Download Revisiting “Rational Design”: Preferences, Power, and the Design of

Survey
yes no Was this document useful for you?
   Thank you for your participation!

* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project

Document related concepts

Socially responsible investing wikipedia , lookup

Foreign direct investment in Iran wikipedia , lookup

Investment banking wikipedia , lookup

Environmental, social and corporate governance wikipedia , lookup

History of investment banking in the United States wikipedia , lookup

Investment management wikipedia , lookup

Investment fund wikipedia , lookup

International investment agreement wikipedia , lookup

Investor-state dispute settlement wikipedia , lookup

Transcript
Revisiting “Rational Design”:
Preferences, Power, and the Design of Bilateral Investment Treaties
Todd Allee
Department of Political Science
University of Illinois
[email protected]
Clint Peinhardt
School of Economic, Political and Policy Sciences
The University of Texas at Dallas
[email protected]
Paper prepared for presentation at the
Workshop on the Politics of Preferential Trade Agreements:
Princeton University, Niehaus Center for Globalization and Governance
April 29, 2010
Introduction
During the past three decades states have signed literally thousands of bilateral and regional
economic agreements with the explicit goal of stimulating greater cross-border trade and investment.
These agreements, which include preferential trade agreements (PTAs) and bilateral investment treaties
(BITs), among others, have received substantial attention from scholars across political science and
economics. Attracting the most attention are the central questions of why governments sign PTAs and
BITs, and what effect these treaties have on trade and investment, respectively. In providing answers to
these questions, existing empirical literatures quite logically treat these agreements as an either-or
proposition - states either have a BIT or a PTA or they do not. Yet there is increasing recognition that
these agreements are not homogenous but instead vary in important ways.
Indeed, trade and investment treaties are carefully negotiated by the signatories, and the resulting
PTAs and BITs range from dozens to hundreds of pages in length. They may include exclude particular
provisions, and language within a given provision often varies. Thus a BIT signed by the United States,
for example, may enshrine different substantive and procedural guarantees than one signed by, say, China
or Egypt and their treaty partners. Legal scholars, being particularly interested in the nuances of the
treaties, have explored differences in the language and provisions across selected treaties. Building on
this foundation, comparatively-minded political economy scholars are pushing the research frontier in a
more generalizable direction by collecting the texts of these economic agreements, coding important
treaty components, and producing quantitative and qualitative measures of treaty variation. On the
empirical front, increasingly rich data sources are beginning to emerge that capture variation within both
PTAs and BITs.
As data become richer and more refined, a new challenge now confronts scholars: how to explain
this important variation across the treaties? One could argue that so far data collection efforts are
outpacing attempts at theoretical development. Cumulative knowledge from the earlier wave of studies
can tell us why two governments would sign a BIT, for instance, but there is little guidance for
1
understanding why they would include some components in their investment treaty yet exclude others?
In other words, it remains unclear why governments design their BITs in the way they do.
Fortunately, the literature on the “rational design” of international institutions (Koremenos,
Lipson, and Snidal 2001; Koremenos 2005; 2007) provides a logical springboard from which to theorize
about variation across economic agreements such as BITs. Upon careful inspection, several of the
“design features” of international institutions for which rational design puts forward propositions are
indeed present in BITs and vary across the treaties. Therefore, the relevant propositions can be applied
fairly straightforwardly to the variation we observe in BITs. In turn, this newfound variation in BITs
provides an excellent domain in which to test quantitatively the conjectures put forward by the rational
design project – and to do so in a manner that addresses many of the criticisms made about the rational
design project’s empirical shortcomings.
In this paper we explore the design of the most important aspects of BITs – procedures for
dispute settlement – and do so using the “rational design” literature as a guide for theorizing. After
identifying important variation across BITs and compiling an original dataset that captures this variation,
we articulate several hypotheses that draw directly from rational design conjectures. At the same time,
we also address criticisms that the rational design framework fails to account for the power and
preferences of the actors who design the agreements.
Quantitative tests of the design of 1,500 unique bilateral investment treaties reveal relatively
modest support for the rational-design inspired conjectures, yet much is gained by applying the rational
design logic to BITs. First, drawing upon the rational design notion of “design features” helps us to
identify the most important aspects of the treaties and to structure our theorizing and data collection more
tightly around the concepts of centralization and flexibility. Second, several rational design conjectures
do possess considerable explanatory power, thereby helping us to understand that governments include
centralized enforcement procedures in treaties when there are strong incentives to cheat and when one
side lacks information about the other BIT signatories. Somewhat ironically, a third contribution emerges
from our efforts to address earlier criticisms of the rational design framework. The bargaining power and
2
preferences of states who sign BITs – which critics of rational design claim are missing from the project’s
framework – turn out to be the strongest and most consistent predictors of BIT design. Treaty language in
BITs ultimately reflects the preferences of the governments who sign them, the interests of important
domestic actors, and the bargaining power of the larger side. All in all, the endeavor to understand BITs
as the result of purposeful design advances scholarship on the rational design of international institutions,
and most importantly, provides a compelling and largely power- and preference-based explanation for
why investment treaties vary in the manner they do.
The Importance and Diversity of BITs
Bilateral investment treaties represent a unique effort to regulate foreign direct investment (FDI)
between pairs of states and their nationals. As such, they attempt to ameliorate the traditional antagonism
between investors and host states that is reflected in developments such as the Hull Rule and the Calvo
Doctrine (see Guzman 1998). For centuries governments have quarreled over issues such as when a
country that hosts foreign investment can legitimately seize (expropriate) that investment, what
constitutes a legitimate versus illegitimate taking, and what compensation is due to the foreign investor
(whether individual or firm) upon expropriation? These and other issues are directly addressed in the
more than 2,500 BITs that pairs of states have signed during the past half-century. In general, all BITs
grant a series of substantive as well as procedural rights to foreign investors (see Muchlinski 2009), with
the explicitly-stated purpose of increasing foreign investment between the signatories.
Given the historical antagonism over foreign investment, it is not surprising that treaty clauses
governing the settlement of investment disputes are widely seen as the most important aspect of BITs. In
fact, treaty provisions regarding investor-state dispute settlement have attracted considerable attention
from legal scholars (e.g., Dolzer and Stevens 1985; Franck 2007a, 2007b; Ginsburg 2005; Sornarajah
2000; Vandevelde 1992; Yackee 2008). Susan Franck (2007b), for one, claims the inclusion of investorstate dispute settlement clauses represents a “sea change” in the legalization of foreign investment. For
the first time, foreign investors can use international law to directly challenge the legality of host state
3
actions through a neutral arbitration panel, and can receive compensation when their complaint is found to
have merit (Paulsson 1995). Because of these investor-state dispute settlement clauses in BITs, Brigitte
Stern, a frequent arbitrator in investment cases, writes that “we are walking with giant steps towards a
general system of compulsory arbitration against States for all matters relating to international
investments, at the initiative of the private actors of international economic relations” (quoted in Williams
2003, 252). The emergence of BITs – with their revolutionary procedures for dispute settlement – is thus
a historical development in international economic relations.
Not surprisingly, scholars have begun to investigate important questions such as why countries
sign BITs (e.g., Elkins, Guzman, and Simmons 2006; Guzman 1998) and what impact BITs have on FDI
(e.g., Egger and Merlo 2007; Hallward-Driemeier 2003; Haftel forthcoming; Kerner 2008; Neumayer and
Spess 2005; Rose-Ackerman and Tobin 2005; Salacuse and Sullivan 2005). Understandably, studies in
this initial wave of research do not distinguish among the treaties and simply record whether or not a pair
of states has a BIT. Yet many of these same authors question this simplifying assumption of BIT
homogeneity, and readily acknowledge that BITs are likely to vary on important dimensions. Neumayer
and Spess (2005, 1571), for example, note that “not all BITs are identical in their provisions” and that
“BITs are also unlikely to be identical in their effect on incoming FDI flows.” Hallward-Dreimeier
(2003, 3), in one of the first studies of BITs, also notes that “(t)oo much attention has been placed on
whether or not a BIT exists rather than on the strength of property rights actually being enshrined in these
agreements.” The pervasive argument is that BITs should increase FDI among signatories because they
represent a “credible commitment” by the parties to respect the terms of foreign investment. This
credibility, however, depends heavily on the ability of aggrieved foreign investors to successfully enforce
the terms of the BIT, which is why the investor-state dispute settlement provisions are so central.
4
Interestingly, a careful examination of the text of BITs reveals that the aforementioned dispute
settlement provisions differ considerably across treaties, in ways that are not always recognized.1
Important treaty variation exists in three areas. First, in some BITs states provide advanced consent to
international arbitration of future disputes whereas in others they do not. The jurisdiction of most
international dispute resolution bodies depends critically on whether states have submitted to their
authority (Guzman 2002), which in practice means that they need a state’s written authorization before
they will even hear a dispute. Although many BITs require this consent be given once a dispute arises,
some treaties actually include both state parties’ advance consent for arbitration to proceed whenever a
dispute arises. This often works to the advantage of aggrieved investors because it speeds up the dispute
resolution process and removes delay-prone arguments over jurisdiction while at the same time reducing
the accused state’s options. Second, BITs can specify one or more among several venues for settling
investment disputes. In addition to domestic courts in the host state, more than a dozen international
arbitration venues are sometimes specified as dispute settlement options. These venues may range from
ad hoc arbitration using UNCITRAL rules to more regimented arbitration through standing bodies such as
the International Centre for the Settlement of Investment Disputes (ICSID), the Permanent Court of
Arbitration, or one of several regional arbitration centers. All else equal, the more venues that are
specified, the more options that parties to a BIT have for addressing their disputes. Third, some of the
aforementioned arbitration venues are stronger and more institutionalized than others. ICSID, for
instance, is a permanent institution that has state delegates, a permanent staff, and the ability to enforce
the awards it renders. Other venues are far less institutionalized, and instead rely heavily on the states to
formulate rules and to enforce awards.
Moreover, these three ways in which BITs vary correspond directly to two of the “design”
features for which the rational design literature provides a set of explanations. Those BITs that require
consent to arbitration on a case by case basis, and that list several venues as dispute settlement options,
1
Recent scholarship in legal journals increasingly acknowledges that these clauses vary across BITs (Franck 2007a,
2007b, Swenson 2005, Yackee 2008), yet other, large-n scholarship on BITs from political science and economics
does not distinguish among treaties.
5
provide greater flexibility to parties. The rational design literature, in turn, has put forward several
conjectures to explain why states would want to design “flexible” international treaties. Similarly,
arbitration institutions like the aforementioned ICSID are heavily centralized: considerable dispute
settlement authority is located in a central, permanent actor that can determine how states resolve their
disputes. Likewise, a substantial number of rational design conjectures attempt to explain such
institutional centralization. Because of this promising conceptual match, it is sensible to turn to the
rational design literature as a basis for theorizing about BIT design.
The Rational Design Project and Its Critics
In an ambitious 2001 International Organization issue on “The Rational Design of International
Institutions,” the editors of the volume, along with several contributors, set out “to offer a systematic
account of the wide range of design features that characterize international institutions” (Koremenos,
Lipson, and Snidal 2001a: 762). This “rational design” issue is tightly anchored around a comprehensive
analytical framework in which the authors propose five dependent variables, or dimensions of variation in
international institutions, among which are the dimensions of centralization and flexibility. To these five
institutional design features they map six independent variables, or “cooperation problems” thought to
influence the manner in which international institutions are designed, resulting in a total of sixteen causeand-effect conjectures intended to explain the rational design of international institutions. This general
theoretical framework is then applied to eight case studies of international institutions.2 On the whole, the
case studies are generally supportive of the framework, and Koremenos, Lipson, and Snidal (2001b:
1056) point out that “nearly 70 percent of the findings are strongly positive.”3 Several conjectures
regarding variation in institutional centralization and institutional flexibility are the most supported, but
the findings in support of rational design are far from universal or definitive. Of the sixteen conjectures,
2
Each case study explores only a subset of the rational design propositions. Kydd’s chapter on NATO expansion
examines only one proposition, while at the other extreme the articles by Richards (on the International Air
Transport Authority) and Morrow (on prisoner of war treaties) examine nine propositions each.
3
See Table 1 of the concluding chapter of the Rational Design issue (Koremenos, Lipson, and Snidal 2001b: 1055)
for a summary of the case study findings.
6
one is not tested in any of the case studies, three are not supported, and four are supported by only one
case study.
Since its inception, the rational design framework has encountered skeptics, with considerable
criticism focused on problems of case selection, quantification of concepts, and the lack of large-n
empirical testing. Both Wendt (2001) and Duffield (2003) are skeptical of the initial case study evidence,
noting that a majority of the studies focus on topics where one might expect the conjectures to apply—a
criticism largely accepted by Koremenos and Snidal (2003: 439). Duffield (2003: 424) notes another
problem with the choice of case studies, namely that “few of the cases fit neatly with the project’s stated
focus on the rational design of explicit, negotiated arrangements.” He then asserts that future tests
“…should be tightly restricted to cases that are indisputably at the heart of the project’s ambit, namely the
design of explicit, negotiated agreements” (ibid., 428).
Not surprisingly, Duffield (2003) calls explicitly for quantitative tests of the rational design
conjectures as a way to address concerns with empirical evaluation. He challenges the consistency of
measurement in the case studies and argues that quantifying the abstract concepts is the best way to move
the project forward. Koremenos and Snidal (2003: 437) in fact acknowledge that “Duffield’s most
valuable critique regards the empirical shortcomings of the Rational Design project” and echo his call for
large-n empirical tests of the theory. Recent work by Koremenos (2005, 2007) advances this goal, but
represents a less systematic test of the rational design conjectures than our endeavor.4
Just as the rational design literature provides a promising platform on which to theorize about
BIT variation, the rich variation we identify in the 1,500 investment treaties we have coded provides an
excellent empirical domain in which these ideas and criticisms can be put to the test. One somewhat
obvious benefit is that BITs are formal treaties, and as such fit Duffield’s criterion for conducting
empirical tests using explicit, negotiated agreements. Second, focusing on BITs holds constant many
4
Neither of the more recent Koremenos pieces (2005, 2007) tests the originally-stated rational design conjectures
directly. Those concepts which reappear from the 2001 volume are generally treated at a higher level of conceptual
aggregation than is found in the original volume. Although the important new ideas (renegotiation costs, risk
aversion, delegation) advanced in these two articles certainly build on and are consistent with the original rational
design project, the well-known conjectures from the 2001 volume are not tested explicitly.
7
potentially conflating factors that otherwise would complicate empirical evaluation. All of the treaties are
negotiated by a pair of states, which removes the need to control for the number of signatories or other
conflating dynamics. Furthermore, the treaties all attempt to regulate and protect FDI and enforce
standards of investor protection, and are similar on nearly all dimensions other than centralization and
flexibility of institutional design. Third, the substantive problems that plague foreign direct investment –
such as credibility of commitment, unexpected shocks, and incentives to violate the agreement – conform
closely to several of the “cooperation problems” identified in the rational design literature. Fourth, the
states that have signed BITs are diverse. More than 175 countries have signed BITs over a period
spanning many decades. As a result, our sample includes both cross-sectional variation in terms of the
domestic political institutions and economic conditions of the countries in the sample, and temporal
variation regarding the use of arbitration, patterns of expropriation, and changes in FDI flows.
Using “Rational Design” to Theorize about BIT Variation
Because the rational design approach appears to be applicable to investment treaty design, it
serves as the basis for theoretical development. Along the way, important criticisms of the approach, as
well as the alternate arguments put forward by critics, also serve as a further platform for theorizing.
According to the overall logic of rational design, states will deliberately include or exclude certain dispute
settlement provisions within a particular BIT, and thus variation across treaties should be purposeful as
opposed to random. Different pairs of governments will intentionally tailor their BITs to address the
particular problems they face.
To understand the possible treaty-design solutions to these problems, we first draw upon the
rational design literature to identify the investment-related cooperation problems that are relevant to BIT
signatories. Next we match these cooperation problems to four rational design propositions that are
relevant to BITs and FDI – three from the original 2001 volume (Koremenos, Lipson and Snidal 2001a)
and one from more recent scholarship on rational design (Koremenos 2007). These four propositions are
8
discussed in the investment treaty context, and they result in four BIT-design hypotheses that are drawn
from the rational design conjectures.
The four rational design propositions that serve as the theoretical foundation are: 5
1) FLEXIBILITY increases with UNCERTAINTY ABOUT THE STATE OF THE WORLD
(F1).
2) CENTRALIZATION increases with UNCERTAINTY ABOUT THE STATE OF THE
WORLD (C2).
3) CENTRALIZATION increases with ENFORCEMENT PROBLEMS (C4).
4) CENTRALIZATION increases with COMMITMENT PROBLEMS (CT).
These four conjectures are identified as the most relevant for BIT design after carefully
considering all of the conjectures related to centralization and flexibility, the two dimensions on which the
BITs in our dataset vary.6 The first three conjectures (F1, C2, C4), which are taken straight from the 2001
International Organization volume and reappear in Koremenos 2007, are among the conjectures that
received the greatest support in the original case studies.7 The fourth conjecture (which we label “CT”) is
an important new rational design proposition that highlights commitment problems – a cooperation
problem that is endemic to investment cooperation as well as many other international issues. Although
the original IO volume did not focus on commitment problems directly, the recent identification of
commitment problems by Koremenos (2007) reflects an attempt to expand the rational design framework.
In this regard, our coupling of a new rational design conjecture with three original ones in our empirical
5
We state the conjectures in this manner for both simplicity and consistency. The first three conjectures are phrased
in the way they are stated in the original “Rational Design” volume (see Koremenos, Lipson, and Snidal (2001a:
788-789, 793). For the fourth conjecture, we re-organize the proposition to be consistent with the three other
conjectures. Koremenos (2007: 194) adds “states facing” before “commitment problems” and then replaces
“centralization” with the more precise language of “delegate(d) dispute settlement provisions.” Although we agree
with Koremenos’s focus on delegation of dispute settlement, we retain the language of “centralization” given our
singular goal of testing the rational design conjectures in a generalizable manner.
6
Four additional conjectures about centralization and flexibility (out of seven total) that appear in the original
International Organization volume are easily omitted. Three of the four omitted conjectures (one on centralization
and two on flexibility) are irrelevant to BITs because they relate to design influences that are held constant in our
design (e.g., the number of actors). A final conjecture from the original edited volume, which relates uncertainty
about behavior to centralization, is deemed inapplicable because such uncertainty is not widespread in BITs due to
the presence of multinational firms who monitor the behavior of foreign governments toward FDI.
7
Proposition F1, in fact, was the most supported conjecture among the original case studies (Koremenos, Lipson,
and Snidal 2001b: 1055).
9
tests enhances both the comprehensiveness and validity of our effort to explain BIT design using the
rational design framework.
Flexibility Increases with Uncertainty about the State of the World
One of the first cooperation problems awaiting states who attempt to establish common
investment rules is uncertainty about the state of the world. Koremenos, Lipson, and Snidal conceptualize
this type of uncertainty as: “…states’ knowledge about the consequence of their own actions, the actions
of other states, or the actions of international institutions” (2001a: 778). This uncertainty about the
implications of an international agreement such as a BIT will vary over time as states gather more
information about the consequences of the agreement. Uncertainty about the state of the world, however,
does not automatically decline over the lifetime of a treaty or regime, since the presence of unexpected
developments or “shocks” can alter the landscape within an issue area (Koremenos 2001, 2005, Mattli
2001, Oatley 2001). Such shocks in the world of FDI – such as dramatic fluctuations in investment
patterns, currency crises, or rapid increases (decreases) in the number of investment treaties or disputes –
could generate more or less certainty about how BITs work, who they benefit most, or how they will be
enforced.
Greater uncertainty should lead BIT signatories to incorporate greater flexibility in their treaties,
ceteris paribus. Since other important features of the treaties are relatively fixed, governments should
seek this flexibility in specifying procedures for dispute settlement.8 The manner in which future disputes
are settled can be highly consequential, since different processes and venues can generate different
outcomes for the parties (Brower 2005). When uncertainty about the state of the world is high,
governments will prefer to have several possible options for dispute settlement and to allow for the choice
8
Treaty renegotiation has been used as an indicator of flexibility (Koremenos 2001, 2005; but see Copelovitch
2008). Yet most BITs are in force for a minimum period of 10 years, which renders this type of flexibility less
relevant.
10
of venue to be made on a case-by-case basis.9 In contrast, when the parties possess greater knowledge
about the implications of their investment agreement, they will be more comfortable including a more
limited range of dispute settlement options in the treaty and granting advanced consent for international
arbitration.10
HF1: BITs will contain more flexible procedures for dispute settlement when uncertainty about
the current state of global investment is high.
Centralization Increases with Uncertainty about the State of the World
The second rational design conjecture predicts that this same uncertainty about the state of the
world also should lead to greater centralization of investment treaties. The problem once again is that
governments who sign BITs may do so in an environment where there is relatively little information
about the myriad consequences of the agreement they are about to sign. In such situations of incomplete
contracting, governments may desire to have one or more centralized international institutions provide
additional information about the state of the world and to clarify ambiguities in the treaties they have
signed (Abbott and Snidal 2000, Garrett and Weingast 1993). Because of the bilateral nature of BITs,
governments facing uncertainty may see considerable benefit in centralizing the crucial function of
dispute settlement as a way to enhance the efficacy of the agreement. Centralization also serves a
valuable pooling function (Abbott and Snidal 1998) by providing information to the signatories, as well
as to other BITs signatories (in the case of transparent centralized bodies such as ICSID). In sum,
delegating dispute settlement authority to centralized international institutions (Abbott, et. al. 2000,
Koremenos 2007) can help governments to obtain “more certainty about how the agreement operates and
a better understanding of its costs and benefits” (Koremenos, Lipson, and Snidal 2001a: 794).
HC2: BITs will contain more centralized procedures for dispute settlement when uncertainty
about the current state of global investment is high.
9
A mid-1990s UNCTAD (1998) study concluded that governments were seeking flexibility in dispute settlement
due to concerns with the enforceability of ICSID awards and the success of UNCITRAL rules in U.S.-Iran tribunals.
10
A more recent UNCTAD (2007b) study, in fact, concluded that governments were including narrower investorstate dispute settlement options in treaties due to information gained through recent ICSID rulings.
11
Centralization Increases with Enforcement Problems
A second cooperation problem that BIT signatories face, one which is common to many types of
treaties, is the concern with whether the treaty will be enforced. Within the rational design literature
enforcement problems are equated with “incentives to cheat” on an agreement.11 Governments sign BITs
knowing that the treaties promise long-term gains, yet certain leaders might have short-term incentives to
seize a foreign asset or to change the terms of investment. This is particularly true for leaders who face
strong economic or political incentives to engage in expropriation or who fear little direct or indirect
retribution for cheating. To address this type of enforcement problem, the rational design logic posits that
one should observe greater centralization – in this case centralization of enforcement – to counter these
incentives. Third-party dispute resolution through established international institutions is a way to deter
cheating by opportunistic governments and to encourage compliance with international agreements (e.g.,
Mitchell and Hensel 2007, Simmons 2000).12 Centralizing dispute settlement in the hands of an
established and institutionalized third-party raises the cost of defection.13 It moves investment dispute
settlement outside of the potentially biased and slow-moving institutions of the defecting state and instead
shifts it to a neutral venue with established rules and timelines (Franck 2007b, 2007c). Knowing this,
potential cheaters might be deterred from defection by the prospect of being challenged before a
centralized enforcement body.
HC4: BITs will contain more centralized procedures for dispute settlement when the temptation to
cheat on the agreement is high.
11
In the original Rational Design volume, Koremenos, Lipson and Snidal define enforcement problems as “the
strength of individual’s actors’ incentive to cheat on a given agreement or set of rules” (2001a: 776).
12
The European Court of Justice (e.g., Alter 1998, Burley and Mattli 1993) and the World Trade Organization’s
(WTO) dispute settlement mechanism (e.g., Rosendorff 2005) have played a central role in enforcing agreements
signed by members of the European Union and WTO, respectively.
13
Abbott and Snidal predict that “hard” legal commitments (such as legalized dispute settlement through an
international organization) should be common in investment agreements, where the incentives for opportunism may
be great (2000: 429).
12
Commitment Problems Lead to Centralization
Bargaining over the terms of FDI is characterized by an endemic commitment problem – the
time-inconsistency problem associated with the obsolescing bargain (Vernon 1971). Before any initial
investment is made, multinational firms possess considerable bargaining leverage vis-à-vis host states, yet
after the initial investment bargaining power shifts to the host government due to the costs sunk by the
firm. This dynamic creates a commitment problem for states that seek to attract foreign investment, since
the anticipated ex post shift in bargaining power makes potential investors leery of investing in states who
cannot credibly commit to uphold the original bargain. For “host” states who seek to attract FDI, then, it
is essential to demonstrate ex ante the credibility of one’s commitment to respect the terms under which
inward investment is made.
To some degree all BITs attempt to address this commitment problem (e.g., Elkins, Guzman and
Simmons 2006, Guzman 1998, Swenson 2005), but the intensity of the commitment problem varies
across treaties. It is particularly acute for those governments who have a poor record of past behavior,
questionable domestic institutions, or about whom little is known. Centralization provides a way for
these governments to ameliorate the most severe commitment problems (Abbott and Snidal 1998). When
a BIT contains language that allows future host-government actions to be addressed directly by investors
through an institutionalized arbitration body, investors become more confident that any ex post
“takings”14 could be dealt with effectively. This ability to turn to centralized dispute settlement increases
investors’ ex ante confidence in the terms of the initial investment. The centralization of dispute
settlement in the hands of bodies like ICSID and the ICC also serves the interests of credibility-seeking
host governments, who can credibly signal that their hands and those of their successors are effectively
tied (e.g., Abbott et. al. 2000, Ginsburg 2005, Koremenos 2007, Yackee 2008). In contrast to
enforcement problems, where governments actually experience rational, ex post temptations to defect on
an agreement, with commitment problems the problem is largely perceptual. Governments who fully
14
“Taking” is the legal term that incorporates both outright expropriations as well as less blatant violations of
investment treaty commitments (Dolzer and Stevens 1995, 98-102; Reisman and Sloane 2004).
13
intend to respect inward investment must still find a way to demonstrate this genuine commitment to
governments with whom they sign BITs as well as multinational firms within those partner countries.
HC4: BITs will contain more centralized procedures for dispute settlement when the need for
credible commitment is greatest.
Criticisms of Rational Design: Bringing in Power and Preferences
Separate from any of the conjectures discussed above, a major criticism of the rational design
project – which we address directly – is its failure to employ control variables and to account for
alternative explanations such as interstate power relations and state interests. Wendt (2001: 1035)
suggests that powerful states embed their practices into the design of regimes, and Duffield (2003: 417418) claims that important power-based and capability-based explanatory variables are absent from the
Rational Design framework. Duffield also criticizes the project for failing to account for the interests of
the actors; that is, the preferences states have over the design of international institutions (2003: 417).
Lake (2009) echoes this concern, claiming that in institutional design approaches “interests are often
treated in an arbitrary or inductive manner and produce, at best, propositions that are hard to falsify”
(240). He then suggests that “rather than building ever more sophisticated models of institutions per se,
scholars of security, human rights, and other important substantive topics might be better served by
developing theories of and focusing attention on the interests of actors” (240). The final three hypotheses,
then, directly address these concerns by emphasizing domestic preferences, government preferences, and
the signatories’ relative bargaining power.
The first of two preference-based variables captures the potential importance of domestic political
interests in investment treaty design. The most relevant domestic actors in this case are multinational
corporations, who have a strong preference for BITs that include the most effective procedures for
challenging foreign government takings. This means they desire more centralized and less flexible BITs.
They want their governments to include greater centralization, since more institutionalized arbitration
procedures are more likely to return swifter, fairer, and more easily-enforced rulings against foreign
violators. Similarly, multinational corporations also prefer to have less flexible BITs, ceteris paribus. In
14
practice, greater flexibility in dispute settlement, in the form of more arbitration options or the need to
provide consent, delays any efforts at dispute settlement. This works against multinational investors and
to the advantage of foreign governments, who can use the flexibility to delay arbitration while debates
about jurisdiction and rules play out. Governments who negotiate BITs should be attentive to the
preferences of these important domestic economic actors.
HdomprefC: BITs will contain more centralized procedures for dispute settlement when a
significant number of large multinational corporations are located in the signatory countries.
HdomprefF: BITs will contain less flexible procedures for dispute settlement when a significant
number of large multinational corporations are located in the signatory countries.
A second preference-based argument emphasizes the preferences of the signatory governments
themselves. Governments of a right-wing, or more free-market, orientation, have the strongest desire to
design BITs that create the most favorable conditions for continued financial openness and the greatest
investment flows. Once again, this is more likely to occur when dispute settlement procedures are
centralized and inflexible. In such situations, treaty violations are most likely to be effectively
challenged, thus providing swift remedies for illegal takings and deterring others.
HgovtprefC: BITs between right-wing governments are more likely to contain centralized
procedures for dispute settlement.
HgovtprefF: BITs between right-wing governments should contain less flexible procedures for
dispute settlement.
Finally, the relative power of the two BIT signatories is included as a predictor of BIT design in
order to incorporate interstate power relations as an explanation for BIT design. One prediction is that
powerful states should prefer, and thus be more likely to obtain, centralized dispute settlement in their
BITs. This seemingly contradicts the traditional realist view (Mearsheimer 1994/1995) and other
accounts of powerful-state attitudes toward legal dispute settlement (Smith 2000). But in the asymmetric
FDI context, centralized dispute settlement serves the interests of powerful states and their investors
because of its greater efficacy in punishing treaty violations. Consequently, it is strongly preferred to ad
15
hoc arbitration or the use of domestic courts in the host state (e.g., Franck 2007b, Schreuer 2005).15
States with a significant bargaining advantage over the other signatory also prefer to have less flexible
BITs, ceteris paribus. Greater flexibility in dispute settlement, in the form of consent requirements and a
wider array of arbitration venue choices, delays any efforts at dispute settlement. This works against the
interests of powerful states and their investors, since it stalls any attempts to address treaty violations
committed by the weaker state due to debates over jurisdiction and procedure.
HpowerC: BITs will contain more centralized procedures for dispute settlement when one of the
signatories is considerably more powerful than the other.
HpowerF: BITs will contain less flexible procedures for dispute settlement when one of the
signatories is considerably more powerful than the other.
Measurement of Concepts
Proper measurement of independent and dependent variables is paramount given the criticism
rational design has attracted for its failure to properly operationalize concepts and to incorporate them in
large-n tests. Duffield (2003), for instance, claims that the original Rational Design volume offers little
guidance for translating abstract concepts like “uncertainty” or “centralization” into precisely measured
indicators. To properly measure concepts like those identified in the hypotheses articulated earlier, we
draw heavily upon our substantive knowledge of BITs as well as related empirical literatures on
legalization, investment treaties, foreign direct investment, and international political economy more
broadly. In most cases we identify and provide justification for a primary operational measure, yet also
employ alternate indicators to check for the robustness of findings.
Among the independent variable concepts, the nature of the enforcement and commitment
problems inherent to FDI compels us to focus heavily on the domestic political conditions within each
BIT signatory, which adds to our data collection burden. These domestic as well as state-level variables
present another challenge: how to combine the parallel country-level measures for the two BIT
15
Similar dynamics were evident in Uruguay Round negotiations to establish the dispute settlement mechanism of
the WTO. The United States and other powerful states preferred a strong, legalized mechanism because it advanced
their interests (Croome 1998).
16
signatories into a single measure? Enforcement or commitment problems may stem from conditions
relevant to either or both states, yet rational design is not clear on how to translate such situations into
empirical tests. Our primary approach is to follow Koremenos (2005) and employ the “weak-link”
approach common to quantitative studies of international conflict, in which the most problematic (highest
or lowest, as dictated by theory) of the two states’ values is selected (e.g., O’Neal and Russett 1997).
Uncertainty about the State of the World
Uncertainty about the state of the world as it relates to BITs and FDI is reflected in two
indicators: one that captures the amount of information states have about BITs and the implications of
BITs, and a second that reflects “shocks” to the prevailing status quo. Because both of these measures fit
the rational design logic but are not correlated, each is included separately and then jointly in regressions
for centralization and flexibility, respectively.
According to the first logic, over time increasing or decreasing amounts of new information,
revealed by waves of treaty-making and actions by governments who have signed BITs, should affect the
degree of certainty governments possess at the time of BIT signing. As the primary way to capture this
uncertainty logic, we include in the regressions a count of the worldwide total number of BITs signed in
the past five years (Number of Recent BITs).16 This indicator provides important information about
recent trends and the resulting level of “certainty” within the FDI regime. A tally of the number of
investment disputes submitted for international arbitration in the past five years (Number of Recent
Investment Disputes) captures a similar logic, and thus is included as a sensitivity check.
The second measure reflects the strand of the rational design literature that equates uncertainty
about the state of the world with “shocks” (Koremenos 2005, 2007). The idea here is that unexpected
events that impact FDI will lead to greater uncertainty among BIT signatories about the current state of
the world in FDI – and thus greater centralization and flexibility in BITs. A tally of global balance of
16
Data on the number of BITs is taken from UNCTAD Investment Instruments Online archive, the primary entity
that systematically collects and publishes the text of BITs, and supplemented with data from ICSID and the authors’
additional research.
17
payments crises during the past three years captures shocks directly relevant to BITs and FDI that will
reduce certainty among BIT signatories. This Recent FDI-Related Shocks (3 years) variable is compiled
from data in Kaminsky and Reinhart (1999) and Kaminsky (2006), and lags of one and five years also are
employed as sensitivity checks. Also employed as a sensitivity check is a measure of Variance in Recent
Global FDI Flows, which captures the statistical variance in worldwide FDI flows over the past three or
five years. The logic here is that inconsistent patterns of recent FDI reflect the total collection of recent
global shocks that make BIT signatories less certain.
Enforcement Problems
Drawing upon the earlier discussion of enforcement problems as “incentives to cheat,” as many
as three distinct variables capture the severity of the enforcement problem for any given BIT. Each
variable is inserted separately, and then with the others, in a series of regressions intended to predict BIT
centralization as a function of the enforcement problem the states face.
The first variable captures each BIT signatory’s “demand” to cheat on the agreement. Political
leaders will have a greater need to engage in some type of expropriation – and thus will be more tempted
to cheat – when they face poor domestic economic conditions. Given short time horizons, leaders who
face immediate economic problems may decide that it is rational to forego the longer-term benefits
associated with increased FDI and instead to violate investor protections as a way to address short-term
economic problems.17 GDP Growth Rate, measured as the amount of GDP growth (decline) in the
country within the past year, captures the degree to which economic conditions in a state are poor and
thus where short-term incentives to cheat present a greater enforcement problem.18 Low growth rates
increase incentives to deviate from BITs and exacerbate the enforcement problem. Several other
indicators reflect the same logic and are employed as sensitivity checks: under high current rates of
unemployment (Unemployment Rate) and inflation (Inflation Rate), leaders become more willing to seize
17
We measure the economic conditions in the signatories at the time of BIT signing, since conditions at that time
represent the two sides’ best guess as to future economic conditions and potential near-term enforcement problems.
18
All indicators in this group are taken from the World Bank’s World Development Indicators.
18
foreign assets and thus present a larger enforcement problem. Alternate indicators for recent trends in
each of three aforementioned rates also are considered as a way to capture the trajectory of these
economic patterns over the recent past.19
A second indicator captures the idea that enforcement problems will be more severe when the
benefits of cheating are greater. Namely, the temptation to cheat will be heightened when global Raw
Materials Prices are higher. In such situations, the value of what might be seized increases, ceteris
paribus, and this in turn intensifies the enforcement problem. We compile data on annual prices (in
constant dollars) of an index of minerals, ores, and metals, taken from the UNCTAD’s Commodity Prices
Database (CPD).
A third enforcement problem variable reflects the sensitivity of the two sides to any costs that
might arise from cheating on the agreement. Signatories with close economic or political ties should have
fewer incentives to defect due to the greater costs of such a move. Most notably, the enforcement
problem will be lessened if the two sides are heavily trade dependent upon one another, since such ties
could be jeopardized in the event of a breakdown in trading relations due to one side cheating on the
terms of the BIT. An indicator of the size of the Bilateral Trade Relationship, taken from the Correlates
of War’s Bilateral Trade data set (v. 2.0), is employed as the primary measure. As a robustness check a
measure of Alliance Ties between the signatories also is employed to reflect reduced incentives to cheat.20
Commitment Problems
Commitment problems within a BIT will be more severe when at least one of the signatories: 1)
lacks credibility-inducing domestic institutions, 2) has no experience hosting investment (since its
commitments cannot be inferred), and 3) has a poor reputation for honoring its past commitments. Once
19
These variables are created by taking the current rate of each economic indicator (growth, unemployment, and
inflation) and subtracting from it the average of the respective rates during the past three years.
20
Data on alliance ties come from the Alliance Treaty Obligations and Provisions (ATOP) Data Set (Leeds et. al.
2002). Our indicator captures whether the home and host states possess any type of alliance tie.
19
again, three respective operational indicators are included in a series of centralization regressions, first
one-at-a-time and then all together.
The first variable showcases the importance of domestic political institutions as commitment
devices. The overarching idea is that certain domestic institutions – those which constrain adventurous
behavior or emphasize the rule of law – can enhance the credibility of a state’s commitment to respect the
terms of the BIT and protect FDI. An important factor is whether political executives are relatively free
to pursue any course of action they choose, such as engaging in some form of expropriation, or whether
their power is limited by other domestic political institutions. The presence of legislative or other actors
who constrain political executives (Henisz 2002) can increase the credibility of a state’s commitment to
respect FDI by reducing fears that an executive will change course and engage in some type of
expropriation. As a result, the political constraints (Political Constraints on Executive) variable from
Henisz (2006) is included as a primary measure. Also substituted for purposes of robustness is the
closely-related Checks on Executive variable from the Database of Political Institutions (Keefer and
Stasavage 2003). Similarly, domestic legal institutions also can reduce the commitment problems
inherent to a BIT. Powerful legal institutions within states that receive FDI can restrain adventurous
leaders from engaging in unexpected expropriations. Therefore, additional measures of Respect for Law
and Order and Lack of Corruption, both taken from the International Country Risk Group (PRS 2007),
also are considered.
A second variable captures the fact that commitment problems will be more severe when little is
known about how the current government will treat inward investment. The lack of information could
reflect a range of political changes, any of which might make foreign actors hesitant to cooperate with
governments who have no proven track record. The Tenure of the Ruling Party, taken from the
Database of Political Institutions (Beck et. al. 2001), captures the length of time the current political party
has been in power, and serves as the primary measure. Among alternate indicators considered is the
Tenure of Executive (the number of consecutive years the current executive has been in power) and
20
Regime Tenure (the number of years the country has been democratic, autocratic, etc.), both of which also
come from Beck et.al. 2001.
A third indicator of commitment problems reflects a situation in which information is known
about a signatory, but that information reflects negatively and creates a larger commitment problem. In
the context of FDI, the commitment problem is exacerbated when one of the BIT signatories has
expropriated assets in the past or violated the terms of similar investment treaties. This type of negative
past reputation for respecting FDI and upholding the terms of BIT should lead to centralized dispute
settlement provisions in the current BIT. Each state’s Recent History of Expropriation, measured over
the past five years, serves as the operational indicator of this logic. Data on expropriations come from
Kobrin (1987) and Minor (1994, and personal correspondence).
Power and Preferences
Creating measures for the alternate arguments about domestic preferences, government
preferences, and relative bargaining power also is an important task that requires some diligence and
creativity. Multinational corporations are the primary domestic interests relevant to BITs, and they
should prefer greater centralization and less flexibility in BITs, as discussed earlier. To assess this
argument about potential influence of these actors on the composition of the BIT, we include in all
regressions an original indicator that captures the percentage of the world’s largest multinational
corporations that are headquartered in the two treaty signatories.21 Similarly, we also include a measure
of whether right-wing governments are in power in the BIT signatories. Such governments are likely to
prefer centralized yet inflexible BITs as a means to advance the cause of greater FDI. Data on whether
right-wing governments are in power in BIT signatories is taken from the World Bank’s Database of
21
Data for this measure are taken from Forbes magazine’s annual list of the world’s largest MNCs. To standardize
the Forbes lists across years (1980-2002), we count all MNCs that have revenues above a constant threshold
(approximately $5 billion in 1980 US dollars). We then identify the percentage of those MNCs in each year that are
from each country in our data set. For 2003 we use the Forbes Global 100 and for 2004-2006 we use the Forbes
Global 200 list. For years before 1980, we use the values from 1980, although the inclusion or exclusion of the pre1980 cases does not affect our substantive conclusions.
21
Political Institutions (Beck et.al. 2001). The final indicator is a count of the number of right-wing
governments between the two signatories (0, 1, or 2).22 Finally, a measure is created for the power-based
argument that emphasizes the relative bargaining power of the two BIT signatories. Our primary
indicator captures the relative economic power of the two states, and is created by taking the GDP of the
lower-GDP state and subtracting it from the GDP of the wealthier state.
Dependent Variables: BIT Centralization and BIT Flexibility
The measurement of the two BIT design features – centralization and flexibility – is somewhat
more straightforward. Considerable guidance is provided by a rich legal literature on investment treaty
arbitration, on which we rely heavily. Data collection represents the greatest challenge. We acquired the
text of nearly 1,500 investment treaties and carefully code each according to the following rules for
determining centralization and flexibility of dispute settlement provisions.
The idea of BIT “centralization” reflects the degree to which the investment treaty includes
procedures for dispute settlement that rely upon established, third-party institutions to enforce the terms
of the BIT. This notion of centralization—centralization of enforcement—is one of the most common
ways to conceptualize centralization within an international institution (Koremenos, Lipson, and Snidal
2001a). Centralization in this context also is consistent with the narrower concept of “delegation” of
dispute settlement authority in the legalization literature (Abbott, et.al. 2000, Goldstein, et.al. 2000).
Our primary measure of BIT centralization captures the degree to which investor-state dispute
settlement provisions in the treaty are institutionalized in the hands of a third party. We code a variable
for the Centralization of International Arbitration as equal to: 0 if no international arbitration is specified
in the BIT, 1 if only ad hoc arbitration is specified,23 and 2 if any institutional arbitration option is listed.24
22
We take the original three-category L-C-R variable and recode it with a value of one for “R” governments, and a
value of zero otherwise.
23
International arbitration procedures classified as “ad hoc” include the use of: UNCITRAL rules, modified
UNCITRAL rules, or treaty-specific rules for dispute resolution.
24
Among the international arbitration options that are classified as “institutional” are: ICSID, International Court of
Arbitration of the International Chamber of Commerce, Arbitration Institute of the Stockholm Chamber of
22
This coding reflects distinctions made by legal scholars who study choices in investment arbitration
(Franck 2007a, 2007b, Slate 1996, Sornarajah 2000) and is consistent with previous scholarship within
the rational design literature on arbitration choices (Mattli 2001).
A first important distinction reflected in this coding scheme is whether investors can pursue
grievances through international arbitration instead of being forced to seek redress through national courts
in the host state. Historically, potentially-biased domestic courts in host states have been inadequate to
solve the commitment and enforcement problem inherent to FDI (Franck 2007c, Guzman 1998). Because
international arbitration not only guarantees neutrality, but provides centralized benefits such as a set of
established rules and awards that have mechanisms for enforcement, it is strongly preferred by
multinational investors and their governments to domestic courts (Franck 2007b, Schreuer 2005). A
second important distinction, which gets at the heart of centralization, is between ad hoc and institutional
arbitration (Mattli 2001). Ad hoc arbitration is a notable step up from purely domestic remedies, yet
under ad hoc arbitration the parties still must navigate most aspects of the case, such as choice of law,
rules for selecting arbitrators, objection, compensation, and award procurement (Slate 1996). In contrast,
those who opt for more centralized arbitration benefit from established rules, quality control, basic
administrative support, hearing rooms, management of the proceedings, selection of arbitrators, facilities,
award enforcement, and legal support (ibid., 52-60; also Franck 2007a, 39).
The second institutional design feature, BIT “flexibility,” reflects the degree to which an
investment treaty provides the signatories with a range of choices for settling investment-related disputes.
Flexibility for resolving disputes is a central issue within BITs, since different pathways for dispute
settlement may return different outcomes or entail different time horizons for dispute resolution. As
noted earlier, there are two ways to conceptualize BIT flexibility. Mattli echoes this notion, claiming that
“(f)lexibility characterizes not only arbitral procedures but also the actual institutions of arbitration…”
(2001: 926).
Commerce, Cairo Regional Centre for Commercial Arbitration, Arab Investment Court, and the Permanent Court of
Arbitration (Hague).
23
The first variable that captures BIT flexibility is the number of arbitration venues from which
signatories may choose in pursing dispute settlement. Several observers of investment arbitration, in fact,
discuss forum choice and forum shopping as a means to achieve flexibility (Franck 2005, 2007c,
UNCTAD 1998). Governments provide greater future flexibility when they specify a range of possible
venues for the arbitration of investor-state disputes instead of limiting themselves to a single or limited
number of venues. This specification of multiple options allows government to retain greater control of
the process in an environment of future uncertainty. This first indicator of Flexibility (Total Number of
Arbitration Options) sums the number of international arbitration institutions – both ad hoc and
institutional – specified in the BIT.25
The second measure of BIT flexibility captures whether the signatories consent ex ante to
international arbitration of disputes. The least flexible treaties contain language in which both state
parties agree in advance to any international arbitration that results from an investment dispute.
Preconsent reduces flexibility by speeding up and systematizing the arbitration process (Yackee 2008)
and removing one of the three typical objections to jurisdiction – both parties’ written consent.26 In the
absence of a preconsent clause, however, the actors have greater control over issues relate to timing, the
need to exhaust local remedies, and other procedural issues. Once again, choice over such matters is
likely to be appealing to BIT signatories who are unsure about future development related to investment
treaties and investment arbitration. Therefore, we code treaties in which the parties must consent to
investment arbitration on a case-by-case basis as equal to 1 for our Flexibility (Consent Required for
International Arbitration) variable and equal to 0 when they instead give their advanced consent to
investment arbitration.
25
See footnotes 24 & 25 for a complete list of all relevant international arbitration bodies (ad hoc and institutional)
used in the count.
26
The other two are ratione personae and ratione materiae (Obadia 2001).
24
Empirical Findings
Attention now turns to empirical tests of the hypotheses described above, most of which are
inspired by the rational design framework, but some of which are inspired by its critics. Five sets of
analyses are conducted on a collection of 1,473 bilateral investment treaties concluded between 1965 and
2006, for which we have coded each treaty according to its degree of centralization and flexibility. The
first three sets of analyses (Tables 1-3) predict BIT centralization, and each one emphasizes a single
rational design cooperation problem that applies to BIT centralization (commitment problems,
enforcement problems, and uncertainty about the state of the world). The remaining two sets of analyses
(Tables 4-5) evaluate uncertainty about the state of the world as a predictor of flexibility, using both
aspects of BIT flexibility discussed earlier (number of arbitration options, advanced consent to
arbitration). All estimations are conducted using poisson, binary probit, or ordered probit estimators with
robust standard errors, depending on the nature of the centralization and flexibility dependent variables.
Additionally, the power and preferences variables are included in every model we estimate. In every case
we also control for changes across time by inserting a series of time-period dummy variables into the
regression, although for purposes of brevity we do not report the estimated coefficients of the time-trend
variables.27
In situations where multiple logics for a rational design concept exists, the primary indicator for
each logic is inserted first one its own, and then later (in the final column) with the primary indicator for
the other logics. Since rational design is not explicit on the issue of how to operationalize concepts in
statistical tests, treating each indicator as both nested and non-nested allows for the most comprehensive
evaluation. Also, in cases where a given logic could be measured in multiple ways, alternate indicators
are substitute one at a time in a series of sensitivity check regressions. The results of these checks are
listed in gray, to denote their secondary status, underneath the primary indicator in the table (which is in
black).
27
We include dummy variables for five-year periods (with 1995-1999 as the baseline) instead of yearly dummies,
since some of our measures are constant across all countries in a given year. Our results are very robust to changes
in the type of temporal controls (yearly dummies, a “year” variable, etc.) as well as baseline time period.
25
Taken as a whole, the estimated findings produce several general conclusions. First, the design of
BITs is certainly not random. Dtates do design BITs in systematic and purposeful ways, as is consistent
with the general thrust of the rational design program. Each table reveals the presence of multiple, strong
predictors of BIT design. Second, the rational design logic explains some aspects of BIT variation, but
not others. Receiving the most support are arguments about enforcement problems leading to
centralization and the impact of widespread uncertainty caused by shocks on BIT flexibility. Admittedly,
however, the overall support for the rational design conjectures is somewhat modest, particularly in
comparison to the support for alternate explanations. Finally, the preferences of states who sign BITs,
and the power relations between them, provide a strong, pervasive explanation for BIT design on both the
dimension of flexibility and that of centralization, in particular. In short, BITs are likely to contain
greater centralization when such a feature is desired by domestic and government interests in a powerful
state, and are much less likely to contain flexibility in either form when those same interests oppose the
inclusion of such options in BITs.
[Table 1 here]
The findings in Table 1 (along with those in Tables 2 and 3) provide a more in-depth illustration
of the workings of power and preferences as explanations for BIT centralization. All three variables
exhibit the predicted relationship with BIT centralization, all are statistically significant, and two of the
three are significant at the 99% level of confidence. In sum, BITs are more likely to have highly
centralized procedures for dispute settlement when multiple large MNCs are based in the signatories,
right-wing governments negotiate the agreement, and when one of the two states is considerably larger
than the other. This powerful result confirms the importance of preferences, or the “interests” of the
actors, as an important determinant of BIT design. Economic actors who anticipate investing heavily in
the partner states will strong prefer centralization, as will pro-market leaders on both sides – resulting in
treaties that reflect the preferences of these actors. Power also appears to be a central explanation of BIT
centralization. When one of the two states sits in a dominant economic position vis-à-vis the other, that
more powerful states is more likely to view itself as the source of investment into the other. As such, it
26
should strongly prefer to include centralized dispute settlement, and due to its greater bargaining leverage,
the treaty will reflect this preference.
Notwithstanding the strong support for power and preference-based explanations for BIT
centralization, the findings across Tables 1-3 reveal some support for the rational-design inspired
hypotheses. First, the inability of governments to commit to uphold the terms of FDI, based on a lack of
information about the government, does increase the likelihood that a BIT involving that country will be
more centralized. Table 1 reveals that when a ruling party in a signatory state has only been in power a
short time, any BIT involving that country is more likely to entail centralized procedures for dispute
settlement (and vice-versa). This finding holds when the Tenure of the Ruling Party is included as the
only “commitment problem” variable (see Table 1, model 1) as well as when it is included with indicators
that reflect the other commitment problem logics (see Table 1, model 4). Second, multiple results in
Table 2 suggest that enforcement problem, defined in two slightly different ways, can result in greater
BIT centralization. When a BIT signatory has witnessed slow economic growth or high unemployment
during the past three years, its BITs will include more centralized enforcement due to the increased
incentives to cheat in the short term. Likewise, some limited evidence suggest that pairs of states that are
less sensitive to the costs of cheating on a BIT, based on an otherwise small trading relationship, are
somewhat more likely to negotiate centralized BITs. This finding does not hold across expanded
estimations, however, (see Table 2, models 3 and 4) and does not apply to the logic of alliance ties, the
alternate indicator for sensitivity to costs.
[Table 2 here]
Perhaps the most supportive evidence for a rational design-inspired conjecture is found in Table
3, where two types of uncertainty about the current state of global FDI lead to greater centralization in
BITs. First, the presence of more recent investment disputes, which signifies heightened uncertainty
about rules for investment and appropriate behavior toward FDI, are associated with greater BIT
centralization (see model 1). Second, greater variance in recent global FDI flows, which suggests
increased unease about future patterns of investment and future FDI development, also leads to greater
27
BIT centralization. These findings lend support to the idea that uncertainty caused by FDI-relevant
“shocks” should affect BIT design.
[Table 3 here]
Although we find support across Tables 1-3 for various rational-design based arguments, one
should not draw any sweeping conclusions about the validity of the rationality-based framework to
predict BIT centralization. Only about one-quarter of the estimated relationships across the three tables
support the relevant hypotheses. Moreover, variables which reflect the power and preferences of the
actors are stronger and more consistent predictors of centralization in BITs, as detailed earlier.
Moving to the determinants of flexibility in BITs, it once again is apparent that the power and
preferences of the negotiating states play a major role in BIT design. For the first element of BIT
flexibility, advanced consent to arbitration, all three power- and preference-based variables play a role in
reducing flexibility, as predicted. The coefficients for all three variables in Table 4 are negative and
statistically significant at conventional levels. Once again, the interests of large multinationals and rightwing governments are reflected in agreements that tend to include, as opposed to remove, advanced
consent clauses. Similarly, less flexibility in BITs (i.e., advanced consent to international arbitration) is
more likely when one of the two signatories is much more economically powerful than the other.
Findings are less supportive, yet still notable, when the second aspect of flexibility, the number of
arbitration venue options, is examined. Table 5 reveals that in BITs where the preferences of MNCs are
relevant, fewer options for dispute settlement are included in the BIT. This lends further support to the
argument that investors would prefer to narrow the options for dispute settlement, since greater flexibility
simply leads to battles over jurisdiction and delayed dispute resolution.
[Table 4 here]
[Table 5 here]
Although none of the rational-design inspired arguments predicts BIT flexibility in the form of
case-by-case consent to arbitration (Table 4), multiple pieces of supportive evidence are revealed in Table
5. Put simply, the presence of various FDI-related shocks, which generate greater uncertainty about the
28
future state of global FDI, are positively associated with greater BIT flexibility in terms of more options
for dispute settlement. States specify more venue options for dispute settlement when there has been a
recent global currency crisis, whether in the past year or past three years (see Table 5). Similarly, states
also design greater venue flexibility into their BITs in the face of another type of global shock: high
volatility in global FDI flows during the past half-decade. In sum, the uncertainty-based arguments that
link shocks as a source of FDI-related uncertainty to greater BIT flexibility receive substantial support.
Discussion and Conclusions
Like the editors of the original Rational Design volume, we are motivated by a simple
observation—that investor-state dispute settlement provisions within BITs vary in important and
systematic ways.28 Our goal has been to advance conceptual thinking on how to best explain the
variations within BITs and PTAs that are increasingly apparent. The literature on the “rational design” of
international institutions emerges as a promising theoretical framework for explaining this design
variation. However, a number of striking conclusions emerge after using the framework as a basis for
theorizing about BIT variation and conducting a series of rigorous empirical tests.
Most notable is the importance of power and preferences in explaining investment treaty design.
Across all of the models we estimate, the preferences of governments and their important domestic
interests are significant predictors of BIT design. The relative power of the negotiating governments also
emerges as a consistent predictor. Ironically, these concepts enter into our theorizing because multiple
critics of the rational design approach stress their importance. What is clear is that anyone working in the
“institutional” or “rational” design tradition must think carefully about the preferences and power of the
actors who design institutions.
The mixed findings for rational-design inspired arguments also are informative. Several of the
more intuitive arguments, such as those about domestic institutions as commitment devices and the
28
On the first page of the introduction to the Rational Design volume Koremenos, Lipson, and Snidal write: “We
begin with a simple observation: major institutions are organized in radically different ways” (2001a: 761).
29
benefits of cheating on BIT, receive no support. It seems that many of the “rational” conjectures about
treaty design either require more refined argumentation or should be dropped in favor of more supported
arguments. Nevertheless, hypotheses about the design implications of “shocks” receive much more
consistent support, as do some arguments regarding the information treaty signatories possess about one
another. These arguments, then, merit future exploration and development.
This paper represents a unique effort to apply rational design theorizing to an important set of
economic treaties that vary in meaningful ways. As such, it offers several insights for future work on
investment treaties, other economic agreements, and institutional design more broadly. One important
contribution is our identification of dozens of measures for abstract concepts (“cooperation problems”)
common to the study of international relations. Although some of our measures are more original than
others, and the strategic context will differ across issue areas, we have done much of the conceptual heavy
lifting to produce an extensive list of indicators that can serve as a starting point for other researchers who
want to measure “uncertainty” or problems with “credible commitment,” even outside the investment
issue area. Several of our indicators use the statistical concept of variance as a way to measure
uncertainty – something that is rare in studies of political economy. We also believe that our method of
measuring trends, such as those in economic conditions, is another promising way to capture actors’
expectations of future behavior. Finally, our findings suggest an elevated importance for domestic
politics as a predictor of treaty design. The small amount of previous quantitative scholarship on rational
design has largely ignored domestic politics, but we emphasize internal factors as being the source of
many of the problems states face when negotiating treaties, and in international cooperation more
generally. Of course, domestic politics also can be the source of state interests, which our findings
demonstrate to be important. Whether the causal pathway from domestic politics to treaty design
represents a “rational” response to an international problem, or merely the channeling of domestic
interests into an international treaty, is a question left to future analysts.
30
Table 1: Ordered Probit Model of BIT Centralization as a function of
Commitment Problems, Preferences, and Power
Model 1
Model 2
Model 3
Model 4
Commitment Problems
Domestic Institutions as Commitment Devices
Political Constraints on Executive (-)
1.24 +
(.284)
Checks on Executive (-) a
.102 +
(.040)
Respect for Law and Order (-) a
.020
(.043)
Lack of Corruption (-) a
.022
(.042)
1.27 +
(.300)
(Lack of) Information about Signatory
Tenure of Ruling Party (-)
-.012 ***
(.004)
Tenure of Current Executive (-) a
-.011
(.012)
-.008 **
(.004)
Negative Past Reputation
Expropriations in past 5 years (+)
.081
(.109)
.092
(.115)
Preferences and Power of Signatories
Presence of Large MNCs (+)
1.93 ***
(.855)
1.37 **
(.824)
1.59 **
(.831)
1.70 **
(.848)
Right-wing Governments (+)
.124 *
(.081)
.207 ***
(.079)
.229 ***
(.077)
.108 *
(.083)
Relative Economic Power (+)
.473 ***
(.153)
.531 ***
(.161)
.553 ***
(.156)
.473 ***
(.157)
==================================================================================
a
Terms in gray are substitutes for terms in black above
*p<.10, **p<.05, ***p<.01 (one-tailed tests)
Time controls included but not reported
Robust standard errors in parenthesis
Number of Observations
1315
31
1274
1315
1274
Table 2: Ordered Probit Model of BIT Centralization as a function of
Enforcement Problems, Preferences, and Power
Model 1
Model 2
Model 3
Model 4
Enforcement Problems
Temptation to Cheat
GDP Growth Rate (-)
-.006
(.008)
Unemployment Rate (+) a
.010
(.012)
Inflation Rate (+) a
.007
(.013)
3-year trend in GDP Growth Rate (-) a
-.023 ***
(.009)
3-year trend in Unemployment Rate (+) a
.043 *
(.027)
3-year trend in Inflation Rate (+) a
-.003
(.004)
.002
(.010)
Benefits of Cheating
Raw Materials Prices (+)
-.002
(.003)
-.002
(.004)
Sensitivity to Costs of Cheating
Size of Bilateral Trade Relationship (-)
-.069 *
(.053)
Alliance Ties (-) a
.203
(.108)
.062
(.055)
Preferences and Power of Signatories
Presence of Large MNCs (+)
1.60 **
(.830)
1.51 **
(.836)
1.71 **
(.880)
1.67 **
(.889)
Right-wing Governments (+)
.230 ***
(.077)
.226 ***
(.077)
.238 ***
(.100)
.237 ***
(.099)
Relative Economic Power (+)
.542 ***
(.157)
.564 ***
(.157)
.513 ***
(.191)
.515 ***
(.191)
==================================================================================
a
Terms in gray are substitutes for terms in black above
*p<.10, **p<.05, ***p<.01 (one-tailed tests)
Time controls included but not reported
Robust standard errors in parenthesis
Number of Observations
1313
32
1315
815
814
Table 3: Ordered Probit Model of BIT Centralization as a function of
Uncertainty about the State of the World, Preferences, and Power
Model 1
Model 2
Model 3
Uncertainty about the State of the World
Amount of Information about Current State of FDI
Number of Recent BITs Globally (-)
.002 +
(.0004)
Number of Recent Invest. Disputes Globally (+) a
.014 **
(.008)
.002 +
(.0005)
FDI-Related“Shocks”
Global Currency Crises, past 3 years (+)
.010
(.017)
Global Currency Crises, past 5 years (+) a
-.0001
(.021)
Global Currency Crises, past year (+) a
.023
(.021)
Variance in Recent Global FDI Flows (+) a
.824 ***
(.369)
-.007
(.017)
Preferences and Power of Signatories
Presence of Large MNCs (+)
1.58 **
(.847)
2.03 ***
(.878)
2.09 ***
(.893)
Right-wing Governments (+)
.235 ***
(.080)
.255 ***
(.079)
.269 ***
(.083)
Relative Economic Power (+)
.617 ***
(.159)
.465 ***
(.160)
.516 ***
(.163)
==================================================================================
a
Terms in gray are substitutes for terms in black above
*p<.10, **p<.05, ***p<.01 (one-tailed tests)
Time controls included but not reported
Robust standard errors in parenthesis
Number of Observations
1315
33
1243
1243
Table 4: Logit Model of BIT Flexibility as a function of
Uncertainty about the State of the World, Preferences, and Power
DV = Flexibility
(Lack of Advanced Consent to Arbitration)
Model 1
Model 2
Model 3
Uncertainty about the State of the World
Amount of Information about Current State of FDI
Number of Recent BITs Globally (-)
-.0004
(.001)
Number of Recent Invest. Disputes Globally (+) a
.006
(.007)
0.0003
(.001)
FDI-Related“Shocks”
Global Currency Crises, past 3 years (+)
.007
(.023)
Global Currency Crises, past 5 years (+) a
-.015
(.029)
Global Currency Crises, past year (+) a
.012
(.027)
Variance in Recent Global FDI Flows (+) a
.076
(.247)
.011
(.024)
Preferences and Power of Signatories
Presence of Large MNCs (-)
-1.49 *
(1.04)
-1.51 *
(1.07)
-1.53 *
(1.07)
Right-wing Governments (-)
-.315 ***
(.089)
-.330 ***
(.091)
-.332 ***
(.091)
Relative Economic Power (-)
-.887 ***
(.223)
-.880 ***
(.231)
-.883 ***
(.231)
==================================================================================
a
Terms in gray are substitutes for terms in black above
*p<.10, **p<.05, ***p<.01 (one-tailed tests)
Time controls included but not reported
Robust standard errors in parenthesis
Number of Observations
1308
34
1236
1236
Table 5: Count Model of BIT Flexibility as a function of
Uncertainty about the State of the World, Preferences, and Power
DV = Flexibility
(Total Number of Arbitration Options)
Model 1
Model 2
Model 3
Uncertainty about the State of the World
Amount of Information about Current State of FDI
Number of Recent BITs Globally (-)
.0003 +
(.0001)
Number of Recent Invest. Disputes Globally (+) a
.001
(.001)
.0003 +
(.0001)
FDI-Related“Shocks”
Global Currency Crises, past 3 years (+)
.010 **
(.005)
Global Currency Crises, past 5 years (+) a
.005
(.008)
Global Currency Crises, past year (+) a
.022 ***
(.007)
Variance in Recent Global FDI Flows (+) a
.130 ***
(.049)
.005
(.006)
Preferences and Power of Signatories
Presence of Large MNCs (-)
-.378 **
(.216)
-.420 **
(.219)
-.404 **
(.218)
Right-wing Governments (-)
.030
(.020)
.030
(.020)
.032
(.020)
Relative Economic Power (-)
.059
(.048)
.055
(.049)
.058
(.048)
==================================================================================
a
Terms in gray are substitutes for terms in black above
*p<.10, **p<.05, ***p<.01 (one-tailed tests)
Time controls included but not reported
Robust standard errors in parenthesis
Number of Observations
1313
35
1242
1242
Bibliography
Abbott, K. & Snidal, D. (1998). Why States Act Through Formal International Organizations. Journal of
Conflict Resolution, 42(1), 3-32.
Abbott, K. & Snidal, D. (2000). Hard and Soft Law in International Governance. International
Organization, 54(3), 421-456.
Abbott, K., Keohane, R., Moravcsik, A., & A.M. Slaughter. (2000). The Concept of Legalization.
International Organization, 54(3), 401-420.
Alter, K. (1998). Who are the Masters of the Treaty? European Governments and the European Court of
Justice. International Organization, 52(1), 121-147.
Baker, J. C. (1999). Foreign Direct Investment in Less Developed Countries: The Role of ICSID and
MIGA. Westport, CT: Quorum Books.
Beck, T., Clarke, G., Groff, A., Keefer, P., & Walsh, P. (2001). New Tools in Comparative Political
Economy: The Database of Political Institutions. World Bank Economic Review, 15(1), 165-76.
Braumoeller, B. & Sartori, A. (2004). Introduction: The Promise and Perils of Statistics in International
Relations. In The Analysis of International Relations, edited by D. Sprinz and Y. Wolinsky, 129-151.
Ann Arbor: University of Michigan Press.
Brower, C. H. (2005). The Place of Arbitration. In International Investment Law and Arbitration:
Leading Cases from ICSID, NAFTA, Bilateral Treaties and Customary International Law, edited by
T. Weiler, 151-167. London: Cameron May.
Burley, A.M. & Mattli, W. (1993). Europe Before the Court: A Political Theory of Legal Integration.
International Organization, 47(1), 41-76.
Copelovitch, M. (2008). Uncertainty, Context, and the Duration of International Agreements. Paper
presented at the 49th Annual Meeting of the International Studies Association, San Francisco, CA.
Correlates of War Project. (2008). Trade Data Set, v2.0. Online, http://correlatesofwar.org.
Correlates of War Project. (2005). State System Membership List, v2004.1. Online,
http://correlatesofwar.org.
Croome, J. (1996). Reshaping the World Trading System: A History of the Uruguay Round. Geneva:
World Trade Organization.
Dolzer, R., & Stevens, M. (1995). Bilateral Investment Treaties. The Hague: Martinus Nijhoff.
Duffield, J. S. (2003). The Limits of “Rational Design.” International Organization, 57(2), 411-430.
Egger, Peter, and Valeria Merlo. (2007). The Impact of Bilateral Investment Treaties on FDI Dynamics.
The World Economy 30(10): 1536-49.
Elkins, Z., Guzman, A., & Simmons, B. (2006). Competing for Capital: The Diffusion of Bilateral
Investment Treaties, 1960-2000. International Organization, 60(4), 811-846.
36
Franck, S. (2007a). Empirically Evaluating Claims About Investment Treaty Arbitration. North Carolina
Law Review, 86(1), 1-88.
Franck, S. (2007b). Integrating Investment Treaty Conflict and Dispute Systems Design. Minnesota Law
Review, 92(1), 161-230.
Franck, S. (2007c). Foreign Direct Investment, Investment Treaty Arbitration, and the Rule of Law.
Pacific Mc George Global Business & Development Law Journal, 19(2), 337-373.
Garrett, G, & Weingast, B. (1993). Ideas, Interests, and Institutions: Constructing the EC’s Internal
Market. In Ideas and Foreign Policy: Beliefs, Institutions, and Political Change, edited by J.
Goldstein and R. Keohane, 173-206. Ithaca N.Y.: Cornell University Press.
Ginsburg, T. (2005). International Substitutes for Domestic Institutions: Bilateral Investment Treaties and
Governance. International Review of Law and Economics, 25, 107-123.
Guzman, A. (1998). Why LDCs Sign Treaties That Hurt Them: Explaining the Popularity of Bilateral
Investment Treaties. Virginia Journal of International Law, 38, 639-688.
Haftel, Y. (Forthcoming). Ratification Counts: U.S. Investment Treaties and FDI Flows into Developing
Countries. Review of International Political Economy.
Henisz, W. J. (2002). The Institutional Environment for Infrastructure Investment. Industrial and
Corporate Change, 11(2), 355-389.
Henisz, W. J. (2006). The Political Constraint Index (POLCON) Dataset. Available at
http://www-management.wharton.upenn.edu/henisz/.
Hirsch, M. (1992). The Arbitration Mechanism of the International Centre for the Settlement of
Investment Disputes. Dordrecht, The Netherlands: Martinus Nijhoff.
ICOW Colonial History Data, version 0.4. Online, http://garnet.acns.fsu.edu/~phensel/icow.html.
Jensen, N. (2003). Democratic Governance and Multinational Corporations: Political Regimes and
Inflows of Foreign Direct Investment. International Organization, 57(3), 587-616.
Kaminsky, G. (2006). Currency Crises: Are they all the Same? Journal of International Money and
Finance, 25, 503-527.
Kaminsky, G. & Reinhart, C. (1999). The Twin Crises: The Causes of Banking and Balance-of-Payments
Problems. American Economic Review, 89(3), 473-500.
Keefer, P. & Stasavage D. (2003). The Limits of Delegation: Veto Players, Central Bank Independence
and the Credibility of Monetary Policy. American Political Science Review, 97(3), 407-423.
Kobrin, S. (1987). Testing the Bargaining Hypothesis in the Manufacturing Sector in Developing
Countries. International Organization, 41(4), 609-38.
Koremenos, B. (2001). Loosening the Ties that Bind: A Learning Model of Agreement Flexibility.
International Organization, 55(2), 289-325.
37
Koremenos, B., Lipson, C. & Snidal, D. (2001a). The Rational Design of International Institutions.
International Organization, 55(4), 761-99.
Koremenos, B., Lipson, C. & Snidal, D. (2001b). Rational Design: Looking Back to Move Forward.
International Organization, 55(4), 1051-82.
Koremenos, B., & Snidal, D. (2003). Moving Forward, One Step at a Time. International Organization,
57(2), 431-444.
Koremenos, B. (2005). Contracting around International Uncertainty. American Political Science
Review, 99(4), 549-566.
Koremenos, B. (2007). If Only Half of International Agreements have Dispute Resolution Provisions,
Which Half Needs Explaining?, Journal of Legal Studies 36, 189-212.
Kydd, A. (2001). Trust Building, Trust Breaking: The Dilemma of NATO Enlargement. International
Organization, 55(4), 801-828.
Lake, D. (2009). Open Economy Politics: A Critical Review. Review of International Organizations
4, 219-244.
La Porta, R., Lopez-De-Silanes, F., Shleifer, A. & Vishny, R. W. (1998). Law and Finance. Journal of
Political Economy, 106(6), 1113-1155.
Leeds, B., Ritter, J., Mitchell, S., & Long, A. (2002). Alliance Treaty Obligations and Provisions, 18151944. International Interactions, 28, 237-260.
Marshall, M. & Jaggers, K. (2005). Polity IV Project Dataset. Arlington, VA: George Mason University.
Mattli, W. (2001). Private Justice in a Global Economy: From Litigation to Arbitration. International
Organization, 55(4), 919-48.
Mearsheimer, J. J. (1994/1995). The False Promise of International Institutions. International Security,
19(3), 5-49.
Minor, M. (1994). The Demise of Expropriation as an Instrument of LDC Policy, 1980-1992. Journal of
International Business Studies, 25(1), 177-88.
Morrow, J. D. (2001). The Institutional Features of the Prisoners of War Treaties. International
Organization, 55(4), 971-992.
Muchlinski, Peter. (2009). The Framework of Investment Protection: The Content of BITs. In The Effect
of Treaties on Foreign Direct Investment, edited by Karl P. Sauvant and Lisa E. Sachs, 37-71.
Oxford: Oxford University Press.
Neumayer, E. & Spess, L. (2005). Do Bilateral Investment Treaties Increase Foreign Direct Investment to
Developing Countries? World Development, 33(10), 1567-1585.
Oatley, T. (2001). Multilateralizing Trade and Payments in Postwar Europe. International Organization,
55(4), 949-969.
38
Obadia, E. (2001). ICSID, Investment Treaties and Arbitration: Current and Emerging Issues.
News from ICSID 18(2), 14-17.
OECD (Organization for Economic Co-operation and Development). (2007). International Direct
Investment by Country.
Paulsson, J. (1995). Arbitration without Privity. ICSID Review, 10(2), 232-257.
Powell, E. J., & Mitchell, S. M. (2007). The International Court of Justice and the World's Three Legal
Systems. Journal of Politics, 69(2), 397-415.
Priest, G. & Klein, B. (1984). The Selection of Disputes for Litigation. Journal of Legal Studies 13, 1-55.
PRS Group, Inc. (2007). International Country Risk Guide: Researcher's Data Set.
Reisman, W. M., and Sloane, R. D. (2004). Indirect Expropriation and Its Valuation in the BIT
Generation. The British Year Book of International Law, 74: 115-50.
Richards, J. E. (2001). Institutions for Flying: How States Built a Market in International Aviation
Service. International Organization, 55(4), 993-1017.
Rosendorff, B. P. (2005). Stability and Rigidity: Politics and Design of the WTO’s Dispute Settlement
Procedure. American Political Science Review, 99(3), 389-400.
Schreuer, C. H. (2005). Investment Arbitration: A Voyage of Discovery. Arbitration, 71(1), 73-77.
Schreuer, C. H. (2001). The ICSID Convention: A Commentary. Cambridge: Cambridge University Press.
Simmons, B. A. (2000). International Law and State Behavior. American Political Science Review, 94(6),
819-36.
Slate, W.K. (1996). International Arbitration: Do Institutions Make a Difference? Wake Forest Law
Review, 31, 41-64.
Smith, J. M. (2000). The Politics of Dispute Settlement Design. International Organization, 54(1), 137180.
Sornarajah, M. (2000). The Settlement of Foreign Investment Disputes. The Hague: Kluwer.
Swenson, D. (2005) Why do Developing Countries Sign BITs? Journal of International Law and Policy,
12(1), 131-154.
UNCTAD. (1998). Bilateral Investment Treaties in the Mid-1990s. Geneva: United Nations.
UNCTAD. (2007a). Foreign Direct Investment Database. Available online via: http://www.unctad.org.
UNCTAD (2007b). Investor-State Dispute Settlement and Impact of Investment Rulemaking. Geneva:
United Nations.
Vandevelde, K. J. (1992). United States Investment Treaties: Policy and Practice. Cambridge: Kluwer
Law International.
39
Vernon, R. (1971). Sovereignty at Bay: The Multinational Spread of U.S. Enterprises. New York: Basic
Books.
Wendt, A. (2001). Driving with the Rearview Mirror: On the Rational Science of Institutional Design.
International Organization, 55(4), 1019-50.
Williams, David. (2003.) International Commercial Arbitration and Globalization: Review and Recourse
against Awards Rendered under Investment Treaties. Journal of World Investment 4(2): 251-76.
World Bank. (2007). World Development Indicators. Washington, DC: The World Bank Group.
Accessed through WDI Online.
Yackee, Jason. (2008). Conceptual Difficulties in the Empirical Study of Bilateral Investment Treaties.
Brooklyn Journal of International Law 33(2): 405-62.
40