عوامل تعیین کننده در پیمان سرمایه گذاری دو طرفه چیست؟

Similar to bilateral or regional preferential trade agreements (PTAs), bilateral investment treaties (BITs) have proliferated over the past 50 years. The purpose of this study is to provide the first systematic empirical analysis of the economic determinants of BITs and of the likelihood of BITs between pairs of countries using a qualitative choice model, in a manner consistent with explaining PTAs. We develop the econometric specification for explaining the two based upon a general equilibrium model of world trade and foreign direct investment with three factors, two products, and trade and investment costs among multiple countries in the presence of national and multinational firms. The empirical model for BITs and PTAs is bivariate in nature and supports a set of hypotheses drawn from the general equilibrium model. Using the preferred empirical model for a sample of 12,880 country-pairs in the year 2000, we predict correctly 88% of all pairs with a BIT and a PTA, 81% with a BIT but no PTA, and 84% with a PTA but no BIT.

“The primary problem for researchers wishing to assess the impacts of policies to promote FDI is that policy adoption is endogenously determined.” ( Aisbett, 2009, p. 396) “The literature on BITs is limited, making it difficult to truly understand the determinants for signing.” ( Tobin and Rose-Ackerman, 2005, p. 15) One of the most notable economic events since World War II is the proliferation of preferential trade agreements (PTAs), including predominantly free trade agreements (FTAs) and some customs unions (CUs). However, the proliferation of bilateral investment treaties (BITs) has been significant as well. For instance, in 2010 the U.S. government had 40 BITs in force while it had only 17 PTAs in force. Fig. 1 presents the numbers of BITs in existence in the world in every year from 1980 to 2007. Moreover, Table 1 indicates the numbers of country-pairs with BITs and with PTAs (including those with both) in the year 2000 for 161 countries. Table 1 shows, for this sample of 12,880 country-pairs, 923 pairs with a BIT but no PTA, 1478 pairs with a PTA but no BIT, and 556 pairs with both.2 Full-size image (20 K) Fig. 1. Number of BITs in the world by year, 1980–2007. Figure options Table 1. Bilateral investment treaties (BITs) and preferential trade agreements (PTAs) across 12,880 country-pairs in the year 2000. PTAs Sum Yes (1) No (0) BITs Yes (1) 556 923 1479 No (0) 1478 9923 11,401 Sum 2034 10,846 12,880 Notes: There are 161 countries (12,880 pairs) in the sample. Sources: WTO and UNCTAD. Table options Yet in contrast to the vast international trade literatures on the theoretical net benefits and costs of FTAs and CUs and on the empirically-estimated effects of FTAs and CUs on trade flows, the literature on BITs is not only considerably smaller but dominated by legal and political science scholars rather than economists (cf., Salacuse, 1990, Vandevelde, 1998, Vandevelde, 2000, Tobin and Rose-Ackerman, 2005 and Buthe and Milner, 2009).3 Consequently, none of these papers address factors explaining BITs' formations using formal theoretical economic models, and few provide econometric empirical analyses. Also, relative to the trade and PTA literature, there are very few papers – some by economists and some by legal/political scholars – that have looked systematically and econometrically at the impact of BITs on foreign direct investment (FDI). Hallward-Driemeier (2003), Tobin and Rose-Ackerman (2005), Gallagher and Birch (2006), and Aisbett (2009) find little economically and statistically significant effect of BITs on FDI flows. By contrast, Egger and Pfaffermayr (2004a), Salacuse and Sullivan (2005), Neumayer and Spess (2005), and Buthe and Milner (2009) find economically and statistically significant effects.4 Furthermore, there is no study trying to systematically explain empirically the economic determinants of BITs — much less one motivated by a formal general equilibrium model. This paper addresses this shortcoming. 5 In this paper, we examine theoretically and econometrically the economic determinants of BITs — in a manner consistent with understanding the economic determinants of PTAs. While BITs have been examined much less in the international economics literature, the motivation for such agreements for FDI is actually quite similar to that for PTAs for trade. While “Friendship, Commerce and Navigation” treaties surfaced as early as the 18th century, modern BITs were effectively created in the late 1950s in response to numerous expropriations of FDIs as well as the limitation of the General Agreement on Tariffs and Trade (GATT) to trade only; (West) Germany concluded the first modern (post-World War II) BIT with Pakistan in 1959. The first modern BITs were intended to reduce for home countries the relative cost of FDI outflows by reducing the risk of “expropriation” by host countries' governments. Hence, the likelihood of a BIT should be positively related to the degree of expropriation risk, other things equal. More recently, BITs have addressed FDI-related issues beyond expropriation risk to promote investment liberalization. UNCTAD (2007) notes that many of the existing BITs guarantee foreign investors fair-and-equitable, non-discriminatory, and “national” treatment. Consequently, more recently BITs have been spurred by host countries as “instruments” of investment liberalization to encourage capital exporting countries to provide FDI inflows to developing and developed capital importing countries, much as PTAs have proliferated as instruments of trade liberalization among and between developed and developing countries. Since the fundamental purpose of a BIT is to encourage FDI flows between country-pairs by reducing the relative cost of FDI and that of a PTA is to encourage trade between country-pairs by reducing the relative cost of trade, economic determinants of BITs may well share many similarities to those of PTAs. Since there has been no previous formal theoretical or econometric model of the determinants of BITs, our starting point is the new literature on the economic determinants of PTAs. This literature, surveyed in Freund and Ornelas (2009), starts with Baier and Bergstrand (2004), or BB, who developed a qualitative choice econometric model of the likelihood of a pair of countries having a PTA in a given year.6 Motivated by a general equilibrium model of world trade with two factors, two monopolistically-competitive markets with national exporting firms, and explicit intercontinental and intracontinental trade costs among multiple countries on multiple continents, the BB model suggests that country-pairs are more likely to have a PTA: (1) the closer together they are; (2) the more remote they are from other markets; (3) the larger their joint economic size; (4) the more similar their economic sizes; and (5) the larger the difference in the pairs' relative factor endowments (up to a point). BB showed that all these economic factors were economically and statistically significant (with expected signs) in explaining cross-sectional variation in country-pairs' probabilities of having PTAs with a pseudo-R2 of 73%. Using a larger sample of 10,585 pairs in the year 2000, Egger and Larch (2008) predicted correctly about 781 of the 1263 pairs with PTAs (or 62%). Their pseudo-R2 was considerably lower at 27% (as expected) due to their much larger and less selective sample. However, the economic determinants of BITs are not likely to be explained by the same econometric model, due to several considerations. First, BITs potentially influence FDI flows. Consequently, while economic size and similarity help to predict PTAs, they may not simultaneously predict BITs; as mentioned, most BITs are between developed and developing countries (and the latter tend also to be economically smaller than the former). Other factors – such as bilateral trade and investment costs and relative factor endowments – are likely to have differing effects on explaining BITs relative to PTAs.7 Since FDI is generated by multinational enterprises (MNEs), a theoretical framework should incorporate MNEs' behavior; consequently, a simple Helpman–Krugman–Heckscher–Ohlin general equilibrium model of trade as in BB is insufficient. An extension of the BB framework to include MNEs, FDI flows, and foreign affiliate sales (FAS), in the spirit of the “Knowledge–Capital” (KC) models of Markusen (2002) and Markusen and Maskus, 2001 and Markusen and Maskus, 2002, is a natural direction. Fortunately, Bergstrand and Egger (2007), or BE, extended the 2 × 2 × 2 KC model to three factors and three countries, and provide a ready framework to address the economic determinants of BITs and PTAs. BE is especially relevant because it is the first general equilibrium model to demonstrate that bilateral FDI and trade are maximized between countries with identical relative and absolute factor endowments, consistent with the large literature on gravity equations that explain very well both bilateral trade and FDI flows. 8 Thus, the first potential contribution of this paper is to use the theoretical framework in BE to generate new comparative statics to show (initially, in the absence of any relative factor endowment differences): (1) how economic size and size similarity of two countries influence their net utility gains (or losses) from a BIT and from a PTA; (2) how bilateral investment and trade costs between two countries influence such gains; and (3) how interactions among these factors influence these gains. Note that, in general equilibrium, the net utility gains from BITs and PTAs are influenced by the behavior of multinational enterprises as well as national (exporting) firms. In the presence of MNEs and general equilibrium, the influences of these economic variables on the net utility gains from a PTA are not necessarily the same as those found in BB, where MNEs and FDI were absent. Second, in reality relative factor endowments are not identical across countries, and such differences matter for economic determinants of BITs because many BITs (not to mention PTAs) are “North–South” in nature, that is, between countries with quite different relative factor endowments.9 The second potential contribution of this paper is to show the relationship using the BE model between relative factor endowment differences between two countries and the net utility gains from BITs and from PTAs. With three factors – skilled labor, unskilled labor, and physical capital – the relationships are complex. However, using traditional Edgeworth boxes and recent developments in specifying properly the relationships between relative factor endowments and bilateral FAS flows in Braconier et al. (2005), our theoretical relationships suggest easily specified empirical counterparts to capture some of the influences of relative factor endowment differences on the net utility gains from BITs and PTAs in the presence of national exporting firms, horizontal MNEs, and vertical MNEs.10 Moreover, we examine the interactive effects of relative factor endowments and investment costs on utility gains from such agreements. Third, guided by the theoretical comparative statics, we specify a bivariate probit model of the probabilities of BITs and PTAs existing between country pairs in the year 2000. We choose to estimate a bivariate probit model because the error terms may be correlated across probabilities, and this provides more efficient coefficient estimates. To anticipate some of the results, we find the following empirical conclusions. First, as much of trade is “intra-industry” and much of FDI is “horizontal,” one would expect that the net utility gains from a BIT and from a PTA are positively related to economic size and similarity. Such results are confirmed here. Second, our theoretical model suggests that a higher initial level of expropriation risk should increase the gains from a BIT, and higher natural investment costs (such as political instability) should decrease the gains from a BIT. Our theoretical model also suggests that a higher initial level of expropriation risk should decrease the gains from a PTA, and higher natural investment costs should increase the gains from a PTA. Using measures of investment costs that should influence FDI and trade oppositely, we find evidence of these “substitution effects.” Moreover, due to the non-linear probit model, we can estimate the sensitivity of the effects of investment costs on the probability of a BIT for a country-pair to their economic size. Third, in the presence of three factors of production, the relationships between relative factor endowments and net utility gains of a pair of countries from a BIT or a PTA are complex, non-linear, and non-monotonic. However, we draw upon the geometric features of an Edgeworth box to introduce a measure of dissimilarity of factor shares that helps explain the importance of relative abundance of skilled labor for increasing the net utility gains from a BIT for a country-pair.11 Alongside another (more standard) measure of deviations of capital–unskilled-labor ratios from the Edgeworth box diagonal, we show empirically how relative factor endowments affect the probabilities of a BIT and of a PTA, and estimate the sensitivity of the effects of investment costs to relative factor endowments. Finally, the bivariate empirical model has a relatively high explanatory power that holds up to an extensive sensitivity analysis, including extensions to “third-country-pair” effects and lagged effects. Moreover, the inclusion of different relative factor endowment variables for the BIT and PTA equations allows for potential identification in a (recursive) simultaneous equations system, which we explore. Using our sample of 12,880 country-pairs in the year 2000, we predict correctly 88% of the 556 country-pairs with a BIT and a PTA, 81% of the 923 country-pairs with a BIT but no PTA, and 84% of the 1478 country-pairs with a PTA but no BIT. The overall pseudo-R2 of 28% for our bivariate probit model is comparable to the 27% pseudo-R2 found in Egger and Larch (2008) for PTAs using a univariate probit model and a similar large sample. The results provide quantitative guidance as to the determinants of BITs and PTAs simultaneously (and also sequentially) and to addressing potentially the endogeneity bias inherent in many previous empirical analyses of the effects of BITs on FDI flows. The remainder of the paper is as follows. In Section 2, we summarize the theoretical BE model and the parameter values chosen for the numerical version of the general equilibrium model. In Section 3, we discuss the general equilibrium comparative static results for the relationships between the net utility gains from a BIT and from a PTA with economic size and similarity, investment costs, and trade costs, assuming identical relative factor endowments across countries. In Section 4, we relax the assumption of identical relative factor endowments and, using conventional Edgeworth boxes and their geometric properties, provide general equilibrium comparative statics to motivate two relative-factor-endowment variables for the empirical analysis. In Section 5, we describe our econometric methodology and data set. Section 6 provides the results from the bivariate probit empirical analysis, including the robustness analysis, marginal response probability estimates, and predicted probabilities. Section 7 concludes.

The purpose of this study was to develop an econometric model that explains the “economic” determinants of BITs — at the same time as explaining PTAs. In the spirit of Baier and Bergstrand (2004), which explained PTAs in the context of a general equilibrium model of world trade with exporters, the model in this study is the first econometric model to explain BITs (along with PTAs) in the context of an explicit general equilibrium model of world production, consumption, trade, and FDI with national and multinational firms in multiple countries. The main conclusions are that the potential welfare gains from and likelihood of a BIT (PTA) between a country-pair are higher: (1) the larger and more similar in GDP are the country-pair; (2) the closer in distance are the two countries; (3) if the two countries are not adjacent (are adjacent) and do not share (do share) a common language; (4) the higher (lower) the degrees of political stability and of expropriation risk of the pair; and (5) the relatively more skilled labor abundant (the wider the relative K/U ratio of) the pair. These factors have economically and statistically significant effects on the probability of a BIT (PTA). While there exist choices of cutoff probabilities in determining the percent correctly predicted of the alternative outcomes, using the unconditional probabilities the preferred empirical model predicts correctly more than 80% of country-pairs with a BIT and PTA, with a BIT but no PTA, and with a PTA but no BIT. Consequently, the model provides a benchmark for incorporating other economic – and especially political science and legal variables – into understanding the determinants of BITs and PTAs.