assumption guiding the subsequent analysis is that states and investors decide strategically whether to participate in investment treaty arbitration. Disputes do not proceed immediately to arbitration but are typically subject to requirements of good-faith consultation and negotiation as well as a three-to six-month "cooling-off" period (Pohl, Mashigo, and Nohen 2012, 17). Disputants have further opportunities to decide whether to settle their differences during the period separating an investor's notice of arbitration from formal dispute registration, which averages 100 days but has extended up to 767 days at the ICSID. The time from registration to the constitution of an ICSID arbitral tribunal averages another 180 days (Sinclair, Fisher, and Macrory 2009). Arbitral involvement is consequently a product of the choices made by political leaders and investors after a conflict is initiated. Given the asymmetries in the potential benefits available to states and investors, the former have the strongest incentives to opt for dispute avoidance, depending on the domestic payoffs of standing up to investors. For investors, in contrast, the strategic calculus is arguably tipped in favor of pursuing a claim, making dispute continuance more likely than discontinuance. The differential access of investors and states to third-party funding of legal costs enhances these propensities. These expectations are consistent with the evidence on dispute outcomes, which shows that final awards are roughly three times more likely than settlements, and settlements nearly twice as likely as discontinuations. The latter typically arise out of the failure of investors to pursue their claim. The main point is not so much that investors always gain by pursuing arbitration, but once they have initiated the process, they are unlikely to decide to abandon their claims, except in response to a favorable settlement agreement.
Drawing on these theoretical considerations, four related hypotheses are advanced to explain variations in investment dispute involvement across Latin America. Looking first at the incentive side of the issue, the propensity for states to become involved in ISDS is expected to vary with the ideology and policy preferences of political leaders. Given the negative reaction of financial markets to the mere prospect of a leftist electoral victory in the region (Martínez and Santiso 2003), the international political economy generates strong pressures for leftist leaders to eschew policies that threaten investors or otherwise undermine business confidence in their countries. Although this prediction resembles those advanced by Popa-Dorobantu (2010) and Pinto (2013), the underlying logic is very different. Both authors argue that leftist leaders favor FDI because of its positive distributive consequences for labor, whereas right-leaning governments advance policies to shelter domestic firms from foreign competition. The relative gains of domestic interests from FDI, however, are arguably secondary in importance to Latin governments relative to concerns about retaining the confidence of international business. The latter problem is particularly acute for governments of the Left but attenuated for those of the Right, whose partisan core is built around business and financial interests. Thus, paradoxically, ideological orientation creates incentives for governments from both ends of the political spectrum to avoid involvement in international treaty arbitration, making dispute participation most likely for centrist governments: H 1 : Governments of the Left are less likely to become embroiled in investment treaty arbitration than other governments.
H 2 : Governments of the Right are less likely to become embroiled in investment treaty arbitration than governments ideologically positioned at the center of the political spectrum.
State involvement in investment treaty arbitration is also expected to vary with foreign exchange constraints, which make politicians particularly sensitive to the risks of losing access to foreign loans, trade, and investment, especially in countries with limited access to international capital markets. This expectation follows literature on Latin America showing that reducing foreign exchange constraints creates opportunities for the adoption of statist and leftist-populist policies, whereas the tightening of such constraints increases pressures for the adoption of more market-oriented economic policies (Kaufman 2011;Remmer 2012;Wibbels 2006;Campello 2012Campello , 2015: H 3 : The more positive the balance of payments, the greater is the likelihood that governments will become involved in ISDS.
Finally, the likelihood of disputes arising under investment treaties is expected to vary curvilinearly with democracy. At the upper end of the authoritarian-democracy spectrum, time-inconsistency problems may lead politicians to pursue short-term political gains at the expense of foreign investors, but this propensity is expected to be checked by political institutions, particularly courts, that constrain executives and offer foreign investors local remedies in the event of a dispute. Where horizontal accountability is more fragile, the opportunities for resolving investor-state disputes via host country institutions are more limited, creating incentives for investors to turn to investment treaty arbitration. The story, however, is not merely one of constraints or opportunities but also one of incentives. Hence dictators may exercise unrestrained political power, but the motivation for challenging foreign investors is likely to be weak, if not entirely absent, under nonelectoral forms of governance. In contrast, political leaders whose survival depends on elections may gain politically by adopting policies that provoke investment disputes, particularly in Latin America, where, as documented earlier, mass publics strongly support nationalistic policies and government economic intervention: H 4 : The probability of state involvement in ISDS is curvilinearly related to regime type, with investorstate disputes most likely at the lower end of the democracy spectrum, where electoral incentives for redistribution and the protection of domestic interests are high but judicial and legislative checks on executive power are relatively weak.
In addition to these four hypotheses, the statistical model also controls for other factors that are expected to shape the probability of dispute involvement: investment treaty participation, passage of time, and extractive resource rents. First, the potential for dispute involvement is expected to vary directly with the number of international investment agreements previously ratified by a country. The odds of a treatybased investment dispute are obviously zero for nations that have not yet ratified a single international investment agreement (e.g., Brazil). The passage of time represents a related issue in the sense that steps toward the creation of enforceable investor protections have become progressively institutionalized over time, strengthening the incentives for investors to file claims and thereby exposing states to a growing risk of dispute involvement.
The third control variable is designed to address the impact of economic structure. Following literature on the "natural resource trap," the clash between the ex ante and ex post incentives for cooperation with foreign investors is likely to be high for investments in extractive industries, which are peculiarly susceptible to time inconsistency problems. Not only is FDI sunk into particular mines or oil fields not readily shifted elsewhere; considerable time is likely to separate the initiation of a project and the realization of profits. Price fluctuations in commodity markets exacerbate the potential for tensions between states and investors. When prices go up, enhancing investor returns, governments may attempt to capture more of the revenue stream than provided for in ex ante arrangements (Hogan, Sturzenegger, and Tai 2010, 2-3). As a result, relationships between host-country governments and private investors can sour, thereby increasing the potential for investor-state dispute involvement in countries with sizable extractive sectors.
To operationalize these theoretical expectations, this article draws on a data set on international investment disputes that builds on two main sources: UNCTAD (2015a) and ICSID (2016). The dependent variable is an event count measuring the number of investment treaty disputes initiated per country-year in the LAC region between 1987 and December 31, 2014. To measure the independent variables, the analysis draws on data from the World Development Indicators (World Bank 2015): the external balance of goods and services as a percentage of gross domestic product (GDP) and total natural resource rents (including oil, gas, mineral, and forest rents) as a percentage of GDP. Two additional variables are drawn from the same source and included in the analysis as a check on the robustness of the results: namely, GDP (lagged and logged) and ratio of FDI to GDP. Democracy ratings are drawn from Polity data, which rank political regimes on a 20-point scale ranging from -10 to 10 (Center for Systemic Peace 2015). For the purposes of this analysis, these scores have been standardized on a scale ranging from 0 to 20. Involvement in IIAs is coded on the basis of UNCTAD's (2016b) International Investment Agreements Navigator. Finally, political ideology is assessed on the basis of codings from the Latin American Electoral Dataset (Murillo, Oliveros, and Vaishnav 2010), updated and supplemented by additional sources on the Caribbean ("Haiti" 2015;"Jamaica" 2015; "Leftist Parties of the World" 2016; Political Database of the Americas 2016). Center-right and rightist leaders are combined into a rightist category, consisting of 42.8 percent of total observations, whereas leaders of the left or center-left are coded as "leftist." The percentage of total observations falling into the latter category equals 22.2, as compared to 35.0 percent classified as centrist.
It should be emphasized that a pivotal advantage of focusing on the Latin American region is the availability of relatively high-quality data on government orientation. Although codings of this type are also provided by the global Database of Political Institutions (Keefer 2015), their availability and reliability are limited. Data on leadership orientation are missing for nearly 45 percent of the observations in the global database. Thirty-two percent of the LAC codes derived from the global database also differ from those of Murillo, Oliveros, and Vaishnav. That these differences are not simply a matter of borderline "judgment calls" is evidenced by the case of Chile, where the boundaries between leftist, rightist, and centrist are unusually well defined. Keefer (2015) nevertheless codes one of the centrist Christian Democratic governments of the 1990-2009 period as centrist and another as leftist; misdates the transition from Christian Democratic to Renovación Nacional governance by one year; misclassifies the right-wing government of Sebastián Piñera as centrist; and inexplicably offers no data for 2014, when the Left returned to power. If an error rate of nearly 40 percent characterizes the codings for one of the most studied and ideologically well-defined party systems in the less industrialized world, the advantages of focusing on a set of cases for which reliable data are available are clear.

Empirical Results
With the exception of small Caribbean nations with populations of less than five hundred thousand, for which no Polity ratings are available, the data set covers all nations of the LAC region: 24 countries, 649 country-year observations, and a total of 194 formally registered treaty-based disputes. As indicated by the descriptive statistics in Table 2, investment treaty disputes tend to be relatively infrequent, with the mean dispute count per country-year equal to only .30. Zeros consequently make up 70 percent of the observations. Given this overdispersion as well as the panel structure of the data, the analysis draws on a random-effects time-series negative binomial event-count model. This model not only is appropriate for the analysis of overdispersed event counts but also is capable of addressing questions about both crosssectional and intracountry variation. One obvious alternative is a model that includes fixed effects, but such a model is useful only when the questions of theoretical interest are limited to the study of the impact of variables that change over time (Williams 2017, 4;Torres-Reyna 2007, 9-10). Because key independent variables, including the lagged trade balance and Polity ratings, change little if at all over time, their impact cannot be assessed via fixed effect estimation. Fixed-effects estimation also makes it impossible to include countries that experienced no change in dispute involvement over the period under analysis, leading to the elimination of nearly 30 percent of the total number of observations. Alternative statistical approaches are discussed below in association with other robustness checks.
The basic results of the statistical analysis are presented in Table 3. Looking first at the coefficients presented in the first column of the table, as hypothesized a positive trade balance significantly increases the probability of an investment dispute. With all other variables set at their observed values and the lagged trade balance held at its lowest level, the predicted number of disputes is only .01. As the trade balance increases to its mean level, this figure rises to .21 and then twenty-fold more, to 4.32 predicted disputes, as the trade balance reaches its highest observed value. The impact of regime ratings is also highly significant, with the estimates conforming to the anticipated curvilinear pattern. As a country's Polity score rises from 0 to 12, the average predicted count of disputes increases from approximately zero to 1.3, but that number declines steadily thereafter. Hence at a Polity score of 12, which denotes a weak democracy, the predicted number of disputes is thirteen times higher than at the top end of the democracy spectrum. This is a pattern of change claimed but not empirically documented in prior research on investment treaty arbitration.
Also consistent with theoretical expectation are the coefficients for leadership ideology, which suggest that governments of the left and right of the ideological spectrum are significantly less likely to be drawn into investment disputes than those at the middle. With all other variables held at their mean, the predicted dispute count for all nonleftist leaders is 5 times that of leftist leaders, and that of non-rightist leaders is 2.2 times that of rightist leaders. What the results clearly imply is that the level of dispute involvement is lower under leftist than rightist leadership, but rightist leadership dampens dispute involvement relative to centrist political leadership. Converting the dummy variables for ideology into a single categorical predictor variable with three levels, ranging from 1 for leftist leaders to 3 for rightist leaders, helps clarify this issue. As suggested by the results presented in the second column of Table 3, in which the indicator variable for centrist ideology (group 2) measures the difference between centrist and leftist leaders and the indicator variable for group 3 the difference between the right and the leftist reference group, leftist ideology is not associated with increased dispute involvement. Instead, centrist leaders are more likely to become involved in disputes than leaders of either the left or the right. Translating these coefficients into incident rates, the dispute incident rate for centrist leaders is more than five times that of leftist leaders, whereas that for rightist leaders is 2.3 times that of the leftist reference group. These findings reflect the pro-business orientation of rightist governments, but they also speak directly to the relevance of "games of confidence" that place leftist governments in countries dependent on international capital inflows under strong pressure to adopt policies that to cater to the preferences of fund managers, bond-rating agencies, and foreign investors (Martínez and Santiso 2003;Campello 2015). These pressures are felt across the full range of Polity scores, making the predicted number  Wald 2 104.05*** 104.05*** 92.69*** 94.08*** 77.73*** 136.08*** Note: Columns 1-4 present results from a random effects negative binomial regression for panel data; column 5 the results for the negative binomial portion of a zero-inflated negative binomial regression model. Standard errors in parentheses. Column 6 reports results for a population-averaged negative binomial model with robust standard errors in parentheses. * p < 0.05. ** p < 0.01. *** p < 0.001. of disputes five times higher for centrist than for leftist governments even at a Polity rating of 20. Figure 1, which traces the impact of leadership preferences across the full range of Polity scores, clearly illustrates this propensity as well as the curvilinear relationship between political regime and dispute participation. Again, this is a new finding not documented by prior research on investor-state disputes.
The results in Figure 1 are very similar to model 3 in Table 3, which probes the robustness of the results to the inclusion of a dummy variable for Argentina. Mainly because of investor claims arising out of the financial crisis of 2001, the country represents an extreme outlier in terms of ISDS involvement. Nevertheless, the dummy variable is not significant, and estimates remain relatively consistent with those presented in the other columns of the table. Column 4 probes the issue of robustness further by including additional control variables for GDP (logged) and the relative importance of FDI flows to the economy, but the results remain consistent with those presented in other columns. Reestimating the results on the basis of a zeroinflated negative binomial model that combines a logit and negative binomial distribution with standard errors clustered by country yields similar results. To facilitate comparison with the other results displayed in Table 3, the results for the logistic component of the model, which addresses the probability of a nonevent, are not displayed. Instead, the column presents results for the event count, which is modeled on the basis of negative binomial regression and yields results that are relatively, albeit not fully, consistent with the other columns in the table. Finally, the last column of the table reestimates the first model on the basis of a timeseries population-averaged negative binomial model with robust standard errors to address potential issues of heteroscedasticity. Again, the results are quite consistent with the other estimations. Table 4 addresses the issue of dispute causation further by exploring key theoretical implications of the hypotheses regarding the impact of electoral competition and government ideology. Clearly, disputes vary enormously with respect to their stakes for both states and investors. The extremes range from claims of a few million dollars in compensation by individual investors in real estate ventures to disputes involving multibillion-dollar claims by multinational corporations with investments in long-term projects of major importance. The costs of confrontations between states and investors are therefore likely to be much higher for disputes in some sectors of the economy than in others, and the political calculus of dispute participants potentially different. Separating out investments in infrastructure and oil, gas, and mining from other dispute categories, I explore these differences. Disputes in the former category include some of the largest and most internationally visible arbitrations, thereby ratcheting up the risks for both states and investors. At the same time, in Latin America the infrastructural and extractive sectors have traditionally been characterized by extensive state intervention and significant domestic political sensitivity to foreign control dating back to the early twentieth century. Thus, to the extent that electoral incentives motivate political leaders to risk investment treaty arbitration, particularly strong evidence of that propensity should emerge from the study of infrastructural and extractive disputes. Likewise, the tendency for leftist political leaders to take a more accommodating stance toward foreign business interests should be particularly marked with respect to the most internationally visible disputes in these two sectors.
The results presented in the first two columns of Table 4 are consistent with both sets of expectations. As compared to disputes in other sectors, the evidence regarding disputes in the oil, gas, and mining sector, as well as those involving infrastructural investments, emphasize the importance of electoral incentives. For infrastructural investments, the average number of predicted disputes peaks at .58 for governments with a Polity rating of 12, as compared to figures of .001 and .04 at the bottom and top of the Polity spectrum, respectively. For extractive disputes, the comparable statistic is .99 for a Polity rating of 11, making the predicted number of disputes at the middle of the Polity spectrum ninety-nine times higher than at the upper end. The pattern for disputes not involving infrastructural or extractive investments is very different, however, with variations in Polity ratings having no statistically significant impact on dispute incidence.
The evidence with respect to ideological orientation is also consistent with expectation. Whereas ideology has no impact on dispute incidence in the residual sectors of the economy, leftist leadership is associated with significantly less dispute involvement in the infrastructure and the oil, gas, and mining sectors. In the former, the predicted number of disputes is five times lower for leftist than centrist leaders, and eight times lower in the oil, gas, and mining sector. These findings clearly emphasize the important role played by both political institutions and ideology in investment treaty arbitration.

Conclusion
Treaty-based legal protections for foreign investors constitute an increasingly important part of the international financial architecture. Although they are designed to encourage investment inflows, the resulting web of legal obligations also imposes new sets of risks, constraints, and uncertainties on hoststate governments. Nowhere is this more evident than in the Latin American and Caribbean region, whose states have been disproportionately challenged by investment treaty claims and by comparatively high rates of investor success in dispute arbitration. Not surprisingly, the political reaction against ISDS and investment treaty guarantees has also been unusually vehement in the region, leading a number of countries to eschew, formally denounce, or otherwise defect from the existing system of investor-state dispute arbitration. Within the framework of these broad regional trends, Latin American states nevertheless differ significantly with respect to their exposure to investment treaty arbitration. For some, investment disputes have been rare events; others have faced a long series of arbitral claims. The findings of this study indicate that these differences can be understood in terms of the incentives and opportunities facing both investors and domestic political leaders. After controlling for variations over time, level of treaty participation, and dependence on natural resource rents, three sets of factors increase the likelihood of dispute outbreak. The first is trade balance: the potential risks of reduced access to foreign capital are attenuated by trade surpluses, thus expanding opportunities for policy choices that may be popular with domestic audiences but also potentially threatening to the interests of foreign investors.
A second important finding concerns the impact of political ideology. Not only is rightist political leadership negatively related to dispute involvement; so is leftist leadership, which means that the propensity for dispute involvement is highest under centrist leaders. These findings underline the importance of international economic constraints on policy choice in Latin America. Governments that are relatively dependent on foreign sources of finance face strong economic pressures to gain and retain the confidence of international investors. These pressures are particularly intense for governments of the left, whose policy preferences deviate most from those of business interests. Buoyant international export markets may mute this tendency, but under other conditions, foreign exchange constraints motivate leftist leaders to signal their commitment to market-oriented policies. Conflicts with international investors, including those aired in ISDS processes, send out precisely the opposite kind of signal, creating inducements for leftist governments to avoid disputes by negotiating differences with foreign investors or otherwise finding ways to avoid formal arbitration proceedings. Consistent with the hypotheses about the role of leadership preferences, these tendencies are particularly marked in the case of unusually visible and high stakes disputes involving extractive and infrastructural investments.
The other consistently important variable to emerge from the analysis is political regime. Regimes situated near the middle of the Polity scale, including democracies characterized by relatively weak constraints on executive power along with "soft" or liberalized autocracies, are particularly prone to dispute involvement. In comparison, regimes ranked at the upper and lower ends of the autocracy-democracy spectrum are less likely to become participants in investor-state treaty disputes, especially highly visible extractive and infrastructural disputes involving unusually high stakes for investors and states alike.
For a region that has struggled over many decades to build and sustain democratic institutions, the significance of these findings can hardly be overstated. Although democracy is the predominant form of governance across the region, the quality of democracy remains an issue because of institutional fragilities that affect governance and the rule of law (Zovatto 2014). It is precisely these fragilities, particularly those relating to the relative independence of the executive branch vis-à-vis political parties and other branches of government, that have generated political incentives and opportunities for clashes with the foreign investment community.
The Latin American and Caribbean experience also reveals the dilemmas that investment treaty arbitration has created for elected leaders, who have increasingly been caught between popular political pressures at home and the constraints of the international investment regime. Operating under the umbrella of treaty protections, foreign investors have gained the right to challenge a broad range of government policies that potentially threaten their profits, including measures designed to protect indigenous rights, the environment, and public health. Investment treaties have repeatedly been portrayed as economically beneficial for less industrialized states. Less widely recognized are their potential political costs, particularly in Latin America, where citizen mobilization around such issues as environmental protection has collided directly with the interests of foreign investors. The findings of this study accordingly underline the challenges posed by investment treaty arbitration for democracies, particularly weak democracies, caught between domestic pressures for democratic accountability and international legal obligations. The growing international backlash against investor-state dispute settlement is a direct reflection of this tension.
Finally, for future research on the international investment regime, the findings of this study underline the potential benefits of disaggregating the international experience to explore variations within and across regions. By, inter alia, facilitating the collection of reliable data on the preferences of both mass publics and political leaders, limiting analysis to a single region creates opportunities for broadening the theoretical focus of research beyond the study of relatively fixed structural and institutional conditions to address the realm of political actors and agency and thereby develop a deeper and more nuanced understanding of the politics of investment disputes.