Transcription

1 WTO Working Paper ERSD Date: March 10, 2014 World Trade Organization Economic Research and Statistics Division Are Stricter Investment Rules Contagious? Host Country Competition for Foreign Direct Investment through International Agreements Eric Neumayer, Peter Nunnenkamp, and Martin Roy Manuscript date: March 10, 2014 Disclaimer: This is a working paper, and hence it represents research in progress. This paper represents the personal opinions of individual staff members and/or external contributors, and is not meant to represent the position or opinions of the WTO or its Members, nor the official position of any staff members. Any errors are the fault of the authors. Copies of working papers can be requested from the divisional secretariat by writing to: Economic Research and Statistics Division, World Trade Organization, rue de Lausanne 154, CH-1211 Geneva 21, Switzerland. Please request papers by number and title. 0

2 Contents 1. Introduction Related literature Theoretical argument and testable hypotheses Research Design IIA provisions and dependent variables Estimation technique, sample and explanatory variables Spatial lag variables Results Robustness tests Conclusion

3 Are Stricter Investment Rules Contagious? Host Country Competition for Foreign Direct Investment through International Agreements Eric Neumayer, Peter Nunnenkamp, and Martin Roy 1 March 2014 Abstract We argue that the trend toward international investment agreements (IIAs) with stricter investment rules is driven by competitive diffusion, namely defensive moves of developing countries concerned about foreign direct investment (FDI) diversion in favor of competing host countries. Accounting for spatial dependence in the formation of bilateral investment treaties (BITs) and preferential trade agreements (PTAs) that contain investment provisions, we find that the increase in agreements with stricter provisions on investor-state dispute settlement and preestablishment national treatment is a contagious process. Specifically, a developing country is more likely to sign an agreement with weak investment provisions if other developing countries that compete for FDI from the same developed country have previously signed agreements with similarly weak provisions. Conversely, contagion in agreements with strong provisions exclusively derives from agreements with strong provisions that other FDI-competing developing countries have previously signed with a specific developed source country of FDI. Keywords: bilateral investment treaties, preferential trade agreements, investment provisions, competition for FDI, spatial dependence JEL classification: F21; F53 1 Eric Neumayer London School of Economics and Political Science Peter Nunnenkamp Kiel Institute for the World Economy 2 Martin Roy WTO Secretariat

4 1. Introduction While the fundamental purpose of a bilateral investment treaty (BIT) is to encourage foreign direct investment (FDI) flows between country pairs (Bergstrand and Egger 2013), the empirical evidence that BITs are effective in stimulating FDI is ambiguous. 2 As noted by Swenson (2005), it is therefore not obvious that developing countries 3 sign BITs simply because these treaties help increase the inflow of FDI. Yet, the number of BITs and other international investment agreements (IIAs) continues to grow even in the absence of conclusive evidence as to the effects of BITs on FDI flows (Sachs and Sauvant 2009: LX). Furthermore, it appears that ever more developing host countries are accepting stricter FDI-related provisions in BITs and other IIAs, notably with regard to investor-state dispute settlement and preestablishment national treatment of foreign investors. This seems to have resulted in an unexpectedly large wave of litigation (Simmons 2014: 13), implying considerable costs and loss of sovereignty of developing host countries. This raises the question of why IIAs continue to be concluded, and what explains the willingness of developing countries to increasingly agree to strict and binding investment rules at the bilateral and plurilateral level. Another element of this puzzle is that developing countries have so far strongly objected to binding multilateral investment rules (Salacuse 2010). As Milner (2014: 7) points out, research on the investment regime has predominantly drawn on theories of competitive diffusion, power politics, and the rational design of institutions. Our contribution is firmly rooted in the first set of theories. We argue that the mushrooming of IIAs and the acceptance of stricter investment rules at the bilateral and plurilateral, as opposed to multilateral, level are mainly defensive moves of developing host countries being concerned about FDI diversion in favor of competing host countries that became parties to IIAs before. Accordingly, the increase in IIAs is fundamentally driven by a self-reinforcing or contagious process: Governments could be signing these treaties because, as more countries conclude more and more of these agreements, they could be afraid that investors may avoid investing in countries that have not signed such treaties (Sachs and Sauvant 2009: LX). Importantly, contagion may also help explain the increasing strictness of provisions in BITs and other IIAs: developing countries are caught in a race to conclude not only more such treaties but increasingly more stringent treaties. The argument that competition for FDI among developing countries drives the conclusion of BITs between developed source countries of FDI and developing host countries of FDI is everything but new (see, for example, Guzman 1997; Elkins et al. 2006; Jandhyala et al. 2011). Our original contribution to this literature is that we refine the theoretical argument and provide a superior empirical test for it. As discussed in Section 2, just a few empirical studies have addressed the diffusion of BITs by modeling spatial dependence in BIT formation. Furthermore, previous studies typically treated BITs and other IIAs as black boxes, ignoring the content and strictness of FDI-related provisions agreed upon between the source and host countries of FDI. Yet, it is exactly the differentiation by content and strictness that allows us to provide a superior test for the competitive diffusion hypothesis. 2 See the collection of papers in Sauvant and Sachs (2009) as well as Kerner (2009), Tobin and Rose-Ackerman 2011, and Allee and Peinhardt (2011). 3 We use the term developing countries as short cut for all countries that are not one of the developed FDI source countries listed in appendix 1. It therefore also includes countries often called countries in transition. 3

5 In terms of refining the theoretical argument, based on emerging, though as yet still mixed, evidence that BITs and other IIAs with stringent investment provisions increase FDI inflows more than treaties without such provisions (Berger et al. 2011, 2013; Büthe and Milner 2014), we argue that a developing country is most concerned about other developing countries concluding IIAs with major FDI source countries if these IIAs contain strict investor-state dispute settlement (ISDS) or pre-establishment national treatment (NT) provisions because these pose the greatest threat in terms of potential FDI diversion away from the country. As a corollary, spatial dependence in the form of FDI-competition driven pressure on a developing country to sign more stringent IIAs should therefore stem from the existence and diffusion of such IIAs with equally stringent provisions in other developing countries competing with the developing country for FDI from developed source countries. By contrast, the existence and diffusion of IIAs with less stringent or no provisions should not drive the diffusion of IIAs with more stringent provisions. Empirically, we are the first to employ existing bilateral FDI stocks as the best and most direct measure of what is at stake in terms of potential FDI diversion as weights in the spatial lags that capture competition among developing countries. Existing FDI stocks thus track much more closely the theoretical argument of FDI-competition driven spatial dependence among developing countries as a major driver of their willingness to sign IIAs and sign increasingly stringent IIAs. Based on estimations from a global sample of 21 developed source countries and 87 developing host countries over the period 1978 to 2004, we show that the increase in IIAs with stricter provisions on ISDS and pre-establishment NT is a contagious process. Importantly, however, contagion of IIAs with weak investment provisions exclusively derives from weak IIAs of FDI-competing host countries, while contagion of IIAs with strong provisions stems solely from strong IIAs of competing host countries. Our stringent and conservative research design shields the estimations from finding spurious evidence for competitive diffusion dynamics. At the same time, it renders it harder to find statistically significant evidence for competing theoretical perspectives. We stress that our findings should be interpreted as buttressing competitive diffusion as an important driver of the international investment regime, not as implying that power politics (e.g., Allee and Peinhardt 2014) or efficient institutional design (e.g., Koremenos 2007) cannot additionally play a complementary role. After reviewing the extant literature in Section 2, we put forward our theoretical argument that results in two testable hypotheses in Section 3. Section 4 describes our research design. We report results from our main estimations in Section 5, from robustness tests in Section 6, and conclude in Section Related literature Important gaps remain even though there is a growing literature on the determinants of BITs and preferential trade agreements (PTAs) that include investment provisions. Until recently this literature failed to address two crucial issues that figure prominently in our empirical analysis: First, it was not taken into account that the scope and depth of investment-related provisions differ considerably across BITs and other IIAs. Second, where an emerging literature started to look into the stringency of BITs and other IIAs, it assumed at least implicitly that such treaties, in particular the conclusion of BITs, were the result of purely bilateral initiatives unaffected by the behavior of other country pairs. This view neglects the impact of spatial dependence on whether a specific pair of source and host country of FDI decide to engage in BIT negotiations, that is, the impact that the treaty concluding behavior of other country pairs has on a specific country pair s willingness to conclude such a treaty. 4

6 Indeed, most of the literature treats BITs as a black box and neglects that the likelihood to conclude a BIT with far-reaching commitments may differ from that of concluding a weaker BIT. Swenson (2005), Elkins et al. (2006), Neumayer and Plümper (2010) and Bergstrand and Egger (2013) are all prominent examples in this regard. An emerging strand of the literature explicitly addresses the content of BITs, however. Effective dispute settlement provisions have received most attention so far. Allee and Peinhardt (2010: 2) note that legal scholars have singled out these investor-state dispute settlement clauses within BITs as perhaps the most important aspect of the treaties. According to Allee and Peinhardt (2010; 2014), developing host countries make more concessions on dispute settlement provisions when negotiating BITs with source countries that enjoy a particularly strong bargaining position. Simmons (2014) complements this finding and argues that host countries are more likely to agree to strict dispute settlement provisions in harder economic times, e.g., in periods of weak economic growth. Importantly, however, spatial dependence in the form of FDI-competition among developing host countries for FDI is not explicitly modeled by these authors. Another strand of the recent literature departs from Baldwin s domino theory of regionalism (Baldwin 1993) to overcome the purely bilateral perspective of analyzing the determinants of BITs and other IIAs. Baldwin (1993) develops a formal political economy model to show that an idiosyncratic event of economic integration among third countries triggers domino effects by changing the cost-benefit calculus of non-members. The triggering event threatens to harm the profits of competing outsiders, thus increasing their inclination to join existing integration schemes or initiate new ones. This process is driven by a peculiar tendency of special interest groups; they usually fight harder to avoid losses than they do to secure gains (Baldwin 1993: 4). Baldwin and Jaimovich (2012) as well as Baccini and Dür (2012) provide empirical analyses of interdependent formation of PTAs. The authors of both papers propose a contagion index to capture the extent to which a PTA between countries A and B changes country C s incentive to conclude a new PTA with either A or B in a defensive move to mitigate adverse effects from trade diversion. Bilateral trade relations are used to construct the spatially lagged contagion measure. However, FDI-related provisions in IIAs and FDI diversion are not considered in these papers. The logic of why countries do not decide in isolation on trade agreements can easily be transferred to the conclusion of BITs in general and the conclusion of BITs with stricter investment-related provisions in particular. As stressed by Baldwin (1997), nonmembers are concerned about trade diversion when their competitors engage in closer economic integration. In the case of BITs, this would imply that an agreement concluded between a pair of a host country and a source country of FDI increases the incentive of a competing host country to engage in BIT negotiations in order to avoid FDI diversion. The BIT boom would feed itself, even if each host country had a preference not to enter into BITs had competitors not done so before. Indeed, Elkins et al. (2006) find that the diffusion of BITs is associated with competitive pressure among developing host countries. These authors rely on network measures of economic competition as well as more indirect evidence on competitive pressures on the host to sign BITs (page 811). Specifically, Elkins et al. (2006) use three alternative spatial weights to proxy for competition among host countries of FDI: the similarity of the destination of exports, the similarity of the export product structure, and the similarity of educational and infrastructural endowments. Similarity in all three dimensions increases the odds of a BIT. Using similar proxies of competition for FDI among host countries, Jandhyala et al. (2011) allow for varying effects over time of these proxies and other determinants of BIT formation. They find that competition among host countries mattered for BIT signing throughout the period of observation, but the effects were stronger in the period than more recently. 5

7 Neumayer and Plümper (2010) refine the analysis of Elkins et al. (2006) by exploring specific channels through which BITs may diffuse. The results of Neumayer and Plümper suggest that the decision of a developing host country to sign a BIT with a developed source country depends only on other host countries BITs with the same source country, rather than other host countries BITs with any source country. In other words, Neumayer and Plümper (2010: 148) find that what they term dyad-specific target contagion matters rather than aggregate target contagion. Lupu and Poast (2013) propose another refinement by modelling the boom in BITs as a multilateral or k-adic process, rather than a dyadic process. 4 Nevertheless, Lupu and Poast (2013) corroborate Neumayer and Plümper (2010) in that host countries conclude BITs with specific source countries in order to divert FDI away from competing hosts of FDI by this particular source country. 5 However, these recent BIT studies analyzing spatial dependence have some common shortcomings that we attempt to overcome. Competition among host countries of FDI is typically proxied by spatial lags using trade relations or geographic distance as weights. Given that BITs and other IIAs raise concerns about FDI diversion in the first place, it is more appropriate to use existing FDI relations as weights as we do in the following. More importantly, none of these studies accounts for the content of BITs and other IIAs. As specified below, we contribute to closing this important gap by considering two essential treaty provisions: investor-state dispute settlement (ISDS) and pre-establishment national treatment (NT). Furthermore, we take into account that such provisions may not only be specified in BITs but also in other IIAs, namely PTAs with investment provisions. 3. Theoretical argument and testable hypotheses Like any theoretical argument that leads to testable hypotheses, ours too is based on a set of assumptions that, in our view, are persuasive, if perhaps not uncontroversial. On the part of developed countries, we firstly assume that these countries unambiguously prefer IIAs and prefer stronger to weaker investment provisions. The case for this assumption is clearest for BITs, which developed countries almost exclusively contract upon with developing countries. For such dyads, developed countries typically enjoy a strongly asymmetrical outward net FDI position such that the benefits almost exclusively accrue to foreign investors from the developed source country and the costs in terms of loss of sovereignty almost exclusively accrue to developing host countries. Our assumption is more problematic for PTAs with investment provisions since these are also concluded among developed countries themselves. These agreements impose some costs in terms of loss of sovereignty onto the developed countries even if developing countries also form part of the agreement witness for example the political controversy in the US and Canada surrounding chapter 11 of the North Atlantic Free Trade Agreement, which allowed Canadian (and Mexican) investors to sue the American government and American (and Mexican) investors to sue the Canadian government. However, our research design is restricted to dyads comprising developed FDI source countries and developing FDI host countries. It is outside the remit of our article and we thus cannot and seek not explain why developed countries join PTAs with investment provisions with other developed countries (Mansfield and Milner 2012). However, conditional on such PTAs potentially existing, it remains true that developed PTA member countries will prefer that developing countries join a PTA with investment provisions and ideally strong provisions as again the outward net FDI position is likely to be asymmetrically in favor of the developed country. We also note that developed countries were on the whole strongly in favor of the failed attempt at creating a multilateral 4 Note that a k-ad stands for a group of states with size k consisting of one source country and a varying number of host countries, including dyads with just one host country. 5 By contrast, other types of contagion appear to have negative effects on the process of BIT formation. 6

8 agreement on investment, which suggests that they are not too concerned about committing to binding investment provisions that investors from other developed countries could take advantage of (Henderson 1999). Secondly, we further assume on the part of developed countries that any single developed country s success in convincing a specific developing country to accept (strict) investment provisions does not create a major competitive disadvantage for other developed countries. For example, if the United Kingdom manages to convince India to accept strict investment provisions, then this is beneficial to UK investors, but not necessarily disadvantageous to German investors. That is not to say that Germany is not interested in concluding similarly strict investment provisions with India. But this interest existed even before and is not necessarily affected by the agreement between the UK and India. 6 This assumption necessarily presupposes that any potential increase in FDI flowing from the UK to India following the conclusion of the agreement with strict investment provisions between the two countries does not crowd out German FDI flowing into India. This, in turn, means that developing FDI host countries are not close to a binding FDI absorption constraint. We do not find this assumption problematic given the hugely under-developed state of the economies of many developing countries and the large potential, at least in principle, for foreign investors to earn high profits from exploiting the economic opportunities that come from economies operating far from any long-run steady state. This second assumption relating to developed FDI source countries allows us to focus on FDI-competition driven spatial dependence among developing countries resulting in what Neumayer and Plümper (2010) dub specific target contagion without the need to simultaneously model what these authors term specific source contagion, which would model competition among developed countries. On the part of developing countries, we assume that all other things equal they prefer not to sign IIAs with investment provisions that curtail their sovereignty to impose conditions on foreign investors since they typically are net FDI importers and would thus predominantly experience the costs without their own foreign investors enjoying much benefit from the investment provisions. Yet, all other things are not equal. First of all, the cost in terms of loss of sovereignty needs to be balanced against any potential increase in inward FDI following from signing such an agreement, which we assume to be beneficial, at least in expectation, for the developing host country. For this to play a part in the benefit-cost consideration of developing countries, the mixed evidence with regards to whether IIAs actually result in more FDI, referred to in the Introduction, is not fatal since all that is needed is that developing country policy-makers believe that these treaties result in more FDI. It must also be true, however, that at least initially developing countries expect the costs to be larger than the potential benefits since otherwise they would all rush to the negotiating table to conclude bilateral IIAs with strict investment provisions. This is also consistent with their refusal to negotiate any multilateral agreement on investment. Instead, what we observe is that some frontrunner developing countries sign such agreements before others. Poulsen and Aisbett (2013) argue that developing countries might have ignored the risks this entails. Whilst we cannot exclude this possibility, another reason is that frontrunners might enjoy an early mover advantage as foreign investment is diverted from locations that refuse to offer provisions favorable to foreign investors toward locations that have committed to such provisions (Guzman 1998). Such FDI diversion will increase the expected benefits of signing IIAs, thus tilting the expected benefit-cost ratio in their favor. This, in turn, creates a negative externality onto other developing countries in the form of FDI diversion. This leads directly to our core argument, namely that developing host countries cannot ignore the behavior of other developing countries with whom they compete for scarce FDI from developed source countries. Every competitor who concludes an IIA with (strict) investment provisions with a specific developed FDI source country poses a threat for a developing country that some of the existing or, more 6 In other words, Germany has little reason to act defensively against a prior move of the UK. 7

9 likely, future FDI from this developed country will be diverted away from it. For this argument to have bite it must be true that either the developed country operates at or close to a binding constraint on the amount of FDI that can flow out of the country or that the quality of FDI differs and the FDI that is more desirable to developing countries comes from investors who are keener on strict investment provisions. This is because without such a constraint or, alternatively, such heterogeneity in the desirability of FDI flows no detrimental diversion can take place. Whether developed countries operate close to such a constraint is hard to say, but some constraint exists of course since FDI needs to be financed by diverting income away from domestic purposes. That FDI is heterogeneous in its desirability to developing host countries is plausible since some FDI will lead to more local job creation and knowledge spillovers than others. Foreign investors who perceive that their investment is particularly desirable to developing host countries in turn have an incentive to expect better investment protection provisions since their investment flows are preferred over those of other investors. Once some developing countries have started signing IIAs, they have set in motion a contagious process that over time induces more and more developing countries to follow suit. Collectively, developing countries would be better off refusing to sign away national sovereignty, which is also why they refuse a multilateral treaty. But individually some took advantage of early mover advantage, leaving those with whom they compete for scarce FDI from developed source countries with little choice than to give in, too. Thus, the pressure on a single country to sign an IIA is the greater the larger the share of competitors that have already signed a treaty. The logic of our argument can also explain the stylized facts of the dynamics of rolling-out investment provisions of different strengths over time (see Figure 1). Developing countries do not favor investment provisions, but they dislike strong investment provisions even more than weak provisions. Hence, in the early periods IIAs with weak provisions will dominate. Yet, the same temptation that induced some developing countries to sign IIAs with weak provisions in the beginning in order to seize a first mover advantage lures them or others into signing IIAs with strong provisions such that at some point these become the dominating type of investment provisions. 8

10 Figure 1: Diffusion of international investment agreements over time End year of period BITs w weak ISDS BITs/PTAs w weak ISDS/NT BITs w strong ISDS BITs/PTAs w strong ISDS/NT Note: For variable definitions, see Section 4. A crucial implication of our argument is that the pressure that comes from the IIA signing behavior of other developing countries with whom a developing country under observation competes in terms of FDI from a specific source country will be exclusive to the specific strength of investment provisions competitors have agreed to. Thus, one s competitors having signed treaties with weak provisions only exerts pressure on a developing country to sign a treaty with weak provisions to remove the previously created competitive disadvantage and avoid FDI diversion. In other words, weak provisions in competitors treaties do not induce developing countries to sign treaties with strong provisions since these are not necessary and carry greater costs. Conversely, one s competitors having signed treaties with strong provisions exerts pressure on a developing country to sign a treaty with equally strong provisions, which are now needed to remove the competitive disadvantage and avoid FDI diversion. Strong provisions in competitors treaties do not induce developing countries to sign treaties with weak provisions since weak provisions are insufficient to counter the competitive advantage that one s competitors have previously created for themselves. Our reasoning therefore results in the following two testable hypotheses: Hypothesis 1. IIAs with weak investment provisions signed by a larger number of one s competitors for FDI from a specific developed source country increases the incentive of a developing host country to also sign an IIA with weak provisions with this developed country, but not an IIA with strong provisions. 9

11 Hypothesis 2. IIAs with strong investment provisions signed by a larger number of one s competitors for FDI from a specific developed source country increases the incentive of a developing host country to also sign an IIA with strong provisions with this developed country, but not an IIA with weak provisions. 4. Research Design IIA provisions and dependent variables While the conclusion of new BITs has slowed down considerably since the early 2000s, possibly because many countries became more reluctant after having experienced their first legal challenges (Poulsen and Aisbett 2013), the cumulative number of all IIAs reached almost 3,200 at the end of 2012 (UNCTAD 2013). BITs accounted for almost 90 percent of all IIAs. In addition, UNCTAD (2013) lists 339 other IIAs, defined as economic agreements, other than BITs, that include investment-related provisions (essentially investment chapters in PTAs). In the following, we analyze the diffusion of BITs on their own, but also of BITs and PTAs with investment provisions together. We thus also analyze the diffusion of IIAs more broadly, not just BITs. 7 The proliferation of IIAs has taken place jointly with a transformation of the content of IIAs, resulting in increasingly strict obligations. Importantly, IIAs differ in whether and to what extent they contain critical legal provisions that diffused over the last two decades (Berger et al. 2013). The two most important features relating to the liberalization and protection of foreign investment appear to be: (i) guarantees of market access for foreign investors, i.e., the extent to which IIAs include provisions on national treatment (NT) in the pre-establishment phase; and (ii) the extent to which IIAs include a strong investor-state dispute settlement (ISDS) mechanism, which is key in ensuring that foreign investments are effectively protected from discriminatory or abusive treatment in the host country. As noted in Section 2, dispute settlement has received more attention in the previous literature than liberal admission rules in the form of pre-establishment NT. The relevant question is whether foreign investors can effectively sue host country governments before an international arbitration tribunal for breaches of treaty obligations, without having to exhaust local remedies or to obtain the host government's prior consent. To capture the variation in such ISDS provisions we follow the classification of BITs by Yackee (2009), 8 which we extended to PTAs applying the same classification. Accordingly, the strongest type of ISDS (coded as 3) offers comprehensive pre-consent concerning the investors possibility to unilaterally initiate binding international arbitration of disputes. Partial pre-consent (coded as 2) restricts this possibility to a limited class of disputes such as disputes on the compensation for expropriation. So-called promissory provisions without guarantee of international arbitration for the investor offer a weaker type of ISDS (coded as 1), while the lack of any ISDS provisions is coded as 0. Over time, the proportion of BITs with strong ISDS has grown significantly; stricter ISDS now tends to be the norm in BITs and other IIAs negotiated since the mid-1990s (see Figure 1). 9 As a result, UNCTAD notes that the number of new cases of ISDS has increased considerably since the mid-1990s (see also Simmons 2014) By restricting our analysis to BITs and PTAs we may miss a very small number of IIAs that come neither in the form of BITs nor PTAs 8 Yackee s classification has previously been employed in Berger et al. (2011; 2013). 9 However, some recent IIAs do not include (strict) ISDS provisions, e.g., IIAs involving the European Union. 10 The total number of known cases exceeded 500 in Investors have challenged a broad range of measures by hostcountry governments, including revocations of licenses, irregularities in public tenders, changes in domestic regulations, withdrawal of subsidies, direct expropriations, and tax measures. For details, see: (accessed: January 2014). 10